Appendix A Notes on Orthography, Phonology and Morphology of the Quranic Consonantal Text

In: Quranic Arabic
Marijn van Putten
Search for other papers by Marijn van Putten in
Current site
Google Scholar
Open Access

A.1 Introduction

This appendix serves as a more detailed discussion for some of the topics of the language of the Quranic Consonantal Text that have come up throughout this book. As previous works on the language and orthography of the Quran have mostly relied on the Cairo Edition, which is not always an accurate reflection of the Uthmanic Text, this appendix aims to add some more detailed discussion to questions of orthography, phonology and morphology of Quranic Arabic. Throughout the book there are several references to this appendix, but I have also included topics of note here which do not receive direct discussion in the book. It is hoped that this appendix can function as a short but useful guide to some of the main features of Quranic Arabic on its own. In some cases, discussions here rely on observations and generalizations of the orthography found in early Quranic manuscripts. Whenever I do so, I refer to Appendix B, which is a list of relevant tables that compares the orthography of certain words across early Quranic manuscripts.

A.2 Orthography

The Quranic orthography was studied in great detail by Werner Diem (1976; 1979; 1980; 1982) in a series of highly insightful and in-depth articles which trace the rise and development of Quranic orthographic practice from its Nabataean Aramaic origins. Diem exclusively relied on the orthography as found in the Cairo Edition, which has occasionally caused him to draw the wrong conclusions about the principles of Quranic orthography as they must have been present in the UT. Quite often, we find that early Quranic manuscripts consistently agree with each other on certain topics of orthographic practice, where the Cairo Edition differs from this practice. In this section I will discuss the main orthographic practices of Quranic Arabic, which will necessarily overlap on occasion with the observations made by Diem.

A.2.1 The Spelling of ā

In Pre-Islamic Arabic written in the Nabataean script, and transitional Nabataeao-Arabic there was no way to write word-internal ā (unlike ī, and ū). With the loss of the glottal stop in Quranic Arabic, the ʔalif gave rise to a new word-internal mater lectionis for /ā/ (Diem 1979, § 60–68; van Putten 2018).1 In the QCT, the use of ʔalif for writing /ā/ is still largely optional, and it is one of the main points of disagreement between different Quranic manuscripts (van Putten 2019c, 281–286). Despite this unstable status of the spelling, several generalizations can be made about its spelling.

In the QCT defective spellings of ā are very common, but highly uncommon in words of the shape CāC and CāC̄ (Diem 1979, § 67). The exception to this being the spelling of the verb ‘to say’ /qāl/ which in early Islamic documents is almost without exception spelled defectively ‮قل‬‎, an archaic spelling retained in this one high frequency word, the same is true for, for example its plural form ‮قلوا‬‎.2 In the CE the special status of the verb qāla has almost completely disappeared, and is generally spelled plene as other verbs of this type. However, ‮قل‬‎ recited as qāla occurs in Q21:112, Q23:112,114 and Q43:24.3

The defective spelling of the feminine plural ending /-āt/ is standard in the Uthmanic orthography. In the CE only ‮بنات‬‎ ‘daughters’ is regularly spelled plene. Besides these there are three specific exceptions: Q41:12 ‮سموات‬‎ ‘heavens’ (versus 189 times that it is spelled ‮سموت‬‎), Q41:16 ‮ايام نحسات‬‎ ‘unfortunate days’ and Q42:22 ‮روضات الجنات‬‎ ‘the flowering meadows of the gardens’. These unique exceptions of these verses do not seem to be a feature of the UT. For ‮الجنات‬‎ /ǧannāt/ ‘gardens’, the plene spelling is regular in early manuscripts, just like ‮بنات‬‎ /banāt/ ‘daughters’, and not just used in the position Q42:22. It appears that if the stem + the feminine plural ending would only be three letters long if the feminine plural ending was spelled defectively, the ending is spelled plene (van Putten 2019c, 284). As for ‮سموات‬‎, ‮نحسات‬‎ and ‮روضات‬‎, these are normally spelled defectively in these positions, following the regular rule of defective spelling of -āt in early manuscripts (see B.1).4

In the QCT, ā two syllables removed from the stress, such as in plural G-stem active participles and duals CāCiCū/īn, CāCiCāt and CāCiCān/ayn (as well as ‮العلمين‬‎ /al-ʕālamīn/) are consistently spelled defectively. Diem (1979, § 67.2; 1980, § 105) notices this rule too, but observes that in the CE, hollow verbs break this patterns and are consistently spelled plene (as are the singulars), e.g. Q7:4 ‮قايلون‬‎ ‘sleeping at noon’. This is however an idiosyncracy of the CE. In early Quranic manuscripts these words simply follow the same rule as other plurals of this pattern and are spelled defectively (van Putten 2018, 108 f.).

The vocative prefix /yā-/ is consistently spelled defectively throughout the Quran, and this is without exception, e.g. ‮يموسى‬‎ ‘O Mūsā’ (Q2:55 and passim), ‮يمريم‬‎ ‘O Maryam’ (Q3:37 and passim), etc.5

Whenever the 1pl. suffix -nā is followed by any other clitic, it is consistently spelled defectively, e.g. ‮رزقنهم‬‎ ‘we provided them’ (Q2:3), ‮ارسلنك‬‎ ‘we have sent you’ (Q2:119).

A.2.2 Questions of Double yāʔ, wāw and ʔalif

Diem (1979, § 37–43) discusses the avoidance of double matres yāʔ and wāw in detail, and argues that the sequences of and are typically written with only a single yāʔ and wāw respectively, whereas other phonetic sequences may still have these two consonants in a row. However, the facts as they appear in the CE are not very representative of the UT, and as a result the analysis does not hold up.

