To cite this article:

Reynolds, Dwight F. "Composition in Performance Arab Style." Classics@ 14. Washington, DC: Center for Hellenic Studies, 2016.

Composition in Performance Arab Style

Dwight F. Reynolds, University of California, Santa Barbara
For researchers who conduct ethnographic fieldwork on oral traditions, one of the most exciting aspects of The Singer of Tales is Albert Lord’s accounts of experimentation in the field, the various ways in which he and Milman Parry interrogated living traditions by comparing performances, setting up encounters between poets, asking poets to repeat performances, asking them to sing about new themes, and so forth, all with the aim of understanding the nitty-gritty details of how poets learned, composed, and performed their tales. Together they developed fieldwork techniques that had a profound influence on all later researchers. Where others before them had been content to collect ‘texts’, Parry and Lord opened up the question of how oral compositions come into being and how they are transmitted from one generation to the next.
These techniques so excited me as an undergraduate, and later graduate, student that when I went into the field to do my dissertation research on the Arabic oral epic poem Sīrat Banī Hilāl (The Epic of the Bani Hilal Bedouin Tribe) in Northern Egypt in the mid-1980s, I sought to replicate those experiments as much as possible. [1] I hoped even to carry their innovative methods a bit further by engaging in an ‘apprenticeship’ with an epic singer and trying to learn to sing a portion of the epic, not in an attempt to become a singer, but rather as a means of exploring and understanding the techniques of oral epic composition and performance. It seemed worthwhile, not only to test the concepts of ‘oral formula’ and ‘composition in performance’ in a different oral epic tradition, but also to replicate the fieldwork methods Parry and Lord had developed in a different cultural context. In effect, I hoped to create a laboratory in which to test their ideas and hypotheses about the processes of transmission, composition, and variation in performance by using the same techniques they had used. I offer this brief summary of the modest results of my own work as an homage to their remarkably innovative fieldwork and as a means of reminding ourselves that Parry and Lord were not only groundbreaking scholars in the conclusions they drew from their materials, but also in how they collected them. It is quite possible that they might not have reached the conclusions they did if they had not conducted their research the way they did, for fieldwork methods to a great extent predetermine the intellectual outcomes they produce.

I. The Structure of Classical Arabic Poetry

The vast majority of written and oral Arabic poetry over the past 1,500 years has been composed within one basic form (qarīḍ in Arabic), in which each verse consists of two equal-length hemistichs separated by a medial caesura, and the rhyme established at the end of the first verse is found at the end of every verse thereafter for the duration of the poem, an arrangement commonly termed ‘mono-endrhyme’:
-------------------- -------------------A
-------------------- -------------------A
-------------------- -------------------A
The interaction between written and oral in the Arabic literary tradition is too complex to treat in detail here; suffice it to say that in the pre-Islamic and early Islamic eras it was quite common to compose and recite poems orally and only later, selectively, set them down in writing. In other words, even the written corpus of classical Arabic poetry that has come down to us is deeply intertwined with the processes of oral composition and performance. [2]
Given that classical Arabic verses typically average between 22 and 32 syllables in length depending on the meter, the medial caesura is virtually a performative necessity, because it creates a brief pause in which the reciter can take a breath. In classical Arabic poetry, it was considered a ‘defect’ to repeat a rhyme word unless it was used with a different meaning, a rule that effectively limited most classical Arabic poems to a maximum of some 120 lines. Although the exact nature of meter in early classical Arabic poetry has remained a source of discussion, the centrality of end-rhyme to the Arabic poetic tradition is unchallenged. [3] Mandatory mono-endrhyme is a feature of the earliest Arabic poetry known to us (c. fifth century CE) and may have existed in Himyaritic in Yemen for a number of centuries before that. [4] With the exception of some polyrhymed strophic song forms that emerged in the Middle Ages, obligatory mono-endrhyme remained a constant feature of nearly all classical Arabic poetic production until the introduction of ‘free verse’ in the twentieth century.
For the first few centuries of the Islamic period, there is little evidence that oral colloquial or folk poetry differed in its basic structure from the classical poetry described above; in other words, even colloquial poetry appears to have been composed as mono-endrhymed poems with medial caesurae. Beginning in the later Middle Ages, however, a number of new poetic genres begin to appear in the written record that are composed in colloquial or semi-colloquial Arabic and are multi-rhymed in a wide variety of patterns. [5] These new forms appear to reflect the diversity of colloquial oral traditions, such as are currently found in modern times in oral folk poetry around the Arab world.
The Sīrat Banī Hilāl oral epic in Northern Egypt and the scattered written fragments of epic verse that have been recorded from the late fourteenth century to the present, however, all use the classical mono-endrhyme with medial caesura form, which therefore sets it apart from most other genres of Arabic oral folk poetry. [6] Sīrat Banī Hilāl epic poetry is thus colloquial in language, but composed in the same form as classical Arabic poetry. It should be noted that narratives about the Bani Hilal tribe are also performed in many regions as folk tales, that is, in prose rather than verse and without musical accompaniment, but these are not of concern to us here. [7] The tradition examined in this essay is one in which the tales of the Bani Hilal have been unified into a single lengthy narrative which is performed almost entirely in sung rhymed verse with musical accompaniment, punctuated with occasional prose passages that are used primarily for scene-setting, rapid summarization of events, or comic interjections. The classical rule that a rhyme-word should not be repeated, however, is not used in oral epic singing, which allows singers to produce sequences of hundreds of verses with the same end-rhyme before switching to a different rhyme. In the twentieth century and today, the tradition of performing the epic in sung verse with musical accompaniment is found only in Egypt, though it may have been more widespread in earlier centuries.
The epic of the Bani Hilal is of prodigious length—a master singer in Egypt can take between 60 and 140 hours to sing his version—and consists of a complicated story involving a cluster of some eight to twelve key characters. The story is divided in performance into segments or episodes (Ar. qiṣaṣ, sing. qiṣṣa) of four to ten hours in duration, typically sung over one or two evenings. With its focus on a constellation of main characters, the overall narrative of the Bani Hilal epic more closely resembles, say, the Iliad, the Nibelungenlied, or the King Arthur cycle, than ‘single hero’ epics such as the Odyssey, Beowulf, or The Song of Roland. Although it has been clear for some time from the fragmentary pre-modern written records and from ethnographic materials collected in the late nineteenth and early twentieth centuries that Arabic epic poetry is to a great extent oral-formulaic in nature, it has also been apparent that oral formulae function rather differently in Arabic than in either the South Slavic epic traditions or the classical Greek tradition. What follows is a description of some of the key differences.

