How was the Qur’an Composed? Part 2 – Assessing Andy Bannister’s Oral Formulaic Approach

The late Nabeel Qureshi gave several lectures at Biola University some years ago, at least one of which focused on the Qur’an. In the lecture, about 1 hour and ten minutes in, he talked about the oral-formulaic view of the Qur’an’s composition. He felt it is the best apologetic against the Qur’an that no one was using.

This fact intrigued me, so I began the research. The primary reason for my research was academic and not apologetic, but I soon discovered it certainly did have an apologetic payoff. In my last blog, I gave evidence that the traditional view and critical views fell short of giving an adequate answer to the Qur’an’s composition.

In this blog, I will finish my two-part series of Qur’an’s composition by summarizing the oral-formulaic view, offered by Andy Bannister. when I stumbled on this approach, I found it to be very insightful. Bannister’s utilization of Oral Literary Theory (OLT) to answer questions that other scholars have yet to answer regarding the oral characteristics found in the Qur’an is possibly ground-breaking. 

Oral-Formulaic Approach

To discover whether a literary work was orally composed, the researcher looks for elements like formulaic language, performance variants, and elusive references located in a text which assumes the audience knows what is being talked about, which are all signs of orality. When combined with strong evidence of an already rich, established oral culture both before and after the document in question was written, all these factors offer strong evidence of orality.[1]

There is also strong evidence that Muhammad borrowed from Biblicist material that was available to him, and transmitted the information orally to his hearers, piecemeal, over twenty-three years.[2] I will present Bannister’s evidence that the words of the Qur’an were composed live in oral performance as Muhammad was answering questions from his audience. The central thesis of the oral-formulaic approach holds that there is formulaic borrowing throughout the Qur’an, resulting in repeated phrasing indicating oral composition. To prove such a thesis, Bannister sets out to explore whether the Qur’an records language originally composed through oral performance, examining the Arabic text by using a computer-based approach to carry out his search.[3]

Computer Analysis 

How does one determine if the Qur’an, or any document for that matter, is an oral document? Bannister attempts to show that OLT provides the theoretical basis for determining whether the Qur’an is an oral document or not.

Historically in OLT, the researcher examines only a portion of the text when doing a formulaic analysis of a book. For example, in determining that Homer’s Iliad and Odyssey were oral works, Milman Parry, the progenitor of the theory, examined the first twenty-five lines of those books. Similarly, the formulaic density for a monumental task like examining Beowulf was determined in the same way. All the analysis mentioned above was done by using small sample sizes to arrive at conclusions. This fact raises a red flag as to accuracy, with the small sample size serving as the criteria needed to make a judgment on the whole text. However, by doing a computer analysis, it “solves the sampling problem.”[4]

First Query

To what extent is the Qur’an a product of orality? To answer this question, Bannister performed his first research query to determine the overall level of formulaic material located in  the Qur’an. To do this, his initial examination was to create a formulaic map by looking for bases and roots found in the Qur’an. The goal was to find out the percentage of root and base sequences that are “repeated elsewhere in the Qur’an.”[5]

In this study, he examined both five-base sequences and three-base sequences.[6] When he used the five-base sequence, he discovered that the Qur’an has a 23.55% formulaic score, whereas a three-base sequence rendered a 52.18% formulaic mark. For a written work to be considered an oral composition, it must surpass the minimum formulaic threshold of 20%.[7] This initial finding points to the possibility that there is high certainty that both five-base and three-base sequences found in the Qur’an reflect the characteristics of oral composition.

Another data point uncovered in Bannister’s three-base analysis is the difference in formulaic density between the Meccan suras and the Medinan suras. The Meccan suras had a formulaic density of 39.4%, while the Medinan suras posted a score of 55.01%. Bannister attributes this discrepancy to a few possible reasons. One possible reason is “perhaps the lower use of formulaic diction in the Meccan suras might indicate Muhammad’s use of pre-Islamic diction.”[8] Another possible reason was that the Meccan and Medinan surahs “were transmitted, collected or redacted differently,” resulting in “these differences somehow affecting the fundamental function of the suras.”[9] He found similar results when examining the root sequences two, three, four, and five roots long.[10]