For the ‮ى‬‎, Diem cites cases such as CE ‮ولى‬‎ /waliyy-ī/ ‘my friend’ (Q7:196; Q12:101); CE ‮يحى‬‎ /yuḥyī/ ‘he revives’ (Q2:73) and CE ‮يستحى‬‎ /yastaḥyī/ ‘he is ashamed’ (Q2:26). However, in early Quranic manuscripts all of these are consistently spelled with two yāʔs, and therefore the UT had two yāʔs (see B.2).6



Q7:196, Q12:101



e.g. Q2:73



e.g. Q2:26






Q15:23, Q50:43



With these forms shown to be innovations of the CE, the amount of examples where a single ‮ى‬‎ is used to write a sequence /yī/ becomes very small, whereas there are several more examples where a double ‮ى‬‎ is used even in de CE, e.g. ‮يحييكم‬‎ /yuḥyī-kum/ ‘he revives you’ (Q2:28; Q8:24; Q22:66; Q30:40; Q45:26); ‮يحيين‬‎ /yuḥyī-n/ ‘he revives me’ (Q26:81); ‮يحييها‬‎ /yuḥyī-hā/ ‘he will give them life’ (Q36:79); ‮حييتم‬‎ /ḥuyyītum/ ‘you are greeted’ (Q4:86); ‮افعيينا‬‎ /ʔa-fa-ʕayīnā/ ‘where we then tired?’ (Q50:15) and ‮عليين‬‎ /ʕilliyyīn/ ‘Elyon’ (Q83:18).

Diem (1979, § 41) considers the outcome of baʔīs (Q7:165) spelled ‮بيس‬‎ to be a reflection of /bayīs/, but it seems doubtful that this is the correct analysis. First, it is not clear that *baʔīs is the word which ‮بيس‬‎ is supposed to represent, as in the canonical reading traditions it is variously recited as bīsin, biʔsin, bayʔasin and baʔīsin (Ibn al-Ǧazarī, § 3150).7 *biʔsin would of course yield /bīs/, for which ‮بيس‬‎ is the only acceptable spelling, and *bayʔasin would presumably yield /bayas/ or /bayyas/, again ‮بيس‬‎ being the only acceptable spelling. But even if Diem is right to assume that *baʔīs is the origin of what ‮بيس‬‎ represents, it is quite probable that the outcome of *baʔīs after the loss of was not bayīs but rather bayyis aligning with the outcome of the CaCīC adjectival pattern of hollow roots such as mayyit ‘dead’ (cf. Blau 1967, § for which, once again, ‮بيس‬‎ would be the expected spelling. For these reasons this word is not a very good example of avoidance of two yāʔs in the sequence .8

The examples that are left, then all have in common that they either have they correspond to the Classical Arabic sequence iyyī/īʔī or āʔī. The examples are given below.

The reading traditions


an-nabiyyīna, an-nabīʔīna (Q2:61 & passim)9


‘the prophets’

al-ḥawāriyyīna (Q5:111; Q61:14)


‘the apostles’

al-ʔummiyyīna (Q3:20, 75; Q62:2)


‘the gentiles’

rabbāniyyīna (Q3:79)


‘worshippers of the lord’

ʔābāʔ-ī, ʔābāʔ-iya (Q12:38)10


‘my fathers’

warāʔ-ī, warāʔ-iya, warā-ya (Q19:5)11


‘behind me’

šurakāʔ-ī, šurakāʔ-iya (Q41:47)


‘my associates’

duʕāʔ-ī, duʕāʔ-iya (Q71:6)


‘my prayer’

ʔisrāāʔīl (Q2:40 & passim)



ǧibrīl, ǧabrīl, ǧabraʔil, ǧabraʔīl (Q2:97, 98)12



mīkāl, mīkāʔil, mīkāāʔil, mīkāāʔīl (Q2:98)13



As the apparent absence of double matres is phonetically conditioned, it seems like they should be considered the result of a genuine phonetic development, rather than an orthographic convention. In the case of the nouns that have a Classical Arabic sequence iyyīna or īʔīna it is likely that we are dealing with a contraction to /-īn/. Diem (1979, § 39) deems this unlikely, as he argues that an oblique plural ‮الامين‬‎ /al-ummīn/ should have had a nominative **‮الامون‬‎ /al-ummūn/. To my mind, it seems perfectly possible to have an asymmetrical paradigm nom. /al-ummiyyūn/ obl. /al-ummīn/ without necessarily undergoing analogical leveling in one direction or the other. This is, in fact, a possibility in the ʕarabiyyah, e.g. ‮أعجمين‬‎ ʔaʕjamīna ‘the non-Arabs’ (Fischer 2002, § 116, note 2).14

As for the nouns that in Classical Arabic end in āʔ followed by the 1sg. possessive marker, it seems likely that the sequence āʔ-ī or āʔ-iya simply collapsed to /ā-y/ after the loss of the hamzah and final short vowels. A trace of this development seems to have been retained in transmission of Ibn Kaṯīr’s reading as warā-ya.