II. Of Rhyme and Reason

Parry and Lord were far ahead of their time in stressing the importance of understanding the processes of transmission and in their exploration of how young men learned to sing epic poems. Unfortunately, I was not afforded the opportunity to observe young singers learning to sing, nor did I have the opportunity to record singers at various stages of their lives as they mastered and developed their craft. Already in the 1980s the sons of epic singers in northern Egypt were going to school and learning other trades rather than doing a traditional apprenticeship of ten or more years with older male relatives.
The singing of the Bani Hilal epic in this region has been almost entirely a hereditary, professional craft restricted to specific social groups, a sharp contrast with the South Slavic traditions where young men learned the craft by attending performances, rather than being trained by family members. Although there had been a few examples in the not-too-distant past of men in northern Egypt who had learned the epic purely out of their love for it, these singers had all passed away before I undertook my research. The most famous singer of this type in northern Egypt was Sayyid Hawwas, who had become a local ‘superstar’ by performing selected scenes from the epic on the western violin, accompanied by several other musicians, with highly amplified sound and at times with painfully strong ‘reverb’, standing on a stage like an urban singer. [8] All of the singers I worked with were instead hereditary professional singers from a single distinct social group, one of several groups in Egypt that have at times been labeled “Gypsies,” but who are not, however, directly related to the Roma of Europe. In this social group, there have been, until the last 30 years or so, two traditional occupations for males: epic singing or blacksmithing. There was no choice between the two—one was either the member of an epic-singing family or a blacksmithing family, and one inherited the occupation of one’s father and grandfather. (Women engage in a variety of different informal trades such as selling vegetables in the marketplaces, weaving baskets or mats, etc., and in recent decades young men have sought a wide variety of jobs including in construction, carpentry, and even the local police force.) Because of this clear knowledge that boys from given families would grow up to be singers, the learning process began quite early, sometimes by the age of five or six.
Older poets all described a very similar process of apprenticeship that involved attending performances with male relatives from an early age, learning to play drone notes on the rabāb (the two-string spike fiddle that is the traditional musical instrument for epic singing), then simple melodies, next learning to sing shorter song genres such as praise songs to the prophet Muhammad and mawwāls (short lyric songs), and finally scenes and later full episodes from the epic itself. Most said that they started out performing the shorter genres during breaks in the epic performance while their father or uncle or grandfather smoked a cigarette and had a glass of tea before picking up the epic tale again. Later they would perform a scene or two during an evening’s performance, and by their late teens they were capable of singing a full episode (i.e. a performance of four to ten hours). Some struck out on their own to perform as itinerant poets at a young age while others continued to play ‘second fiddle’ to an older male relative for many years.
After several months of attending and recording performances of oral epic singing in the northern Egyptian village in which I did my dissertation fieldwork, I decided to expand my research by attempting to learn to sing the epic from a master poet. Ethnomusicology as a discipline has long stressed the value of learning to perform the traditions one studies on the principle that there are many elements about musical structures, ideas, and techniques that only become apparent in actual performance and because it is necessary to understand a performer’s conception of what he or she does and observe how the tradition is transmitted. [9] My very first lesson, however, was an abysmal failure—my teacher sang perhaps 20 to 25 verses and then asked me to sing them back to him. I, unfortunately, could scarcely remember the first few words of the first verse. On the other hand, having just heard the rather simple melody 20 or more times, I had no difficulty playing the tune back to him on the rabāb. He found this quite surprising and at a later point asked me pointedly why I was able to play the instrument so easily when, in his words, my tongue was “made of wood.” When I first asked permission to record our lessons, he refused since he naturally only wanted good, full performances to be recorded, but I finally persuaded him. I would then spend hours after my lesson listening to the tape over and over again, and although I at first tried to memorize my lessons orally, I simply was not able to learn more than a few verses at a time in this manner. I eventually accepted the fact that I was hopelessly literate and resorted to writing down passages on scraps of paper that I could refer to during the day to jog my memory. Mine was therefore not a traditional apprenticeship—but it was never intended to be.
What taking ‘lessons’ did allow me to do was to test the boundaries of acceptable variation. For example, I soon found out that my teacher did not necessarily expect me to sing passages back to him in my next lesson to the same melody he had used. If I chose a known melody that worked with the length of the line, he voiced no objections. As any scholar of oral tradition might have predicted, he also did not have a word-for-word or even line-for-line recollection of what he had taught me in previous lessons—if I got the story right (in other words, retold the events of the story as he had taught them to me), everything was fine. I am convinced that he was also unaware that at times I memorized passages as closely as I possibly could, producing a nearly word-for-word replication in which I memorized even the slight melodic variations from his performance. This type of ‘verbatim’ imitation drew no comments from him.
But then I tried to change something that did matter to him. Battle scenes in the epic are very repetitive. The heroes always challenge each other to single combat, they always taunt and insult each other before beginning to fight, the bad guy always appears to get the upper hand at first, and the good guy always wins in the end. So one time I substituted passages from a battle-scene in one episode for a scene in another. My teacher immediately cocked his head and said to me, “That is not from this episode.” I asked how he knew this and the answer was deceptively simple—the rhyme was wrong. As we discussed the matter, he was astonished that I had never noticed that each episode had a set rhyme scheme. When I asked him what he meant he immediately rattled off the rhyme scheme of the episode I was learning: it starts in one rhyme, and then, when such-and-such happens in the narrative, it changes to a different rhyme, and later when so-and-so sings his poem about X, that poem is in a different rhyme, and so forth. For several months I had been listening, recording, transcribing, and translating texts and I had failed to notice the single most important structural element in the minds of the poets—the rhyme scheme! Of course I had paid quite a bit of attention to rhyme, but I had failed to notice that the same poet from one performance to the next, and even different poets performing the same episode, followed a traditional, inherited rhyme scheme.

