Measuring variation in the early tradition

In this first blog I begin with a discussion of what our methods might be able to show us about the early transmission of the written tradition. The case I take up involves the Muwaṭṭaʾ (“the well-trodden path”) of Imam Mālik b. Anas (d. 796), which is also a founding book for Islamic law and Prophetic tradition (Hadith).

What follows is quite detailed. Not all blogs will ask as much of readers. But I think it is worthwhile to start with the early written tradition, where a lot of the debates and confusion lie. In particular, I am curious to probe how much variation there is between different transmissions of an author’s text.

Anyone interested in how the written Arabic tradition emerged, in the decades before and after Mālik’s lifetime, needs to take seriously the ideas of Gregor Schoeler, who has illuminated the complex ways that the written tradition developed. His most important insights relate to the inseparable mixture in early Islam of modes of oral and written transmission. This situation is reflected in the existence of hypomnema (draft notes and notebooks), which passed into syngramma (actual manuscript books). What began as aides-mémoire became in many cases, by the ninth century, fixed, “published” books. Schoeler was not the first to offer such nuance, but his explanation was the most forceful and clearest, at least in Europe and North America (his work has been published in German, French and English). By contrast, much earlier scholarship opposed oral and written transmission.

Schoeler then has made us think harder about the cultural meaning of “books” in the early periods of Arabic and Islamic history. Were hypomnema properly speaking “books”? And likewise authorship – where transmission is complicated, do we need to be more careful in attributing ideas to an original author or a text, rather than a student who may have changed it? Similarly, where students maybe had different lecture notes covering the same topics, for example, what authority do we assign to the original teacher/“author”? How do we distinguish his voice from that of his student/transmitters? And generally speaking, how are we meant to read, cite, and understand the many works from the early periods that sit on shelves now – including the Muwaṭṭaʾ?

The complexity of the Muwaṭṭaʾ’s transmission has been known for centuries: the Mālikī jurist Qāḍī ʿIyāḍ (d. 1149), for example, referred to 20 different transmissions of the Muwaṭṭaʾ, although he said his contemporaries knew 30. Modern editors and scholars often mention large numbers, including Yasin Dutton (in his 1999 book, The Origins of Islamic Law: where he lists 9 or 10 “versions,” either in complete form or as fragments). Mālik reportedly updated his lectures throughout his life, this explaining the changes. But despite summaries of differences (eg. by Ignaz Goldziher), there has been very little by way of detail proposed on how precisely these texts differ.

We can now furnish far more evidence than previously for these differences – and also for differences in the transmission of other texts. This should provide a new evidentiary basis for addressing the bigger questions and help us to engage with what have often been tentative proposals.

I have begun to compare three different texts bearing the Muwaṭṭaʾ title pulled originally from Shamela and now in the OpenITI collection. They are credited to:

  1. Yaḥyā ibn Yaḥyā al-Laythī (d. 234/848-49), who is by far the best known transmitter, and the basis for the modern English translation of the Muwaṭṭaʾ by Aisha Abdurrahman Bewley. Yaḥyā’s relationship with Mālik has been debated in scholarship (there is one view that he studied with Mālik), but as Maribel Fierro has argued, Yaḥyā was “undoubtedly the pupil of Mālik’s pupils such as the Egyptians Ibn Wahb (d. 197/812) and Ibn al-Ḳāsim (d. 191/806).”[1] Yaḥyā is considered to have introduced Mālik’s Muwaṭṭaʾinto al-Andalus. His transmission of this work became canonical in the Islamic West and was also widely diffused in the East. It formed the basis of many commentaries. Our electronic edition is based on the printed edition, which was edited by Muḥammad Fuʾad ʿAbd al-Bāqī (Beirut: Dār Iḥyā al-Turāth, 1985). The text in our corpus contains 146,475 words (OpenITI URI [uniform resource identifier]: 0179MalikIbnAnas.Muwatta.Shamela0001699-ara1).
  2. Abū Muṣʿab Aḥmad b. Abī Bakr al-Zuhrī (d. 242/856). Abū Muṣʿab, from Medina, was reportedly the last person to have related the Muwaṭṭaʾ from Mālik. Our edition is that edited by Bashshār ʿAwwād Maʿrūf and Maḥmūd Muḥammad (1412 AH). This electronic version is based upon a manuscript from Hyderabad, India that has recently been edited and published. Various portions of this transmission also exist in manuscript form in Tunis, Qayrawān, Damascus, and Dublin. Word count: 155,402 (OpenITI URI: 0179MalikIbnAnas.Muwatta.Shamela0008140-ara1).
  3. Muḥammad al-Shaybānī (d. 189/805). Al-Shaybānī was a younger contemporary of Mālik, and also a transmitter from Abū Ḥanīfa (and indeed, often reckoned a founder of Ḥanafism). One of the ostensibly great virtues of this text is that al-Shaybānī was a contemporary of Mālik, and since Goldziher, at least, al-Shaybānī’s text has been described as a versionof the Muwattaʾ.[2] Our edition is that edited by ʿAbd al-Wahhāb ʿAbd al-Laṭīf (Beirut?: Maktaba al-ʿIlmiyya, n.d.).Word count: 63,443 (OpenITI URI: 0179MalikIbnAnas.Muwatta.Shamela0016050-ara1)