This leaves us with ‮اسريل‬‎ ‘Israel’, ‮جبريل‬‎ ‘Gabriel’ and ‮ميكيل‬‎ ‘Michael’. At first sight one might want to read these as /ʔisrāyil/, /ǧibrāyil/ and /mīkāyil/. However, because ‮اسريل‬‎ stands in a /UR/ rhyme eight times (Q7:105, 134; Q26:17, 22, 59, 197; Q32:23; Q43:59), such a reading would break the rhyme. The reading that would be consistent with both the rhyme and the spelling is, in fact, /ʔisrīl/, paralleling the development that we see in the majority reading of ‮جبريل‬‎ as /ǧibrīl/. By extension it seems probable that ‮ميكيل‬‎ is to be understood as /mīkīl/.15

While double yāʔ avoidance when spelling does not appear to have been an orthographic principle, this seems to be different for double wāw avoidance when spelling (cf. Diem 1979, § 40). In post-consonantal position, the sequence /wū/ is indeed written with a single wāw. This is exemplified by forms of the verb lawā ‘to distort; to turn around’: ‮يلون‬‎ /yalwūn/ ‘they distort’ (Q3:78), ‮تلون‬‎ /talwūn/ ‘you will [not] turn around’ (Q3:153) and ‮تلوا‬‎ /talwū/ ‘you distort’ (Q4:135).16 We can likewise see this avoidance of two wāws in word-initial position we find ‮ورى‬‎ /wūrī/ ‘was concealed’ (Q7:20). It seems likely that we can also count ‮فاوا‬‎ /fāwū/ ‘so retreat!’ (Q18:16) and ‮الغاون‬‎ /al-ġāwūn/ ‘the deviators’ (Q26:224). The pronunciation of ‮داود, دواد‬‎ ‘David’ is difficult to determine, so it is not entirely certain whether that should be interpreted as an example of double wāw avoidance (see A.2.8).

Diem takes ancient sequence *aʔū(na) of III-ʔ stems in the plural as having developed to /awū(n)/. It seems likely however that III-ʔ and III-w/y stems have merged completely and these should rather be read as /aw(n)/. From spellings such as ‮يستهزون‬‎ /yastahzūn/ < *yastahziʔūna it is clear that at least the *iʔū sequence has merged completely with III-w/y stems. Indeed, in the reading traditions we see this with some of these verbs, with etymological -aʔūna forms, e.g. ʔabū Ǧaʕfar’s yaṭawna ‘they step’ < *yaṭaʔūna and murǧawna ‘postponed’ probably < *murǧaʔūna (see § 6.5.5).

Diem likewise analyses the adjectives ‮روف‬‎ ‘compassionate’ and ‮يوس‬‎ ‘despairing’ as evidence of aʔū > awū being represented by a single wāw. Once again one has to wonder whether this is a correct identification. For ‮روف‬‎, Diem implicitly assumes that the Ḥafṣ reading raʔūf is the origin of the word represented, and thus reconstruct /rawūf/, however, all other Kufan readers read raʔuf (Ibn al-Ǧazarī, § 2731), which would presumably yield /rawuf/ or perhaps /rawf/ after the loss of the hamzah, where spelling with a single wāw would be expected. ‮يوس‬‎ is universally recited as yaʔūs, but yaʔus is reported in Arabic lexicography (Lisān 4945c), thus likewise opening up the possibility of the reading /yawus/. If his assumption that these come from CaCūC adjectival patterns holds up, however, these may indeed be good examples of being spelled with a single wāw, assuming that *aʔū did not yield /awwu/ rather than /awū/ in this position.

To this he adds several other probable examples of this orthographic practice like al-mawʔūdah ‮الموده‬‎ /al-mawūdah/ ‘the buried alive girl’ (Q81:8) and ‮يوده‬‎ /yawūdu-h/ ‘it tires him’ (Q2:255). After the loss of hamzah ‮روس‬‎ ‘heads’ probably became /rūs/ (Q2:279) as the plural of /rās/ in analogy to /sāq/ pl. /sūq/ (incidentally also read as suʔūq, see § 6.4.10), but /ruwūs/ cannot be excluded.

Like double wāw, the sequence of double ʔalif is avoided. This is clearest in the case of nouns that end in ʔalif mamdūdah, followed by an indefinite accusative. Rhyme confirms that such sequences where indeed pronounced with two syllables, e.g. ‮انشا‬‎ /inšāʔā/ ‘a creation’ (Q56:35), yet they are spelled with only one ʔalif.

This same avoidance is found with the question particle ‮ا‬‎ /ʔa-/. When it combines with words that start with /ʔa-, ʔi-, ʔu-/, it is generally spelled with just a single ʔalif, e.g. ‮انتم‬‎ /ʔa-ʔantum/ (or /āntum/?) ‘are you?’ (Q2:140), ‮انك‬‎ /ʔa-(y)innaka/ ‘are you?’ (Q37:52), ‮انزل‬‎ /a-(w)unzila/ ‘has it been revealed?’ (Q38:8). Occasionally however, such sequences are spelled phonetically rather than morpho-phonemically, in which case a glide is written in the place of the word-initial vowel that followers the question particle, e.g. ‮اينكم‬‎ /a-yinna-kum/ ‘do you?’ (Q6:19), ‮اونبيكم‬‎ /ʔa-wunabbī-kum/ ‘shall I inform you?’ (Q3:15). Both spellings may even occur in a single verse, e.g. ‮ايذا متنا وكنا ترابا وعظما انا لمبعوثون‬‎ /a-yiḏā mutnā wa-kunnā turābā wa-ʕiẓāmā a-(y)innā (or innā)17 la-mabʕūṯūn/ ‘When we die and become dust and bones, will we be resurrected’ (Q56:47, cf. also the identical phrase in Q23:82 and Q37:16, where /a-(y)iḏā/ is spelled ‮اذا‬‎).