III. Oral Formulaic Systems

Milman Parry defined an ‘oral formula’ as “a group of words which is regularly employed under the same metrical conditions to express a given essential idea.” [10] This definition certainly holds true for traditions that feature strict metrical constraints, but what about poetic traditions in which meter is not the primary structuring device of epic verse? [11] In Arabic oral epic poetry in northern Egypt, end-rhyme is strictly observed, as is the overall length of each line (typically 26–30 syllables), but the metrical system is rather flexible. Classical Arabic poetry as it has come down to us in written versions possesses a complex system of quantitative meter based on alternating patterns of long and short syllables, but even in this seemingly rigid system it is completely acceptable at most points within a verse to substitute two short syllables for one long syllable or vice versa; the classical metrical system was thus fixed but somewhat elastic in nature. This metrical elasticity is even more apparent in the sung tradition of oral epic poetry, where poets have additional musical resources that further allow the ‘stretching’ of a single syllable via melisma (singing a single syllable over several notes) or the ‘squeezing’ of several syllables into a single note. In this tradition, which has notably long verses and a relatively malleable metrical structure, it is therefore not surprising to find the most easily identified oral formulae clustered at the end of lines where the singers face their greatest compositional challenge—the generation of end-rhyme. The poets of Northern Egypt refer to this as al-mashī bi-l-ḥarf, walking or going along with the “letter,” perhaps better rendered as “sticking to the letter” in colloquial English.
One of the techniques that poets have developed for generating end-rhyme while singing their tales is the use of what might be termed ‘oral-formulaic systems’ (clusters of related oral formulae) that consist of a stem-phrase plus a choice of different rhyme words that all express a similar idea. This is a mechanism that allows poets to express the same or very similar ideas with different rhymes, the way South Slavic oral formulae allow poets to express the same idea in different metrical contexts. They also use closely related stem-phrases of different lengths, so that one can use the formula to fill out anywhere from a few syllables at the end of a verse to an entire hemistich of twelve to fifteen syllables. The linchpin of the system, however, is always the rhyme word.
A common example of this system is an expression in the archaic diction of the epic for losing control of oneself from grief, anger, surprise, or some other intense emotion, which is very similar to the English phrase “to lose one’s mind.” This idea occurs almost exclusively in final (i.e. rhyme) position and the many variants of this idea form an ‘oral formulaic system’ that can be used to express the same idea in all of the most common rhyme schemes. Below are examples of this simple formula attached to ‘stem phrases’ of various lengths so that the same idea can be expressed in various different rhymes in phrases that range from six to eleven syllables:
I. His mind from him + rhyme word
A. il-‘aql-I [12] minnuh: tāh / rāḥ / ṭār / hām / iḥtār [6-7 syllables]
His mind [13] from him: strayed / departed / flew / wandered / grew confused

B. il-‘aql-I min dimāghuh: tāh / rāḥ / ṭār / hām / iḥtār [8-9 syllables]
His mind from his skull: strayed / departed / flew / wandered / grew confused