Second Query

Unfortunately, the results of the first query do not provide conclusive evidence, by Parry’s standards, that the text is an oral one. Needed is a further examination of the formulaic systems found in the Qur’an because there needs to be “a stronger indicator of orality in a given text … not merely of repetitions, but of systems, networks of similar formulas strongly suggesting the existence of an entire formulaic diction, a poetic language as it were, lying behind a text.”[11]

Among the litany of statistical details this query uncovered, the results produced three critical findings. First, the query found that formulaic material occurs regularly throughout the Qur’an when doing a verse-by-verse query. Bannister examined the ten most formulaic suras. The most formulaic sura in the Qur’an, Sura 61, has a formulaic density of 77.8%, with no verses falling below 55.44%, except one verse. Verse 13 has a formulaic density of only 11.11%. This discrepancy is where Richard Bell’s analysis of the Qur’an is particularly helpful to explain why the formulaic density may be low in just one sura where the others are so rich in formulaic language. The other nine of the ten most formulaic suras in the Qur’an exhibit similar characteristics. For example, Sura 69, having the third highest formulaic density, consists of suras with high formulaic density, with three exceptions, while verse 36 has no formulaic density.

Second, an explanation is offered as to why there is a drop in formulaic density in individual suras. Bell’s hypothesis is that the Qur’an was delivered piecemeal throughout twenty-three years of the revelations received by Muhammad, then when compiled, the revelations were stitched together to form the Qur’an we have today. The areas where there is either low or no formulaic density may be where the compilers of the Qur’an added phrases so the information they collected could make sense. Bannister selected eight suras, four Meccan and four Medinan suras, where he compared the computer results with those of Bell. The comparison yielded results that give strong evidence that the low number of verses which have little to no formulaic density points to the possibility that there has been some redaction.

Third, there is a significant difference between Meccan and Medinan suras. The query showed that the Meccan suras were less formulaic than the Medinan suras. For example, the average formulaic density for Meccan suras was 39.4%, while the average for Medinan suras was 55.01%. The computer shows apparent “stylistic differences” between the two sets of suras.[12] One possible reason for this is that the “‘Meccan’ group of suras are more fragmentary and have undergone a more complexed process of transmission, and redaction than have the ‘Medinan’ suras.”[13] One possible answer regarding the Medinan suras is that they “may perhaps represent more coherent, longer sets of material, perhaps better reflective of the original oral mode in which the Qur’anic material was first preached and taught.”[14] This is only a hypothesis, and more research needs to be done to find concrete answers.

Third Query

Whereas the first two queries turned up considerable formulaic diction, the third query is where Bannister searched for formulaic systems, which are “networks of formulas, showing the kind of flexibility which might have allowed its performer to adapt formulaic phraseology quickly and easily for a variety of contexts.”[15] One caveat regarding this query is that it was not based on sequences that were exactly the same Arabic bases and roots because when attempting to discover systems, a new approach is needed. Relying on Parry and Lord, Bannister asserts that when one locates formulaic systems, one is almost assured that the document under examination is rooted in oral performance.[16] To garner this conclusion, Bannister looked for relationships between the sequences found in the Qur’an by using pattern matches to look for systems, as well as using them to analyze frequency and diversity and then search for root sequences to solidify his conclusion.

The search found that not only were there “direct verbal repetitions found, but formulaic ‘families,’ groups of formulaic sequences with 80% or 60% of their roots in the same position.” The results pointed to a significant amount of formulaic families located in the Qur’an.

Fourth Query: Iblis and Adam

The question arises, namely, can the information generated by the computer be trusted? To test the material, Bannister used a line-by-line analysis done by hand, which was the same method used by Parry and Lord, and many of the oral literary scholars after them, and then cross-checked the information with a concordance, to examine whether the Qur’an was orally composed. This provides the control needed to evaluate the computerized study.[17]

Bannister chose the Iblis and Adam story, mentioned seven times in the Qur’an, as the sample from the Qur’an to provide the material for his inquiry. This story depicts Allah as the creator. Angels are commanded to bow before Adam to which they did. Iblis, however, refuses to obey Allah’s command, so Allah, in return, questions his disobedience, then goes on to curse him because he criticizes mankind. Iblis responds by asking Allah for a respite until the Day of Judgment, to which Allah obliges. Despite the grace shown to him, Iblis decides he will lead men and women astray. The story concludes with a designation to send Iblis and those who follow him to hell.[18]