A.2.3 ʔalif al-Wiqāyah

A place where the orthography of the QCT diverges rather sharply from Classical Arabic orthography is in its use of the so-called ʔalif al-wiqāyah. In Classical Arabic, an ʔalif is written after word-final wāw only when this wāw denotes the verbal plural ending (Wright 1896, § 7a). In the QCT, its use is much more widespread, and regularly appears after any word-final /ū/ or /aw/, regardless of whether it is the plural verb or not (Nöldeke et al. 2013, 418 f.). This highly morphological spelling of Classical Arabic is thus an innovation. Examples of the broader use of the ʔalif al-wiqāyah are, e.g. ‮كفروا‬‎ /kafarū/ ‘they disbelieved’ (passim), ‮مشوا‬‎ /mašaw/ ‘they walked’ (Q2:20), ‮يدعوا‬‎ /yadʕū/ ‘he calls’ (Q2:221);18 ‮ملقوا ربهم‬‎ /mulāqū rabbi-hum/ ‘meeting of their lord’ (Q2:46); ‮ناكسوا روسهم‬‎ /nākisū rūsi-hum/ ‘the hanging of their heads’ (Q32:12). The relative pronoun /ḏū/ which in the CE follows the Classical Spelling ‮ذو‬‎, is consistently spelled ‮ذوا‬‎ in early Quranic manuscripts (Déroche 2009, 65).

There is only one case in the QCT where ʔalif al-wiqāyah is not used for word-final /-ū/, where we would expect it to be spelled, namely ‮يعفو‬‎ /yaʕfū/ ‘that he forgive’ (Q4:99) (see B.3). An exception to the general rule that whenever word-final /-aw/ occurs it should be written with ʔalif al-wiqāyah, are cases where a /w/ immediately precedes. Thus we find ‮اوو‬‎ /ʔāwaw/ ‘they gave shelter’ (Q8:72, 74) and ‮لوو‬‎ /lawwaw, lawaw/19 ‘they turn aside’ (Q63:5). This orthographic practice is lost in the CE, but is consistent in early Quranic manuscripts (see B.4). There are two other words that end in /-aw/ words which in the CE are written without ʔalif al-wiqāyah, one of them certainly had the ʔalif al-wiqāyah in the UT, namely, ‮سعوا‬‎ /saʕaw/ ‘they strove’ (Q34:5) and another whose data is a bit more ambiguous, as several very ancient manuscripts have the ʔalif al-wiqāyah while (mostly) later ones lack it, namely: ‮عتو(ا)‬‎ /ʕataw/ ‘and they became insolent’ (Q25:21), see B.4.

Nöldeke et al. (2013, 418 f.) object to the possibility that the ʔalif al-wiqāyah is intended to represent the phonetic value /ū/ and /aw/, and instead suggest that “every final ‮و‬‎ is followed by an ‮ا‬‎” and “exceptions to the rule can be easily explained”. However, one of the main exceptions is not addressed at all: All nouns that end in a consonantal /w/, either when preceded by a consonant, or when part of word-final /uww/ are consistently spelled without ʔalif al-wiqāyah. Examples of word-final -Cw are: ‮العفو‬‎ /al-ʕafw/ ‘the surplus; the forgiveness’ (Q2:219; Q7:199), ‮باللغو‬‎ /bi-l-laġw/ (Q2:225; Q5:89; Q25:72), ‮اللغو‬‎ /al-laġw/ (Q23:3; Q28:55), ‮لغو‬‎ /laġw/ (Q52:23) ‘idle talk’, ‮لهو‬‎ /lahw/ (Q6:32; Q29:64; Q31:6; Q47:36; Q57:20), ‮اللهو‬‎ /al-lahw/ (Q62:11) ‘amusement’, ‮البدو‬‎ /al-badw/ ‘the desert’ (12:100). Words that end in word-final -uww are: ‮عدو‬‎ /ʕaduww/ ‘enemy’ (Q2:36, and passim), ‮العدو‬‎ /al-ʕaduww/ ‘the enemy’ (Q63:4), ‮بالغدو‬‎ /bi-l-ġuduww/ ‘in the mornings’ (Q7:205; Q13:15; Q24:36), ‮لعفو‬‎ /la-ʕafuww/ ‘surely oft-pardoning’ (Q22:60; Q58:2), ‮عتو‬‎ /ʕutuww/ ‘arrogance’ (Q67:21). The rule as formulated by Nöldeke et al. does not account for this, whereas the phonetic definition (which they object to): wāw+ʔalif al-wiqāyah denotes /ū/ or /aw/, does.

Diem (1979, § 47) tries retain the orthographic rule formulated by Nöldeke et al. while taking these forms into account. The orthographic rule he formulates, however, is sufficiently complex that it would take a linguist to be able to spell correctly. He suggests that the ʔalif al-wiqāyah is only used of the ʔalif could not be mistaken for the indefinite accusative. This does a reasonable job at explaining laġwun ‮لغو‬‎ (Q52:23) versus laġwan ‮لغوا‬‎ (Q56:25), although even this requires a rather complex process of the scribe of needing to work through counterfactual readings, in order to ensure the ʔalif does not get written accidentally. But it becomes especially difficult to square with the fact that the definite form does not take the ʔalif al-wiqāyah either, e.g. al-laġwi ‮اللغو‬‎ (Q23:3), a context where writing the ʔalif al-wiqāyah could never lead to a confusion with the indefinite accusative.

Moreover, Diem’s rule is based on the mistaken assumption that luʔluʔ ‘pearl’ distinguishes the indefinite accusative luʔluʔan ‮لولوا‬‎ from the other cases ‮لولو‬‎ for luʔluʔun and luʔluʔin. This, however, is an idiosyncrasy of the CE. In the UT, this word always received the ʔalif al-wiqāyah also in the nominative and genitive form (see B.6).