C. il-‘aql-I min juwwa dā rāsuh: tāh / rāḥ / ṭār / hām / iḥtār [10-11 syllables]
His mind from inside his head: strayed / departed / flew / wandered / grew confused
The very simple cluster of formulae in set A, when used at the end of the verse, makes it possible to express the same basic idea with many different rhymes. Demonstrated above are -āh, -āḥ, -ār, and -ām, with an additional option of a two-syllable final word within the ‘r’ rhyme (iḥtār), all of which are appended to a five-syllable stem phrase (il / ‘aq / l-I / min / nuh). In set B, the same rhyme words are appended to a seven-syllable stem phrase, and in set C to a nine-syllable stem phrase. There are many other rhyme words that can occur in this position such that this particular formulaic system allows oral poets to express this same idea in almost any rhyme environment. But there are also closely related formulae that express very similar ideas with slightly more complex or different imagery. These occur somewhat less frequently than the phrases in the example above:
II. His mind departed from …
il-‘aql-I rāḥ min: ḥimāh / il-maṭāriḥ
His mind departed from: its sanctuary / these places
III. His mind …
‘aqluh: fāriq il-mīzān / ghāyib / ikhtalaṭ bi-jinān
His mind: tipped the scale / (is/was) absent / mixed with madness
Thus if South Slavic oral formulae exist to assist the expression of ideas with “the same metrical contexts,” we might make a slight alteration of the definition for these Arabic formulaic systems as existing to help poets express ideas in the same ‘rhyme contexts’. The change is quite logical—in South Slavic oral epic, meter is the primary structural constraint upon the singer, whereas in Arabic the primary structural constraint is end-rhyme.
These systems of verse-final formulae were readily apparent to me even in the early part of my research. The work of Parry and Lord had already provided an effective model for identifying, marking, and calculating the percentages of such formulae, and these were therefore the first elements of compositional technique I was able to identify. Their research also suggested one of the first fieldwork ‘experiments’ that I conducted. Every now and then, I took to asking a poet to re-sing a passage immediately under some pretext so I could study the amount of variation. I would, for example, say, “Oh no! That tape was bad,” or offer some other excuse, and then ask if the poet could please re-sing the part of the story that he had just finished.
One time it suddenly occurred to me to ask Shaykh Taha Abu Zayd, my teacher, if he would repeat a section—but on a different rhyme. This occurred during a performance in a private home in front of about a dozen adult male listeners and I made my request while Shaykh Taha was taking a break, sipping tea, and smoking a cigarette. He stopped, looked me squarely in the eye and asked, “What rhyme do you want?” He had sung this section on the rhyme /-ār/, a rather common rhyme, so I chose another common rhyme /-ām/. [14] He again looked me in the eye, which made me feel that I was being a rather impertinent, perhaps even disrespectful student, but he immediately picked up his rabāb and re-sang 65 verses of poetry without a pause, changing the rhyme in every verse, but with almost no changes in the narrative. [15] I was amazed and enthusiastically asked if he could do it again on a third rhyme, but the audience members objected, saying, “Doktoor, what didn’t you understand? We have already heard this part twice!” The poet looked at me and smiled, as if to underline that although he and I were aware of what he had just done, the audience members were totally oblivious to the virtuosity that had just been demonstrated (a thought that he confirmed in our conversation the following day), and were instead focused entirely on the story. The resulting two versions of the same scene provide ample testimony to Shaykh Taha’s mastery both of the art of epic singing and of the versatility of the system of end-rhyme formulae. Six verses from these two parallel performances are compared below; each performance was a half-hour in length (i.e. the side of one cassette tape) and they were almost exactly the same number of verses in length, though Shaykh Taha started the second performance with a few introductory verses that were not part of the original version. Verbatim repetitions are underlined, nearly verbatim repetitions are marked with broken underlining, passages that have been changed are unmarked, and the rhyme phrase appears in English between slashes and thereafter the rhyme word appears in Arabic transliteration.
In this scene, the hero Abu Zayd [16] (also known as Salama) is arriving in a city after a long journey across the desert and he poses as an epic poet (a common motif in this tradition); he introduces himself to the group of men he finds there, and is asked to sing, but while he is singing, the city is attacked by enemy forces and the alarm is sounded. Abu Zayd then doffs his poet’s garb, mounts a horse, rides with the men of the city into battle and slays their enemy. Thus the “lowly poet” is revealed to be in truth a valiant hero, and vice versa. [17] The audio recordings of these two performances can be accessed at the Sirat Bani Hilal Digital Archive (—the passages below are quoted from the final fifteen minutes of recording 118-B and the first fifteen minutes of recording 119-A
Text / Line   Rhyme
Text 1 / Line 5: Abū Zayd reached them: “My greetings to the Arabs,
     My greetings to those on the right, yes, and are repeated /to those on the left/
  Ṭabb ‘aleyhum Abū Zeyd: “Salāmī ‘al-‘arab / salāmī ‘ala l-yumne aywa ti‘ūd yasār”  
Text 2 / Line 11: My greetings to you, O Bedouin,
     My greetings to those on the right and I repeat /my greetings/
  “Salāmī ‘aleykum ayyuhā l-bawādī / salāmī ‘al-yumnā wi-a‘īd is-salām  
Text 1 / Line 6: Abū Zayd recounted to them what had befallen him;
     They welcomed him /from right and left/
  Ḥakā lhum Abū Zeyd bi-mā qad jarā luh / dōl raḥḥabū min yamīn wi-yasār  
Text 2 / Line 13: Abu Zayd recounted to them what had happened to him,
     And they welcomed him, O Nobles, not just with /words/
  Ḥakā lhum Abū Zeyd bi-mā qad ḥaṣal luh / wa-raḥḥabū yā agāwīd min ghayr-I kalām  
Text 1 / Line 8: What is your craft, O Uncle, and what is your profession?”
     So Salāma [= Abū Zayd] said, “I am a poet /of princes/”
  “Mā ṣan‘itak yā ‘amm wi-ēh mihnitak?” / fa-qāl Salāma, “Dā anā shā‘ir al-umār”  
Text 2 / Line 14: What is your profession and what is your craft?”
     He said to them, “I am a poet /I measure words/.”
  “Mā mihnitak wi-ēh ṣan‘itak?” / fa-qāla, “Shā‘ir anā bawzin il-kalām  
Text 1 / Line 9: They said to him, “Poet, take out your rabāb,
     Let us hear, man, praise of the Prophet, /the Chosen One/.
  Gālū [18] luh, “Shā‘ir, ḥill-I rabābak / sammi‘nā yā rāgil madḥ in-nabī il-mukhtār  
Text 2 / Line 16: They said to him, “Poet, O Uncle, take out your rabāb,
     Let us hear praise of the Arab Prophet, /the High, the Leader/.”
  Qālū luh, “Shā‘ir yā ‘amm-I, ḥill-I rabābak / sammi‘nā fī madḥ in-nabī il-‘arabī al-‘ālī al-miqdām  
Text 1 / Line 11: Abū Zayd was making poetry when suddenly the horses of the foreigners
     Came out from the mountains, /men of King Ṣaymūl, strong of refuge-seeker/ [19]
  Abū Zeyd biyish‘ir illā wi-khēl il-‘agam ṭālla min ig-gabal / rigāl malik Ṣaymūl ‘azīz il-jār  
Text 2 / Line 19: Abū Zayd was making poetry when suddenly the horses of the foreigners
     Came out from the mountains, /each warrior like seven at the time of weighing/
  Abū Zeyd biyish‘ar illā wi-khēl il-‘agam ṭāllū min ig-gabal / min kull-I fāris sab‘ waqt-I mīzān  
Text 1 / Line 14: Put up your rabāb—Ah! Ah! Ah!—O poet of the Arabs
     The horses of the foreigners have come to us, they have seized /the regions/
  Tāwī rababak āh āh āh yā shā‘ir il-‘arab / khēl il-‘agam gātnā itmalakū l-ashṭār  
Text 2 / Lines 21 and 22: O poet of the Arabs, listen as we speak to you,
     Stop your rabāb and cut short your words
The horses of the foreigners have come to us, right and left,
     The foreigners have come to us and have seized /the fronts/