All three of these transmissions were among those used by Dutton in his The Origins of Islamic Law.

Two issues do become apparent when we compare the electronic versions[3] of these different texts using our software, passim.

The first issue has been generally known but neither in extent nor in detail. And that is that the wording of the Yaḥyā ibn Yaḥyā’s transmission is close to that of Abū Muṣʿab. But the order is significantly different. Our data visualization, just below, compares the two transmissions.

Graph 1: Yaḥyā ibn Yaḥyā’s transmission is laid out on the x-axis on the top in 100-word segments, and Abū Muṣʿab’s is laid out similarly in the bottom graph. The yellow lines represent where passim, detects common passages of texts (the longer the red line, the longer the passage).

The difference in order can be seen by looking at the yellow lines. When they cross, we can detect rearrangement – there are at least seven different parts of the texts that are so rearranged:

Graph 2: As in graph 1, but circles indicate divergences in sequence between common passages.

Such data and graphs are just the start of enquiry, but they show the close relationship. In terms of overlap, we find that about 77% of Yaḥyā’s text can be found in that of Abū Muṣʿab, and that about 72% of Abū Muṣʿab’s text can be found in that of Yaḥyā (the different percentages arise because the two books are of slightly different sizes – Abū Muṣʿab’s book is larger, so the common passages represent less of it). On their own, the data and visualization says nothing about the stemmatic relationship between the texts. But they do open up areas for investigation using the aligned texts that accompany our visualisations as we can see precisely where the differences lie, including at the bigger structural level. This means that so-called “close” reading can now be guided by “distant” reading.

The second issue to become apparent is that it is very hard to reckon al-Shaybānī’s text as a straight-forward “version” or “recension” of those of Yaḥyā and Abū Muṣʿab. If we compare al-Shaybānī’s text to either book, the density of relationship is much less. Our data shows specifically that less than 25% of Shaybānī’s text matches that of either of the other Muwaṭṭaʾs. Likewise, only about 10% of these (longer texts) matches the Muwaṭṭaʾ attributed to al-Shaybānī.

Graphically, in fact, al-Shaybānī’s text looks a lot more like a commentary on the Muwaṭṭaʾ than a straight forward version of it.

Graph 3: al-Shaybānī’s text is laid out on the x-axis above, and the Muwaṭṭaʾ transmitted by Yaḥyā is below.

Graph 4: al-Shaybānī’s text is again laid out on the x-axis above, now with the Muwaṭṭaʾ transmitted by Abū Muṣʿab below.

I say that al-Shaybānī’s text looks like a “commentary” because often a commentator reuses words from a text, followed by his/her words that explain it. Graphically, the commentator’s words would result in white on the graph. So, you would expect to see shared passages but not continuously. For example, in Graphs 5 and 6, just below, we have comparisons of a self-proclaimed commentary on the Muwaṭṭaʾ al-Tamhīd li-mā fī al-Muwaṭṭaʾ min al-maʿānī wa l-asānīd, by the Cordovan Ibn ʿAbd al-Barr (d. 463/1070) — with the transmissions of the Muwaṭṭaʾ by Yaḥyā and Abū Muṣʿab.

Graph 5: Yaḥyā’s text of the Muwaṭṭaʾ is laid out on the x-axis above, with Ibn ʿAbd al-Barr’s commentary below (OpenITI URI: 0463IbnCabdBarr.TamhidMuwatta.JK000585-ara1).

Graph 6: Abū Muṣʿab’s Muwaṭṭaʾ is in the top part of the graph, and Ibn ʿAbd al-Barr’s commentary is laid out below.

