Thread by @PhDniX

I figured it would be nice to do a thread on the presentation I gave at IQSA, so here goes!

This talk focuses on the reading traditions as present in manuscripts, a unexplored topic that I hope to devote my time to in the coming years (if funding bodies will it...).

When we talk about Quranic manuscript it's important to think of its three layers. A modern Quran has full vowels and consonantal dots.
The vowels are much later than the consonantal skeleton which in the very earliest manuscripts hardly ever receives consonantal dotting.

The ambiguities that the earliest manuscripts bring with them, mean that a variety of different readings are possible with the same consonantal skeleton (rasm). And thus many readings proliferated. Seven were only canonized in the 4th Islamic century by Ibn Mujāhid.

Today only two (or three) main readings are popular. The most dominant at all is that of Ḥafṣ the transmitter of the Kufan ʿĀṣim. After that Warš ʿan Nāfiʿ (North Africa) and in third place Qālūn ʿan Nāfiʿ (mostly Libya and Tunisia). But all 7 (and 10) are considered canonical

So what exactly is a "Reading" that the rasm allows? It is a complete system, if you follow a reading you are to follow both its ʾUṣūl "general principles" and Farš al-Ḥurīf "specific variants".

An example of the general principles is Ibn Kaṯīr's reading of the pronouns. He always added an -ū vowel to plural pronouns:

humū, ʾantumū, ʾilay-kumū, ʿalay-himū etc. This is absolutely regular and must be applied consistently if following his reading.

An example of a "Specific variant": quran.com/19/19 is read as follows by the 2 most common readings:
Hafs reads: ... li-ʾahaba ... “I am but a messenger of your lord, so that I will give you a pure son”
Warsh reads: ... li-yahaba ... “[...] so that he will give [...]”.

Ibn Mujāhid in operates centuries after the readings he canonizes. Literature suggests that before his time many other readings besides his seven were used in recitation.

Where they the most popular? How popular were others?

Ibn Mujāhid is the earliest extant systematic description. It is difficult to gain insight into the systems of readings before his canonization. While we have some earlier authors, their descriptions are often fragmentary and difficult to evaluate.

So how do we learn more?

The obvious step forward is to use primary source material. There are hundreds of (mostly unstudied) vocalised manuscripts, that represent reading tradition of that stem from before the time of the canonical readers right up to the canonization of the 7.

To illustrate what we can learn from Vocalised manuscripts, I will start by illustrating some findings about Arabe 334a, of which I have recently published an edition in the Journal of Islamic Manuscripts.
doi.org/10.1163/187846…
academia.edu/41060428/Arabe…

In its general principles, there's immediately one feature that stands out in the manuscript (something already noticed by @CellardEleonore), unlike any of the canonical readings the third person masculine pronoun is always -hū and never -hī/-hi.
academia.edu/36426137/Cella…

Other features are recognisable. The plural pronouns, for example are always long (similar to Ibn Kaṯīr and ʾAbū Jaʿfar), but the behaviour of the singular pronoun disqualifies it from being either of these two readings.

Also in specific variants, the manuscript is recognisably non-canonical. The manuscript has a reading exclusive to Yaʿqūb among the canonical readers (one of the 3 after the 7 canonical readers) and other variants that Yaʿqūb never has. So also in these variants uncanonical.

It is worth pointing out what I mean by 'non-canonical'. The tradition speaks of readings that don't follow the standard Uthmanic rasm, this is not the type of non-canonical readings I'm talking about Arabe 334a follows the standard rasm, so is both Uthmanic AND non-canonical.

If one looks at the focus on non-Uthmanic readings both in recent western scholarship that those were extremely dominant. But these do not appear in manuscripts. It is hard to understate how dominant the Uthmanic Rasm is.
However, Uthmanic non-canonical readings were very common!

So how does one go about identifying such non-canonical readings across so many hundreds of manuscripts?

To a carpenter everything looks like a nail.

So as a linguist I focus on the grammar of the Quranic readings. But I think this yields good results.

Grammatical forms such as the shape of the plural pronouns, or the treatment of hamzah or ʾimālah occur extraordinarily often in manuscripts, so you can quickly scan manuscripts at a relatively low resolution and still identify many non-canonical readings.

When we find that the generla principles are non-canonical it seems you can predict that the specific variants are that too. That is, we don't find non-canonical grammar with canonical wording.

To show off my method of "Grammar as diagnostic", I take a look at Arabe 330f.

In this manuscript we find that the lām of command does not lose its vowel if wa- or fa-precedes. Thus we find fa-li-taqum, wa-li-taʾti etc. This is different from all canonical readers but is an option mentioned by grammarians and early maʿānī al-qurʿān literature.

These non-canonical manuscripts are not just strange one-offs of some addled vocaliser who went crazy rather than putting down a standard reading, made something up himself. There are clear and recurrent patterns and they are VERY common.

Of 350 manuscripts I examined 50 share a system. The all have: unharmonized -hu with one exception they all have bi-hi.
Such manuscripts consistently also have long pronouns (and very often are written in the B.II style).

This 'mystery reading' is more common in the manuscripts than some of the canonical readings.
While traditionally popular readings like ʾAbū ʿAmr and Warš ʿan Nāfiʿ occurs a lot in these manuscripts, this mystery reading clearly occurs way more often than Ibn Kaṯīr for example.

The nice thing about this mystery reading is that it is so numerous enough that we have large overlapping sections of text, and check if the specific variants are identical too, and indeed they are! There is thus potential to reconstruct this reading lost to the literary sources.

These manuscripts with the "Unharmonized + bi-hī" reading mostly occur in B.II and D.IV manuscripts which are dated around the 3rd Islamic century, consistently after the lifetime of the canonical readers. Which brings us to the question: When did the Canon become the canon?!

It does not seem to be the case that Ibn Mujāhid's seven were accepted as the absolute only possibly acceptable readings during his lifetime. The Quran of Amajur (endowed in 262 AH), has a non-canonical reading (see: min baʿdi-hū).

cudl.lib.cam.ac.uk/view/MS-ADD-01…

Even later, the Quran of ʿAbd al-Munʿim (endowed 298 AH) likewise has a non-canonical reading (ḥaqqa jihādi-hū). As it stands, it is unclear when these non-canonical readings end, although sometime after Ibn Mujāhid his seven do become very dominant.

Vocalised manuscripts can serve as a primary source of the readings that PREDATE canonisation.
These manuscripts record reading system that are not recorded in any systematic way in the literary sources.
We must use these sources to better understand the history of the readings.

A side note that I posed at the conference: If we claim that the language of the Quran is "Classical Arabic", what do we mean by that?
The tradition gives 20 mutually exclusive answers as to what that language is, and manuscript are set to bring many more conflicting answers.

Before I forget, a big shout out to @therealsidky with whom I've been working through these vocalised manuscripts by studying their pronominal system. It's 100% a collaborative effort that I am able to discuss the 50 manuscripts that have a "unharmonized/bi-hī" reading.

@therealsidky Most manuscripts mentioned here have been digitized. I've linked some as I went along, but here are links to some more:
Arabe 334a: gallica.bnf.fr/ark:/12148/btv…
Arabe 330f: gallica.bnf.fr/ark:/12148/btv…
Arabe 329d: gallica.bnf.fr/ark:/12148/btv…

@therealsidky Arabe 5122: gallica.bnf.fr/ark:/12148/btv…
Or. 6184:
hdl.handle.net/1887.1/item:15…
Landberg 834: resolver.staatsbibliothek-berlin.de/SBB000092DF000…