  Yā shā‘ir il-‘urbān w-isma‘ nikallimak / baṭṭal rababak wi-qill il-kalām
Khēl il-‘agam gātnā yamīnan wi-maysara / gātnā il-‘agam w-itmalikū l-aqdām
Most of the verses in this performance were converted from one rhyme to another through a simple substitution of the final rhyme word or phrase utilizing precisely the type of ‘oral-formulaic system’ demonstrated above. But in a few cases, Shaykh Taha made no attempt to ‘convert’ the image from his first performance by changing the rhyme word, but rather shifted to an entirely different image that nevertheless described the same action. These passages, upon inspection, turn out to be some of the most common stock motifs of the epic, for example, the hero dressing, arming himself, and then riding into battle. Below are the four verses from the first performance and the two and a half verses in the second performance that describe these actions:
Text 1 / Lines 24-28:
Salāma [= Abū Zayd] dressed, possessor of audacity,
     Tall of stature, sugar of the skirmish,
He put on brocade, and then silk, and then chainmail,
     Then he girded himself with an Indian [sword] that splits stones,
Plates of steel on the chest of Salāma,
     With them he deflects the arrows of the heathens,
Plates of steel, plates and arm-guards on his right arm,
     With them he deflects attack when taking revenge.

Text 2 / Lines 30-32:
When the Hilālī [= Abū Zayd] had dressed with resolve,
     In his hand was a Yemeni [sword] with a well-sharpened blade-edge.
God is Great! Abū Zayd when he mounted!
     Like the sail of a ship on the seas as it wanders.
Or a tall dovecot tower filled with doves. [20]
It turns out that the poets possess what might be called ‘templates’ for the most common stock scenes in different rhyme schemes. If one is singing in the rhyme -ār, the ‘dressing for battle’ scene leads almost inevitably to the image of a sword that “splits stones” [yiflaq al-aḥjār], an image that is not found in other rhymes, perhaps for lack of other words that fit the image of “splitting.” It is a ‘rhyme-specific’ formula. The substituted image in the second performance (“a well-sharpened blade-edge”) in contrast, does generate a number of other possibilities, due to the richness of Arabic vocabulary for swords, blades, edges, etc. (nearly 40 terms for swords and blades occur in the epic); it is part of an ‘oral-formulaic system’.
In sum, the experiment of asking Shaykh Taha to re-sing a section of the epic on a different end-rhyme revealed several things about the role of oral formulae in this tradition: 1) oral formulae occur most frequently at the end of verses, that is, in ‘rhyme position’; 2) many of these formulae operate within formulaic systems that allow poets to substitute different end-rhymes with a fair amount of ease by changing only one or two words; 3) there are also certain formulae that occur in rhyme-specific environments, that is, they occur only with a specific rhyme; and, 4) sequences of formulae constitute a form of ‘template’ for expressing stock scenes in a particular rhyme. If all stock scenes were sung using the same images and phrases with changes only in the rhyme words, Arabic epic singing would be boring and repetitive in the extreme, so this final element, the different rhyme ‘templates’ as I have termed them, plays an important role in creating a richer, more varied set of phrases, images, and details for embellishing common narrative scenes. I only came to realize the centrality of these rhyme templates, however, when I tried out another fieldwork ‘experiment’, again inspired by the example of Parry and Lord.