The reason Bannister chose the story was because of the “narrative structure of the Quranic versions of the Iblis and Adam story.”[19] The story of Iblis and Adam is not original to the Qur’an, as this story is depicted in both Jewish and Christian traditions. The core of the story is found in both Biblicist traditions as well as in the Qur’an. What makes these tellings prime for examination is that the Iblis and Adam passages in the Qur’an are told in various lengths, ranging from one to fourteen verses. Each telling retains the core elements of the story, but “what makes the increased length is the ornamentation of the core with additional narrative detail.”[20] This follows the thesis of OLT where the poet is thought to improvise when telling the story by ornamenting details to tell a more compelling story.

Bannister computed the results of his line-by-line analysis gainst the computerized results, by searching for “verbal repetitions and system-based formulas … using a concordance.”[21] The line-by-line analysis found that the total formulaic density for base sequences ranged from 45.45% to 91.75%, while the total formulaic density for root sequences ranged from 48.89% to 89.23%, which is well above the 20% threshold.[22] The line-by-line analysis confirmed the computerized data, with a noted difference. The line-by-line analysis yielded results that were higher than the automated results. For example, the computer-based analysis regarding the total formulaic density for base sequences ranged from 43.86% to 85.57%, while the range for the total formulaic density for root sequences ranged from 53.85% to 95.5%.  This fact means that the automated results increase one’s confidence in their accuracy, that in fact, much of the Qur’an was composed extemporaneously using oral formulae.[23]

Current and Further Research

Any serious contribution to the field happens when other scholars utilize pertinent information from the original idea and use it to make a further impact. One example of someone taking Bannister’s work and using it to develop a new understanding of the Qur’an, in this case, its theology, is Mark Durie’s recent book, already mentioned, titled The Qur’an and Its Biblical Reflexes: Investigations into the Genesis of a Religion. He argues that there are implications to understanding how the Qur’an’s theology is to be understood because “the woven oral-formulaic character of the Quranic text impacts the way a theological analysis of the Quran must be presented.”[24] If the Qur’an was compiled in the way Bannister suggests, “the Qur’an then offers a descriptive model of the performer’s theological mental map,” making “Qur’anic Theology … a cognitive construct.”[25]

Bannister’s work also provides a way to understand how the theological ideas in the Qur’an grew, developed, and changed over time throughout its composition, shedding new light on how things like the Qur’an’s understanding of eschatology, Qur’anic chronology, its understanding of monotheism, and the doctrine of Rasulogy and Prophetology developed over the ministry of the Messenger.[26] To answer the question as to why these theological ideas evolve, Durie draws upon the rules of OLT by pointing out that “the formulaic repertoire of an oral performer is not static, but is constantly evolving. Each new performance has the potential to bring innovation and alteration to the repertoire.”[27] As someone poses a new question to the Messenger, an answer is done on the fly, and in doing so, the Messenger can create a fresh new idea that builds on old ideas. Durie offers “textual evidence [linking] very different phases in the production of the Quran … .”[28]

Other possible fruits of Bannister’s work could be used to answer questions that scholars continue to ask about the oral elements of the Qur’an. For instance, the groundbreaking Quran Gateway, developed as a result of Bannister’s study, is a handy tool for scholars to access online.[29] It is “an independent and scholarly resource” providing tools to enable “researchers to examine, search, and analyze Islam’s core text …” as well as  “historical manuscripts.”[30] It can help answer questions that need answers as well as pose new questions.

Also, OLT can be useful for examining needed answers to questions posed regarding the content of Arabic poetry, the Sira, and Hadith literature. To analyze the oral-formulaic content of these documents, someone needs to develop software tools similar to Bannister’s computerized analysis of the Qur’an, so these books can be analyzed in the same way as the Qur’an.