Since indeed the use of ʔalif al-wiqāyah in these words is most readily explained phonetically, it being used whenever it is vocalic /ū/ or diphthongal /aw/, whereas when it is consonantal it is spelled without, it seems to me that contrary to the popular belief, the ʔalif al-wiqāyah does represent a phonetic value, rather than it being a purely orthographic practice (and certainly not a ‘word-divider’).

The reason why /ū/ and /aw/ are treated the same may be up for debate. First, it is of course possible that Quranic Arabic had lost final /aw/ of the verbs. In many modern dialects, e.g. Damascene Arabic, the final weak ending -aw has been lost completely and merged with , e.g. katabu ‘they wrote’ and banu ‘they built’ not **bano (Cowell 1964, 55, 61). It is possible that these merged in Quranic Arabic although a more conservative reconstruction seems prudent.

Another point of comparison here is the treatment of diphthongs in the Old Arabic as reflected in the Safaitic inscriptional corpus. Safaitic orthography never writes vowels with matres lectionis. Thus, /ū/ is never expressed with ⟨w⟩. Perhaps surprisingly, the diphthong /aw/ is treated the same, and is likewise never expressed in writing whereas consonantal /w/ is expressed with ⟨w⟩. Thus, to the speakers of the Safaitic Old Arabic dialect, the diphthong /aw/ was treated as a true diphthong, that is more similar to a long vowel than a vowel+consonant sequence (Al-Jallad 2015, 37 f.).

The treatment of /aw/ and /ay/ as being distinct from other consonantal uses, and more akin to the long vowels, is also something we see in their treatment in the Arabic grammatical tradition. Thus, the ḥurūf al-līn are the use of ʔalif, yāʔ and wāw when a vowel precedes, in words like: nār ‘fire’—envisioned as /naAr/, dār /daAr/ ‘house’, fīl /fiyl/ ‘elephant’, qīla /qiyla/ ‘it is said’, ḥūla /ḥuwla/ ‘it was changed’ ġūl /ġuwl/ ‘ogre’, bayt ‘house’ and ṯawb ‘garment’ (Lisān, 4117c).20

In light of this it seems quite likely, and phonologically plausible that the ʔalif al-wiqāyah was used as a tool to write word-final ‘vocalic’ uses of wāw, i.e. /ū/ and /aw/ as opposed to consonantal uses of wāw.

Another argument that Nöldeke et al. bring up to not take this as a phonological spelling, but rather a ‘place wāw after every wāw’ rule is that it is placed after verbs in the subjunctive, such as ‮يعفوا‬‎, ‮تعفوا‬‎ (Q2:237), ‮لتتلوا‬‎ (Q13:30), ‮لن ندعوا‬‎ (Q18:14), ‮ان اتلوا‬‎ (Q27:92) ‮ليربوا‬‎ (Q30:39), ‮ليبلوا‬‎ (Q47:4) ‮نبلوا‬‎ (Q47:31), which according to them must be verbs ending in -uwa not . This presupposes that the Quranic reading traditions are an accurate representation of the language of the QCT, and final short vowels were not lost in such verbs. Neither of these assumptions are justified. The fact that these verbs are treated exactly the same as verbs that end in in Classical Arabic rather speaks in favour of the loss of the final short vowels, something that I have also argued on different grounds in Chapter 7 and Van Putten & Stokes (2018).

An exceptionally difficult issue is the treatment of the ʔalif al-wiqāyah in roots that originally contained hamzah. While some of these behave exactly as expected, it is especially the historical sequences *-aʔu and *āʔu that paint a rather complex picture. Nöldeke et al. (2013, 419) object to seeing the ʔalif al-wiqāyah as a phonetic marking for /ū/ and /aw/ as against consonantal /w/, because many words of the type have final hamzah. This, again, presupposes that the Quranic reading traditions are an accurate reflection of the language of the QCT, which certainly in the case of the hamzah cannot be accepted. It is quite clear that Quranic Arabic had lost hamzah completely (see § 5.2) which has given rise to many forms of artifical and pseudocorrect hamzah use all throughout the reading traditions (see Chapter 6 and § 3.6.1). In the following sections we will discuss the different contexts where ʔalif al-wiqāyah appears where the words etymologically contained a hamzah.

A.2.3.1 ʔalif al-wiqāyah for Stem Final *uʔ

When it comes to stem final *uʔ, regardless of what vowel would historically follow, the word is always spelled with ʔalif al-wiqāyah. Thus ‮امروا‬‎ /imrū/ ‘man’ (Q4:176), ‮لولوا‬‎ /lūlū/ ‘pearl’ (Q52:24), ‮اللولوا‬‎ ‘the pearl’ (Q55:22; Q56:23). In the Cairo Edition some these forms of ‘pearl’ are spelled without ʔalif al-wiqāyah, but this not original to the UT, see B.6.

In the case of the indefinite ‮لولوا‬‎ ‘pearl’, the spelling is thus ambiguous whether it represents nominative/genitive /lūlū/ or accusative /lūluwā/. This ambiguity has indeed lead to disagreement in the Quranic reading traditions where the word may be read both as a genitive luʔluʔin, lūluʔin and as an accusative luʔluʔan, lūluʔan (Q22:23; Q35:33, see Ibn al-Ǧazarī, § 3652).