IV. Two Poets Compared

The example below presents a comparison of two different poets singing the same episode—the birth of the hero Abu Zayd. The epic is performed with short passages in spoken prose in between much longer passages of sung verse. The transition from prose to sung verse can only take place at the beginning of a character’s speech or song, that is, the junction between speaking and singing is always at the opening of a quotation of direct speech. Once he has begun to sing, however, the poet can use sung verse for first-person speech, dialogue, reported narrative, external commentary, or description. He can switch from sung verse back to spoken prose whenever he wishes, and this is sometimes used to speed up a performance by summarizing events quickly or for the insertion of some comic material, both of which are more easily done in spoken prose. In the chart below, it is easy to see that these two poets use almost exactly the same rhymes for the same narrative material, and also insert their brief prose passages in nearly the same places. It should be noted, however, that whenever a performance is interrupted, a poet must use a prose introduction to get back into the story, so not all prose passages are the result of the poet’s aesthetic choices, but may instead reflect external factors in the performance situation.
Rhyme Scheme Comparison
Shaykh Taha Shaykh Biyeli Story
prose prose 1) Introduction of characters; Rizq has married eight women, but does not yet have a son to be his heir.
-āXiḥ -āXiḥ [21] 2) Rizq’s complaint about his state; he marries Ghanim’s daughter, but she bears a girl, then a crippled boy; he sends her back to her father; Rizq goes hunting with Sarhan; in the desert a dervish shaykh tells Rizq to go to Mecca and marry Khadra, daughter of the Sharif of Mecca, and she shall bear him a son.
prose prose 3) The Bani Hilal tribe arrives in Mecca and are welcomed by the ruler, Qirda; Rizq wishes to marry his daughter.
-ālhā -ān [22] 4) the Hilali Qadi [judge] proposes marriage and describes prose the dowry; they marry; seven years pass and they still have no children.
" -ālhā [23] 5) Ghanim tells Rizq to divorce Khadra and marry another.
-āh -āh 6) Rizq goes to Khadra and confronts her; she greets him and calms him saying that all is in the hands of God; Rizq grows angry and strikes her; she leaves her tent and goes to Shamma, another barren woman of the tribe; they go to the desert and wish upon birds—Shamma for a son as beautiful as a white bird, Khadra for a son as strong as a black bird she sees; her servant wishes also for a son; at their return they are reconciled with their husbands.
prose prose 7) Shamma bears Ḥasan, Khadra gives birth, but the boy is black; Rizq is ridiculed by the men of the tribe who say that Khadra has committed adultery with a black slave; Khadra’s servant bears a son to Rizq’s servant, Najah.
-āra X [24] 8) Najah goes to his master Rizq with the good news.
-āh -ālī 9) Rizq grows angry and confronts his wife; they argue.
-ālī " 10) She protests her innocence, but he casts her out.
-āh prose 11) They propose to divide the herds by throwing a spear; a dervish shaykh appears and takes the spear; he casts it and wins for Khadra all of Rizq’s livestock and half of Ghanim’s.
X [25] " 14) Rizq in his grief leaves the tribe to live alone among the wild beasts of the desert; Khadra seeks refuge with the other nobles of the tribe but is refused.
-āh " 15) Khadra is escorted into the desert by a shaykh.
" -āh 16) Khadra complains of her fate; they come to a cross-roads and she chooses the road to the Zahlan tribe, the enemies of the Bani Hilal; the shaykh cannot take her further and leaves; she is alone and lost in the wilderness with her newborn son.
The above comparison covers 30 minutes of performance and Shaykh Taha’s version includes 184 verses of rhymed poetry. [26] We can see that although the rhyme scheme is not absolutely fixed, these two poets do share a basic rhyme scheme in their minds. They shift from prose to poetry and back again at the same points in the narrative. At a few points one poet uses prose when the other uses verse, but during the verse passages, there are only a few verses where the end-rhyme in the two performances are not in agreement.
In discussions with the poets about rhyme schemes, I came to realize that they possessed an additional concept that I had failed to grasp: farsh al-qiṣṣa or “the bedding of the story,” by which they mean that the overall story is based on a sequence of common rhymes that are rather easy to improvise in, but that specific poems stand out against this background because they are composed in rhymes that are more unusual and more difficult to use. In essence, there is an ‘aria and recitativo’ structure to the epic. The ‘arias’ are in more difficult rhymes and the ‘recitativo’ passages are in easy rhymes. It is also clear that the ‘recitativo’ sections (which include all of the more banal stock scenes such as travels, descriptions of gardens, descriptions of beautiful maidens, battle-scenes, and so forth) also vary more from one performance to the next in their wording and length—exactly as we understand oral-formulaic composition in performance from the work of Parry and Lord. But there also exist contrasting passages, the ‘arias’, which vary much less, and are also much better known by the audience. These ‘arias’ appear most frequently at dramatic or highly emotional points in the narrative. Audience members wait for these moments with almost palpable anticipation, and poets will occasionally take the opportunity to liven up the performance by singing the best known verses several times over, encouraging listeners to sing along. [27] I several times heard audience members criticize the performance of one of these ‘arias’ for having left out a favorite line or group of lines, a level of familiarity and consistency that is not possible with the ‘recitativo’ sequences. It is not that these ‘arias’ contain no oral formulae—they do; but because the rhymes in these passages are used much more rarely and for shorter periods of time, their oral formulaic quality is much less evident. [28]
The “bedding” of “The Birth of Abu Zayd” consists of two rhymes /-āh/ and /-ān/, to which the poets return again and again. The opening ‘aria’ is a shakwā or complaint against Fate and Destiny, sung by the father of the future hero who has sired a dozen daughters but has no son to carry on his name. This passage is composed in a double rhyme (i.e. where the two final syllables rhyme) in the pattern /-āXiḥ/ (where X stands for any consonant, for example, wāḍiḥ, sāyiḥ, maṭāriḥ); later, when the mother of the hero is accused of adultery and thrown out of the tribe because her son is black, she sings a poignant defense of her honor and requests God to protect her and her son in another double-rhymed ‘aria’ on the pattern /-alī/, and so forth. Some of these ‘arias’ are even further embellished by internal rhymes.
This overall rhyme structure also helps explain my inability to replicate one of the experiments that Lord recounts in The Singer of Tales. Try as I might, I could never get the poets I worked with to compose on new themes, and even more baffling to me was that they knew episodes from the epic that they could recount in prose, but which they said they could not sing in poetry. The poets even repeatedly referred to these episodes in their introductions to evening performances. For example, they might say: “Now after Abu Zayd had stolen the famous Mare of Jabir al-‘Uqayli and married his daughter ‘Alya, he traveled back to the camps of the Bani Hilal tribe …” and here the performance began. I was particularly eager to record a performance of this particular episode because Lady Anne Blunt and her husband Wilfred Scawen Blunt had recorded and translated a version in Egypt in the nineteenth century with which I hoped to compare a version recorded from oral tradition a century later. [29] But to my frustration, my teacher, and all of the other poets in the village, simply said: “I don’t sing that episode.” They could tell me the story, in full detail, in prose, but consistently said they could not sing it. The work of Parry and Lord, however, made me believe that they should be able to sing it, so I pressed my teacher until he finally said, “I have never heard it sung—I don’t know its rhymes.”
It is interesting to note that Parry encountered a somewhat similar situation where poets refused to sing rhymed poems when they had forgotten the rhyme, but this was because their method of oral-formulaic composition did not include techniques for creating rhyme. Unrhymed poetry was never sung exactly the same by two different singers, but rhymed poetry had to be memorized: “The poem in rhyme, however, must be learned by heart, and when forgotten it cannot be re-improvised on the instant, since the rhymes present too great an obstacle to such improvisation.” [30] Epic poets in Egypt, however, are able to improvise in rhyme because their oral-formulaic methods of composition are specifically geared to that goal. Their objection was rather to trying to recreate an episode by creating new rhymes since they did not know the traditional rhyme scheme.
In the ensuing discussion with Shaykh Taha, it became clear that he felt strongly that one did not just go about creating new episodes—this went against his own sense of transmitting what he had been taught by his father and grandfather. This poet was a master improviser within a performance, as we have seen above, but ‘creating’ an episode went against his concept of the epic itself. His statements that he was transmitting what he had been taught were not just a rhetorical stance, rather they represented a very real understanding of the tradition, and this element makes the Arabic process of ‘composition in performance’ rather different from that so wonderfully and richly documented by Parry and Lord among South Slavic singers.