No doubt, the Iliad and Odyssey are two of the most influential works in the Western literary canon, typically attributed to Homer. If Homer orally composed these works, given their complexity, it points to his genius. For instance, “the oral poet had an abundant repertoire of epithets diversified enough to provide an epithet for any metrical exigency that night arise as he stitched his story together.”[31] Consequently, Bannister’s research could lead scholars to formulate fresh insights into the person of Muhammad, elevating him to the status of poetic genius instead of the illiterate messenger who serves as a mere receptor of Allah’s message.[32] One must keep in mind, if the oral-formulaic analysis is true, that not only is there an oral tradition behind the text, there is also an oral bard who originally performed the information found in the text. This makes the Qur’an a human book and not a divinely inspired one.


The conclusion drawn from Bannister is that there is a high degree of likelihood that the composition of the Qur’an was the product of an oral performer. The computer analysis produced by Bannister provides scholars with material to analyze in the Qur’an that was not available to scholars previously.



[1] Bannister. YouTube lecture


[2] The following statement attests that the Bible written in Arabic came after the death of Muhammad. It is also unlikely that other Biblicist material was translated in Arabic as well. “Nevertheless, as we have seen, there is as yet no convincing evidence for the existence of any extended part of the Bible in written Arabic prior to the rise of Islam. …  the earliest time at which it would have been feasible for Arabic-speaking Jews and Christians to undertake a translation of the Bible (or parts of it) into written Arabic was the mid to late seventh century. The project would at that point have been possible either in tandem with, or in response to the Muslim project after Muhammad’s death to collect and to publish the Arabic Qur’an as a fully written scripture. But it is more likely that the first written Bible translations were made in the eighth century, and outside of Arabia.” The Bible in Arabic: The Scriptures of the “People of the Book” by Sidney H. Griffith (Princeton: Princeton University. 2013), 90


[3] Bannister. An Oral-Formulaic Study of the Qur’an. 243


[4] Bannister. An Oral-Formulaic Study of the Qur’an. 131


[5] Bannister. An Oral-Formulaic Study of the Qur’an. 139


[6] According to Bannister, “Three bases represent arguably the shortest length where a meaning phrase or formula can be constructed. To ensure the thoroughness of the analysis, larger base sequences were chosen. Hence the five-base sequence recorded here.” (139)


[7] Lord. The Singer of Tales.


[8] Bannister. An Oral-Formulaic Study of the Qur’an. 180


[9] Bannister. An Oral-Formulaic Study of the Qur’an. 145


[10] Bannister. An Oral-Formulaic Study of the Qur’an. 133


[11] Bannister. An Oral-Formulaic Study of the Qur’an. 155

[12] Bannister. An Oral-Formulaic Study of the Qur’an. 192


[13] Bannister. An Oral-Formulaic Study of the Qur’an. 192


[14]  Bannister. An Oral-Formulaic Study of the Qur’an. 192


[15] Bannister. An Oral-Formulaic Study of the Quran. 237


[16] Bannister. An Oral-Formulaic Study of the Qur’an. 207


[17] Bannister. An Oral-Formulaic Study of the Qur’an. 243


[18] Bannister. An Oral-Formulaic Study of the Qur’an 8


[19] Bannister. An Oral-Formulaic Study of the Qur’an 7


[20] Bannister. An Oral-Formulaic Study of the Qur’an 10


[21] Bannister. An Oral-Formulaic Study of the Qur’an. 244


[22] Bannister. An Oral-Formulaic Study of the Qur’an. 248


[23] Bannister. An Oral-Formulaic Study of the Qur’an. 266


[24] Mark Durie. The Qur’an and Its Biblical Reflexes: Investigations into the Genesis of a Religion. (New York, Lexington Books, 2018), 33


[25] Durie.

[26] Durie. 34

[27] Durie.

[28] Durie.


[29] The Quran Gateway ( goes beyond the expertise of Andrew Bannister, and has a consortium of scholars who assume the role of “advisory council.” The scholars serving in this role are Mehdi Azaiez, Emran El-Badawi, François Déroche, Fred Donner, Alba Fedeli, Asma Hilali, Davidson MacLaren, Thomas Milo, Gerd-R Puin, Gabriel Said Reynolds, Peter Riddell, and Nicolai Sinai. Accessed on February 16, 2019.




[31] Walter J. Ong. Orality and Literacy. (New York, Routledge, 2000), 23


[32] This idea was taken from a conversation I had with Andrew Bannister on Oct. 24, 2019


Leave a Reply

Your email address will not be published. Required fields are marked *