A.2.3.2 Treatment of Stem-Final *ūʔ

In nouns, etymological sequences of stem-final *ūʔ behave exactly the same as stem-final /uww/, and thus are spelled without ʔalif al-wiqāyah: ‮قرو‬‎ /quruww/ (< qurūʔ-) ‘menstruations’ (Q2:228) ‮سو‬‎ /suww/ (< *sūʔ-) ‘the wickedness of …’ (Q2:49, and passim), ‮بالسو‬‎ ‘wickedness’ (Q2:169, and passim), ‮السو‬‎ ‘wickedness’ (Q4:17, and passim). Of course, in the indefinite accusatives, these receive a final ʔalif as the mark of the indefinite accusative, e.g. ‮سوا‬‎ /suwwā/ ‘wickedness’ (Q4:110)

In the verbal system, however, we find these spelled with ʔalif al-wiqāyah in the two instances that it occurs. What is recited in the reading tradition as la-tanūʔu is spelled ‮لتنوا‬‎ ‘would be a burden’ (Q28:76) and what is recited as tabūʔa is spelled ‮تبوا‬‎ ‘that you bear’ (Q5:29). This is likely the result of analogical leveling due to a partial paradigmatic levelling of the II-w, III-ʔ imperfect paradigm with the III-w paradigm:


Hamzaless Arabic









*yabuww >> yabū











The merger of II-w, III-ʔ verbs with III-w verbs towards ending in /-ū(n)/ may perhaps be visible in ‮ليسوا‬‎ /li-yasū/ (Q17:7) which is variously read as li-yasūʔū ‘so that they will sadden’, li-yasūʔa ‘so that he will sadden’ and li-nasūʔa ‘so that we will sadden’ (Ibn al-Ǧazarī, § 3447). If the majority reading yasūʔū was indeed the grammatical form intended, then it seems that the Quranic Arabic pronunciation of this was /li-yasū/.

A.2.3.3 Treatment of Word-Final *āʔū

Unlike Classical Arabic spelling, II-w/y, III-ʔ verbs in the in the perfect 3pl.m. form are spelled without an ʔalif al-wiqāyah, e.g. ‮وباو‬‎ /bāw/ ‘they returned’ (Q2:61); ‮فاو‬‎ /fāw/ ‘they returned’ (Q2:226),21 ‮جاو‬‎ /ǧāw/ ‘they came’ (Q3:184) and also ‮راو‬‎ /rāw/ ‘they saw’22 (Q2:166; Q7:149; Q10:54; Q12:35; Q19:75; Q28:64; Q34:33; Q37:14; Q40:84, 85; Q42:44; Q62:11; Q72:24). The last of these is spelled ‮راوا‬‎ in the Cairo Edition, but this is not original to the UT, see B.7. As word-final /ū/ is otherwise always spelled with ʔalif al-wiqāyah, this suggests that word-final *āʔū shifted to /āw/, rather than **āwū as Diem (1979, § 65) suggests.

Of exceptional status is ‮اسوا‬‎ ‘they did evil’ (Q30:10; Q53:31) which is universally recited as ʔasāʔū and thus we would rather expect the spelling **‮اساو‬‎. But indeed, in early Quranic manuscripts, the spelling is as it is found in the Cairo edition (see B.16). This spelling thus seems to suggest a pronunciation /ʔasaw/ rather than /ʔasāw/. As this is the only C-stem perfect in the 3pl.m. of stems of this type, it is difficult to be sure about this analysis.

A.2.3.4 Word-Final *aʔū

Plural hamzated verbs that historically end in *aʔ-ū are likewise spelled with the ʔalif al-wiqāyah and are presumably pronounced /-aw/: ‮تبروا‬‎ /tabarraw/ ‘they disown’ (2:167), ‮فادروا‬‎ /fa-draw/ ‘so avert!’ (3:168), ‮اقروا‬‎ /iqraw/ ‘recite!’ (Q69:19), ‮فاقروا‬‎ /fa-qraw/ ‘so recite!’ (Q73:20).

One verb lacks the final ʔalif al-wiqāyah: ‮تبوو‬‎ /tabawwaw/ ‘they settled’ (Q59:9), thus showing similar behaviour as the verbs without an original that have /w/ before a final /-aw/, like ‮اوو‬‎ /ʔāwaw/ ‘they gave shelter’ (Q8:72, 74) and ‮لوو‬‎ /lawwaw, lawaw/ ‘they turn aside’ (Q63:5) discussed above.

A.2.3.5 Word-Final *aʔu(n)

As for *aʔu sequences, verbs are overwhelmingly spelled with wāw and ʔalif al-wiqāyah, with a couple of exceptions where it is simply spelled with ʔalif, e.g. ‮يبدوا‬‎ /yabdaw/ ‘he begins’ (Q10:4, 34 (2×); Q27:64; Q30:11, 27), ‮تفتوا‬‎ /taftaw/ ‘you will not cease’ (Q12:85), ‮يتفيوا‬‎ /yatafayyaw/ ‘it inclines’ (Q16:48), ‮اتوكوا‬‎ ‘I lean’ /atawakkaw/ (Q20:18), ‮لا تظموا‬‎ /lā taẓmaw/ ‘you will not be thirsty’ (Q20:119) ‮يدروا‬‎ /yadraw/ ‘he knows’ (Q24:8), ‮ما يعبوا‬‎ /mā yaʕbaw/ ‘will not concern himself’ (Q25:77), ‮ينشوا‬‎ /yunaššaw/ ‘is brought up’ (Q43:18), ‮ينبوا‬‎ /yunabbaw/ ‘will be informed’ (Q75:13). There are three exceptions to this general rule, namely ‮يستهحزا‬‎ /yustahzā/ ‘it is being ridiculed’ (Q4:140), ‮يتبوا‬‎ /yatabawwā/ ‘he settles’ (Q12:56), ‮نتبوا‬‎ /natabawwā/ ‘we settle’ (Q39:74).

For nouns, the ʔalif spelling is more common, although the spelling with ʔalif al-wiqāyah occurs as well. Thus for *al-malaʔu ‘the chieftains’ we see: ‮الملا‬‎ /al-malā/ (Q7:60 66, 75, 88, 90, 109, 127; Q11:27; Q12:43; Q23:33; Q28:38; Q38:6) and ‮الملوا‬‎ /al-malaw/ (Q23:24; Q27:29, 32, 38). The other noun, from *nabaʔu ‘the news of’, on the other hand, occurs more often in the wāw + ʔalif al-wiqāyah spelling: ‮نبوا‬‎ /nabaw/ (Q14:9; Q38:21; Q64:5) but ‮نبا‬‎ /nabā/ (Q9:70). The indefinite form *nabaʔun ‘news’ is likewise spelled ‮نبوا‬‎ (Q38:67) (see B.8, B.9).

The presence of these spelling with final wāw and ʔalif al-wiqāyah seems to have an important implication for the relative chronology of final short vowels and the hamzah, as it requires that hamzah was lost before the final short vowels were lost. The forms that are simply spelled with ʔalif are perhaps analogical levelling of the default form, as verbs that end in do not usually show a distinction between the imperfective and aorist/apocopate, and likewise nouns that end in do not usually show a distinction between the nominative versus the accusative after the loss of final short vowels.23

A.2.3.6 Word-Final *āʔu

An especially vexing case of the issue of the ʔalif al-wiqāyah in words that etymologically end in ʔalif mamdūdah followed by the nominative or imperfect *-u. First of all, it should be said that unlike the reflexes of *aʔu—where the distribution is almost 50/50—the vast majority of the words in this group are simply spelled with the final ʔalif. However, there are 18 cases in the CE where a spelling with wāw + ʔalif al-wiqāyah shows up. However, a closer look at the data in early Quranic manuscripts shows that not all of these can be successfully reconstructed with that spelling in the UT. B.10, B.11, B.12, B.13, and B.14 tabulate the attestations of the relevant words as they appear in early manuscripts. Here I will give a summary of the conclusions we can draw from this examination. Below, I have also included a few cases where an unusual spelling occurs where the CE has ʔalif.






‮ابنوا الله‬‎

‮ابنا الله‬‎



‮انبوا ما‬‎

‮انبوا ما‬‎



‮انبوا ما‬‎

‮انبا ما‬‎ (probably)



























‮علموا بنى اسريل‬‎

‮علما بنى اسريل‬‎








‮ٮشاو‬‎ or ‮ٮشوا‬‎











‮دعوا الكفرين‬‎

‮دعا الكفرين‬‎



‮جزوا الظلمين‬‎

‮جزاو الظلمين‬‎



‮جزوا الذين‬‎

‮جزاو الذين‬‎



‮جزا من‬‎

‮جزاو من‬‎



‮جزا المحسنين‬‎

‮جزاو المحسنين‬‎



‮جزوا سييه‬‎

‮جزاو سييه‬‎



‮جزوا الظلمين‬‎

‮جزا الظلمين‬‎

Of the 19 words spelled with the ʔalif al-wiqāyah, only six appear to have been spelled as such in the UT, with one (‮نشوا‬‎ Q11:87) being somewhat unclear. In five cases ǧazāʔu is not spelled as ‮جزوا‬‎ or ‮جزا‬‎ but as ‮جزاو‬‎ instead, whereas the spelling ‮جزوا‬‎ is entirely absent. The normal spelling of ʔalif mamdūdah nouns with simple ʔalif remains the majority spelling however (10 cases).

All cases of the ‮او‬‎ spelling are nouns in construct (a place where final short vowels appear to have been retained), and thus ‮جزاو‬‎ may very well represent /ǧazāwu/, with optional weakening of stem final hamzah, whereas ‮جزا‬‎ in identical context would represent /ǧazāʔu/. With a clitic following, this noun is variously spelled with and without the final glide in the UT (A.4.11). This spelling would then align with the proposed theory here that wāw not followed by ʔalif al-wiqāyah represent consonantal /w/. The only fly in the ointment is ‮انبوا‬‎ (Q6:5) which is likewise stands in construct but has the ʔalif al-wiqāyah. A possible solution is to not read this as a ʔaCCāC plural, but rather as a ʔaCCuC plural, i.e. *ʔanbuʔu > ʔanbū, which would explain this spelling. Admittedly, however, this solution is rather ad hoc.

The remaining words with the ʔalif al-wiqāyah spelling are all diptotic CuCaCāʔu plurals that do not stand in construct. Among these nouns, spellings of this type are fairly common with, five times appearing with the wāw+ʔalif al-wiqāyah in the nominative, and 19 cases where it is spelled with ʔalif. Rabin (1951, 110, §w) speculates (following Vollers) that these forms represent /aḍ-ḍuʕafō/ with a final vowel /ō/, and he seems to think that there is no special relationship between this spelling and the nominative. The fact that we never see such spellings in non-nominative contexts (which are by no means uncommon) however make this rather unattractive to assume that the original case vowel plays no role here.