The poets who sing the Arabic oral epic poem of Sīrat Banī Hilāl learn a basic narrative structure coupled with the technique of ‘composition in performance’ that allows them, for example, to expand or contract passages in response to their audience’s reactions by a method very similar to the techniques described by Parry and Lord. But there are also at least two additional levels of structure to Sīrat Banī Hilāl that do not seem to exist in the South Slavic tradition: first, there is a historically transmitted rhyme scheme for each episode that the singers preserve and pass on from generation to generation; and, second, there are individual passages (that I have here termed ‘arias’) that are more tightly—and more distinctively—composed, that remain more stable over time, and that appear to be somewhat less formulaic than other passages. With this essay, I would therefore like to pose the question to scholars of other oral poetic traditions whether the texts and performances they study are entirely of the same texture, cut from the same type of cloth, so to speak, or whether there are passages that stick out from the overall flow of the poem, similar to these ‘arias’, that is, passages that are more fixed, or are in some other manner distinctive and separate from the larger poem.
Parry and Lord were ground-breaking scholars not only in the conclusions that they drew from their research, but also in the manner in which they conducted their research. Without their innovative and experimental research techniques, they might never have discovered the key ideas that have proved so fruitful and so productive in the study of so many different traditions, oral and written, the world over. Their ideas about ‘composition in performance’ and ‘oral-formulaic composition’ have been compelling contributions to the study of oral poetry; but it should not be forgotten that their fieldwork methods are equally deserving of notice and emulation.


‘Abdallah, Y. 1988. “Der Sonnengesang von Saba: ein Stück religiöser Literatur aus dem antiken Jemen.” Die Königin von Saba: Kunst, Legende und Archäologie zwischen Morgenland und Abendland, ed. W. Daum, 185–192. Stuttgart and Zurich.
Blunt, A., and W. S. Blunt, eds. 1892. The Celebrated Romance of the Stealing of the Mare. London.
Blunt, W. S. 1914. The Poetical Works of Wilfred Scawen Blunt: A Complete Edition II. London.
Caton, S. C. 1990. “Peaks of Yemen I summon”: Poetry as Cultural Practice in a North Yemeni Tribe. Berkeley.
Finnegan, R. 1977. Oral Poetry: Its Nature, Significance, and Social Context. Cambridge.
Heinrichs, W. 1987. “Poetik, Rhetorik, Literaturkritik, Metrik und Rheimlehre.” Grundriss der arabischen Philologie II: Literaturwissenschaft, ed. H. Gätje, 177–207. Wiesbaden.
Hood, M. 1960. “The Challenge of ‘Bi-Musicality.’” Ethnomusicology 4:55–59.
Larkin, M. 2006. “Popular poetry in the post-classical period, 1150–1850.” Arabic Literature in the Post-Classical Period, eds. R. Allen and D. S. Richards, 189–242. Cambridge.
Lord, A. B. 1995. The Singer Resumes the Tale. Ed. M. L. Lord. Ithaca.
Parry, A. 1971. The Making of Homeric Verse: The Collected Papers of Milman Parry. Oxford.
Parry, M. 1930. “Studies in the Epic Technique of Oral Verse-Making. I. Homer and Homeric Style.” Harvard Studies in Classical Philology 41:73–147.
Reynolds, D. F. 1995. Heroic Poets, Poetic Heroes: The Ethnography of Performance in an Arabic Oral Epic Tradition. Ithaca.
———. 1991. “The Interplay of Genres in Oral Epic Performance: Differently Marked Discourse in a Northern Egyptian Tradition.” The Ballad and Oral Literature, ed. J. Harris, 292–317. Cambridge MA.
———. 1989. “Tradition Replacing Tradition in Egyptian Oral Epic-Singing: The Creation of a Commercial Image.” Pacific Review of Ethnomusicology 5:1–14.
Slyomovics, Susan E. 1987. The Merchant of Art: An Egyptian Hilali Oral Epic Poet in Performance. Berkeley.
Sowayan, S. A. 1985. Nabati Poetry: The Oral Poetry of Arabia. Berkeley.
Weil, G. 1958. “‘Arūḍ.” Encyclopedia of Islam, new ed., vol. 1, fasc. 11, 667–77. Leiden.


[ back ] 1. The research described here was made possible by the generous support of grants from the Center for Arabic Studies Abroad (CASA 1980–81, 1982–83); the Harvard Society of Fellows (1986–90); a Fulbright-Hays international dissertation grant (1986–87); and an American Council of Learned Societies (ACLS) Digital Humanities grant (2008–09).
[ back ] 2. ‘Classical Arabic’ refers here to a specific register of formal Arabic that is defined by its adherence to a set of grammatical norms, including a complex system of desinential case-endings. It contrasts with ‘colloquial Arabic’, a term that refers to the various spoken forms of Arabic, which differ from ‘classical Arabic’ in a number of ways, among which is the general lack of a case system. In this regard, colloquial Arabic dialects differ from classical Arabic in ways similar to how the Romance languages differ from Latin. The rules of classical Arabic prosody were codified already in the eighth century CE and were strictly adhered to with remarkable consistency until the twentieth century; whereas colloquial poetry, on the other hand, appears always to have been more flexible. Both ‘classical’ and ‘colloquial’ Arabic poetry can be composed, performed, and transmitted orally, so the linguistic register of an Arabic poem provides little evidence as to whether it was composed orally or as a written text.
[ back ] 3. Whether pre-Islamic poetry was composed within a strict metrical system or whether the extant metrical patterns were conceived of at a later date and early compilers then “regularized” the versions that have come down to us, as well as the role of stress-access versus quantitative meters, all remain topics of vigorous debate. See Weil 1958 for a concise and reliable summary; also Heinrichs 1987.
[ back ] 4. See ‘Abdallah 1988.
[ back ] 5. See Larkin 2006.
[ back ] 6. The Southern Egyptian tradition of singing Sīrat Banī Hilāl in rhymed quatrains seems to be a unique regional variant—thus references to “Arabic oral epic singing” in the remainder of this essay do not necessarily hold true for the Southern Egyptian tradition. For an account of pre-twentieth-century written records, see Reynolds 1995:1–13; for a study of the Southern Egyptian quatrain tradition, see Slyomovics 1987.
[ back ] 7. There are also proverbs and riddles that derive from the tale of the Bani Hilal.
[ back ] 8. See Reynolds 1989.
[ back ] 9. See Hood 1960.
[ back ] 10. M. Parry 1930:80 (= A. Parry 1971:272).
[ back ] 11. A number of scholars have questioned the applicability of a strict definition of ‘oral formula’ across cultures; see, for example, Finnegan 1977:58–73; Sowayan 1985:183–195; Caton 1990:86–99.
[ back ] 12. The “–I” here represents an obligatory helping vowel which occurs to prevent the clustering of three consonants.
[ back ] 13. Lit. “the mind” since Arabic often does not use possessive endings for body parts (i.e. “the hand” for “my hand”).
[ back ] 14. In Arabic, a rhyme is referred to only by its rhyming consonant, not by the vowels that precede or follow it, so I literally asked Sheikh Taha to sing on an “m-rhyme” (Ar. Mīm).
[ back ] 15. Looking back at this moment after even just a few days, I realized that my request might well have been perceived as disrespectful, primarily because it took place in front of others, and that it might have been more diplomatic to have made this request in private during a lesson. I was, however, simply carried away with the excitement of the moment. We may certainly wonder whether the various tasks Parry and Lord placed before their singers might also have been perceived as lacking in appropriate decorum.
[ back ] 16. Shaykh Taha Abu Zayd, like a number of other epic singers, incorporates the name of one of the heroes of the epic into his own personal name, in this case because it was the name of his father.
[ back ] 17. I have analyzed the interplay of poets and heroes in the Bani Hilal epic in detail in Reynolds 1995.
[ back ] 18. Shaykh Taha alternates between the more classical pronunciation of the verb ‘to say’ (pronounced with an initial ‘q’) and the colloquial pronunciation (pronounced with a ‘g’).
[ back ] 19. An epithet that indicates that he is so powerful that even the strong seek refuge with him in times of need.
[ back ] 20. Shaykh Taha here sang two rhyming hemistichs in a row, which he did not consider an “error” because the rhyme fell into place.
[ back ] 21. This is a two-syllable rhyme in which the two vowels and the final consonant are set, but the penultimate consonant is variable, marked here as X (for example, wāḍiḥ, rāyiḥ, maṭāriḥ).
[ back ] 22. Shaykh Biyeli interpolates a six-verse section on the rhyme –ān before breaking into a rapid prose summary of the subsequent events, perhaps because he started off on the wrong rhyme. He then switched to the “proper” rhyme a few moments later.
[ back ] 23. The disjunct /-ālhā/ section in these two performances is a comic passage in which the speaker describes what types of girl one should and should not marry. Shaykh Taha places it in the mouth of the Hilali Qadi who is addressing the ruler of Mecca on Rizq’s behalf; Shaykh Biyeli places it in the mouth of Ghanim who is advising Rizq to divorce Khadra since she has borne him no children and to take another wife.
[ back ] 24. This short scene is not present in Shaykh Biyeli’s performance.
[ back ] 25. This scene is not found in this performance by Shaykh Taha, but is part of other performances sung by him.
[ back ] 26. The audio recordings of both performances are available online at: Shaykh Taha 101A and Shaykh Biyeli 003A. Shaykh Taha’s performance is also presented there in Arabic transcription and English translation, as well as in the “Virtual Performance” section where one can hear the performance and read a synchronized line-by-line text.
[ back ] 27. For an analysis of one such passage, see Reynolds 1991:308–311.
[ back ] 28. Cf. Lord’s analysis of “oft repeated blocks of lines (a stable, but not fixed, core of wording)” in Lord 1995, Chapter 2 (summarized by M. L. Lord on pp. 62–63).
[ back ] 29. Blunt and Blunt 1892, reprinted in Blunt 1914:129–217.
[ back ] 30. A. Parry 1971:442.