Closer reading is required to say more of a specific nature about how we should in fact classify al-Shaybānī’s text, but the data does raise questions, especially about the relationship between it and Mālik’s teaching. One possibility is that al-Shaybānī’s text is a commentary. But al-Shaybānī had direct access to Mālik. Was he commenting already in Mālik’s lifetime? It would be of note for the tradition at large to foreground the possibility of commentaries preceding finalized versions. But there are other possibilities. For example, was al-Shaybānī’s text possibly an early version, and those of Abū Muṣʿab (also likely a contemporary) and Yaḥyā the result of different participation in lectures, and the subsequent formation of a consensus? More digital files of other editions and manuscript transmissions would be helpful. And surely specialists in Islamic law and tradition can offer their own, expert readings of the data as it pertains to al-Shaybānī’s text and the transmission of the Muwaṭṭaʾ.

It is noteworthy that the data we find on the Yaḥyā and Abū Muṣʿab Muwaṭṭaʾs is consistent with the sort of complex environment described by Schoeler, where one would expect to find both a lot of precision (insofar as students transmitted faithfully) and change (insofar as Mālik adjusted his readings across time). And it is also noteworthy that Schoeler seems to have intuited problems with referring to al-Shaybānī’s text as a “recension” of the Muwaṭṭaʾ. He uses the term with some obvious misgivings, and in fact, his description of al-Shaybānī’s text makes it sound a lot like a (critical) commentary on the Muwaṭṭaʾ.[4] Here, I think, intuition would benefit from a new evidentiary basis.

Often, discussions of texts such as the Muwaṭṭaʾ tend to revolve around questions of “authenticity.” Instead, I think we can ask ourselves – seeing textual variations such as this – far more interesting questions relating to the development of the written tradition. These pertain to the varieties of ways of (re-)using existing written texts, the malleability of the written tradition in different times and locations and for different types of texts, and about how producers of texts concretely worked. We have multiple “versions” of many other early books in our corpus and also fragments of books reproduced in other books for which we have electronic files. These should also be studied alongside those for the Muwaṭṭaʾ to engage with these bigger issues.

In future blogs, then, I will consider other cases.

So, Schoeler makes us think harder about the cultural meaning of “books” and “authors,” with one scenario in mind. But this was just a start. What others can we discern?


[1] Fierro, “Yaḥyā b. Yaḥyā al-Laythī,” Encyclopaedia of Islam, second edition.

[2] As Aḥmad ibn ʿAbd al-ʿAzīz Al Mubārak says in his introduction to Bewley’s translation: “Another version of the Muwaṭṭaʾ worthy of mention is that of Imām Muḥammad ibn al-Ḥasan ash-Shaybānī which has great distinction. It includes the transmission of many traditions intended to support his madhhab and the madhhab of his Imām, Abū Ḥanīfa. Sometimes he mentions that Abū Ḥanīfa agrees with Mālik regarding the matter under discussion.” Joseph Schacht lists the “recension” and “edition” of Shaybānī in his elaboration of the Muwaṭṭaʾ’s history in Encycopaedia of Islam articles.

[3] The predominant way of generating digital texts has been through double-keying, which is extremely accurate. For a study on the matter (albeit for German), see Susanne Haaf, Frank Wiegand, and Alexander Geyken, “Measuring the Correctness of Double-Keying: Error Classification and Quality Control in a Large Corpus of TEI-Annotated Historical Text.” Journal of the Text Encoding Initiative, no. Issue 4 (March 8, 2013). 

[4] Schoeler writes in The Genesis of Literature in Islam (trans. Shawkat Toorawa, 2009; p. 78) of “the two most important recensions,” and that the “second [recension, after that of Yaḥyā b. Yaḥyā al-Laythī] is the recension of, or rather the reworking (Fr. remaniement) by, the Ḥanafī Muḥammad ibn al-Ḥasan al-Shaybānī (d. 189/804), who was a student of Mālik’s, one that distinguishes itself above all by its critical comments about Mālik and about the teaching of law in Medina. Al-Shaybānī’s comments appear at the end of most of the chapters and are not always in agreement with Mālik’s juridical opinion, or with the ḥadīths Mālik quotes. Notwithstanding the fact that he is a transmitter of Mālik, al-Shaybānī constantly has recourse to the juridical opinions of his Ḥanafī colleagues, which very often contradict those of Mālik, and to the opinion of his teacher, Abū Ḥanīfah (qawl Abī Ḥanīfah), with which he always agrees.”




Share this: