Nicolai Sinai (2017, 111) notes that a Muslim tradition (seemingly found in itself recognises a shift in Qur’anic terminology, seeing ‘O you who believe’ (yā-ayyuhā lladhīna āmanū) as a Medinan phrase, and ‘O you people’ (yā-ayyuhā l-nās) as a Meccan phrase.
Nöldeke (2013[1860/1909], 99) claimed to detect a shift in the usage of al-raḥmān as a title for God, from once in the first Meccan period (Q 55:1), more frequently in the second Meccan period (about fifty times in the second Meccan period, and sixteen times in Q 19), rarely in the third Meccan period (Q 13:30, 41:2) and never in Medina. His reconstruction full does depend upon how one dates certain occurrences (Q 78:37 as second Meccan, Q 2:143, raḥīmun, as Meccan), and the distinction between al-Raḥmān as a title and as an adjective (Q 1:1, 59:22; Nöldeke seems to have missed Q 2:163). However, even if one disputes the odd verse here and there, this wouldn’t necessarily challenge the overall trend; indeed, Noldeke concedes the rare exceptions to his schematisation. But as one of, though not the only, criteria for dating surahs, in order to avoid circularity one must check and weigh the other arguments given for dating these surahs. But even if one disputes exactly where in the Meccan period to date surahs, Noldeke’s division between Meccan and Medinan seems (at least for these surahs in question) seems to typically match the Muslim tradition. At the minimum, therefore, the abundance of this term in the Meccan period and its scarcity in the Medinan period seems to be well established.
There may well also be plausible historical explanations for why a divine author might change the usage of certain words according to historical circumstance. Perhaps ‘O you people’ is appropriate in addressing a fairly uniform pagan audience in Mecca, while ‘O you who believe’ becomes necessary in Medina when a substantial number of both believers, unbelievers and hypocrites are all addressed simultaneously. Nöldeke (99) noted that some Muslim commentators on Q 17:1110 claimed that Allāh and al-Raḥmān were two separate deities; perhaps this misunderstanding caused the term to be dropped.
Broader data (inc. mean verse length)
However, as the number of such examples multiplies, and as we consider different types of data, this would increasingly suggest that genuine stylistic development is at work. Sinai (2017, 113) argues that an increase in mean verse length is a good foundation for a chronology of the Qur’an. He (118, 121-122) notes that shifts in this mean verse length correlates with another feature, viz. the way that those surahs begin, whether with the ‘Vocative’, ‘Praise of God’, ‘Reference to ‘the Scripture’ and/or the ‘sending down’ (n-z-l) of revelations’, ‘Isolated letter(s)’, ‘Temporal clause(s)’ and ‘Oath(s)’.
After introducing a few more factors (‘terminological and thematic features’, ‘correlation between mean verse length and formulaic density’, and Sadeghi’s work), Sinai (122) summarises:
The upshot of all this is that the mean verse length of Qurʾanic surahs is smoothly covariant with a host of independent stylistic, terminological, literary, and thematic features. As Behnam Sadeghi has underscored, this correlation is far too pervasive to be coincidental and therefore demands an explanation – and the simplest explanation is arguably to assume that the texts now compiled in the Qurʾanic corpus reflect different stages of a process of literary development, a process in the course of which a large number of distinct parameters would have undergone gradual and concurrent change.
Behnam Sadeghi (2011)
Behnam Sadeghi’s (2011) work, consciously building upon the pre-computer age work of Iranian scholar Mehdi Bazargan, appears significant, even if the details are difficult for me and many others in the humanities to grasp. In a nutshell, he notes the smooth development over time in the usage of a number of textual units (morphemes) in correlation with the change in average verse length.
His results are based on ‘the 28 most common morphemes in the Qurʾān’, ‘114 other common morphemes’ and ‘a list of 3693 relatively uncommon morphemes’ (Abstract, 210). The advantage of considering morphemes is that many of those in Lists A and B can be separated from any consideration of topic or theme. That is, we can avoid the objection that the subject matter is determining the choice of words, and thus different historical circumstances in Medina explain differences in the Qur’anic idiom. As Sadeghi (252) writes:
Feature List A, given in Table 5, consists of the top twenty-eight most common morphemes in the Qurʾān. It can be seen that these features are ones that can be used independently of the subject matter at hand; that is, they are in principle non-contextual. They also tend to be function features (pronouns, case-endings, etc.), as opposed to content features. … Feature List B, provided in Table 6, consists of 114 morphemes that are function features or otherwise are relatively non-contextual, and which are not included among the top twenty-eight most frequent features. These are among the most frequent 536 features in the Qurʾān. I eliminated many elements…that are firmly tied to specific themes. (Emphasis original)
Sadeghi (278) acknowledges that List C ‘reflects not only style, but also subject matter.’ He claims, however, that ‘one may choose to treat this method [of list C] as experimental, ignoring it where it contradicts the approaches of the last two sections. But where it agrees with them, there is independent corroboration, as it is highly unlikely for the agreement to be coincidental.’ Discovering a lack of corroboration may be due to the different proportions of subject matters affecting the usage of infrequent words; this is the ‘noise’ that one might expect to find. However, detecting the same ‘signal’ as in Lists A and B, is unlikely to happen by chance, and hence any corroboration should be noted. And Sadeghi (279) does claim that such corroboration is largely what we find.
Sadeghi (283) writes in his conclusion: ‘The only discernable explanation for the observed concurrent smoothness is chronological development. One who denies this conclusion has the burden of explaining the pattern in some other way.’ It is above my pay-grade to know whether Sadeghi’s findings are statistically valid and significant; maths is not my forte. However Sadeghi makes, what seems to this lay-man, a good case that one can track stylistic development in the Qur’an. Furthermore, Nicolai’s (2017, 118, 121-122) observation of correlations between mean verse length and the way in which surahs begin is itself persuasive.
Nöldeke, Theodore, Friedrich Schwally, Gotthelf Bergsträßer, and Otto Pretzl. The History of the Qur’ān. Translated by Wolfgang H. Behn. Leiden: Brill, 2013.
Sadeghi, Behnam. “The Chronology of the Qur’ān: A Stylometric Research Program.” Arabica 58, no. 3 (2011): 210-99.
Sinai, Nicolai. The Qur’an: a historical critical introduction. Edinburgh: Edinburgh University Press, 2017.