However, it is similarly unlikely to take these spelling as representing /āwu/, or even /āʔu/ (as suggested by Diem 1981, § 184a; and Nöldeke et al. 2013, 422). In contexts much less ambiguous than the very specific context of CuCaCāʔu-plurals, it seems to be clear that with such a sequence the spelling ‮او‬‎ would be expected, at least usually (see the reflexes of *āʔū, and *ǧazāʔu above). I would tentatively suggest that for reasons currently not entirely clear, the outcome of diptotic *CuCaCāʔu indeed is /CuCaCō/, creating a diptotic case distinction not dissimilar to the sound masculine plural with /CuCaCō/ in the nominative and /CuCaCāʔ/ (or /CuCaCā/?) in the oblique. Quranic Arabic then represents a stage where such nominatives have mostly, but not entirely, been analogically levelled.

The difference in behavior of the diptotic plurals may very well be because of their lack of nunation. Thus *-āʔu, *āʔa became /-ō, -ā/, because there was no nunation to guard this contraction, whereas *-āʔun, *āʔin, *āʔan were exempt from this contraction and became /-āʔ, -āʔ, -āʔā/. This may also explain why *ʔawliyāʔu- when followed by pronominal clitics appears to behave as ending in ʔalif maqṣūrah /ʔawliyā-hum/ rather than ʔalif mamdūdah **/ʔawliyāwu-hum/ (see A.4.11). Without further data this hypothesis will have to remain speculative.

A.2.3.7 ‮الربوا‬‎, ‮ربا‬‎

A final word whose spelling appears to contain an ʔalif al-wiqāyah is ‮الربوا‬‎ ‘usury’ (Q2:275 (3×), 276, 278; Q3:130; Q4:161), which in the indefinite appears spelled as ‮ربا‬‎ (Q30:39) (see B.15). This alternation between ‮وا‬‎ and ‮ا‬‎ spelling may at first glance seem similar the treatment of the *CuCaCāʔu plurals discussed above. However, unlike the nouns above, this spelling is not unique to the nominative, but is found in all cases but the nominative, e.g. ‮الذين ياكلون الربوا‬‎ “those who devour usury (acc.)” (Q2:275), ‮مثل الربوا‬‎ “like usury (gen.)” (Q2:275).

It is quite unclear what the etymological background of this word is and how to interpret it. In the Quranic reading traditions it is either read as ar-ribā or ar-ribē (Ibn al-Ǧazarī, § 1974), but on the basis of the spelling with wāw, it has been argued that it should be /ar-ribō/ (Rabin 1951, 105; Nöldeke et al. 2013, 418). As Rabin points out, this is an opinion, already endorsed by al-Zamaxšarī (Kaššāf, I, 319). However, there are no other stems with /ō/ as a word-final vowel, regardless of case—and there is no obvious etymological origin for this vowel to appear in this context. As shown by Van Putten (2017a), *awV- yields /ā/ in Quranic Arabic, not /ō/, and thus we would expect *ar-ribawa to have yielded **‮الربا‬‎ /ar-ribā/, or perhaps even ‮الربى‬‎ /ar-ribē/.24 Due to the unusual position of this word, Rabin (1951, 109, §u) seems justified in assuming that the word is likely a borrowing of some kind, but the exact linguistic origin remains unclear.

A.2.3.8 Summary

The table below summarizes the distribution of the different spelling of ʔalif al-wiqāyah and wāw (as well as the spellings ‮او‬‎ and simply ‮ا‬‎). Excluded from this table are several highly frequent particles which are never spelled with ʔalif al-wiqāyah. These are ‮هو‬‎, ‮او‬‎ and ‮لو‬‎. In the cases where the distribution is not absolute, I have shaded the cell with the dominant spelling.

*ū, *uwa
























3 (*-waw)(+1?)

1 (*-waʔū)







It should be clear that the two spellings ‮وا‬‎ and ‮(ا)و‬‎ are in quite strict complementary distribution, clearly suggesting a phonetic origin for these spellings. The only environment where such spellings appear to be in competition is in the *-āʔu(n) sequence. However, as we saw above even here the two spellings appear to be mostly in complementary distribution, where ‮او‬‎ is reserved for triptotic nominative nouns in construct, and ‮وا‬‎ is reserved for diptotic nouns in the nominative definite and indefinite form. It therefore seems quite reasonable to suggest that indeed ‮وا‬‎ is used to write /-ū/ and /-aw/, whereas ‮و‬‎ marks word-final consonantal /w/.

A.2.4 Spelling of la- ‘Indeed’ as ‮لا‬‎

In the Quran the asseverative la- is frequently spelled ‮لا‬‎ before 1sg. form of the verb.25 It is attested once in the CE in ‮لااذبحنه‬‎ /la-ʔaḏbaḥanna-h/ ‘I will surely slaughter him’ (Q27:21), but attested in quite a few more places in Early Quranic Manuscripts, for example ‮فلااقطعن‬‎ /fa-la-ʔuqaṭṭiʕann/ ‘So surely I will cut off’ (Q20:71 in SM1a); ‮لااكيدن‬‎ /la-ʔakīdann/ ‘surely I will plan’ (Q21:57 in W, T26); ‮لااملن‬‎ /la-ʔamlānna/ ‘I will surely fill’ (Q32:13 in W,27 T;28 Q38:85 in BL); ‮ولااغوينهم‬‎ /wa-la-ʔuġwiyanna-hum/ ‘and surely I will mislead them’ (Q15:39 in Arabe 334c); ‮ولاامرنهم‬‎ /wa-la-ʔāmuranna-hum/ ‘I will surely command them’ (Q4:119 in W, Arabe 330b); ‮ولاادخلنكم‬‎ ‘I will certainly admit you’ (Q5:12 in Arabe 324c);29 ‮لااقتلنك‬‎ /la-ʔaqtulanna-k/ ‘I will surely kill you’ (Q5:27 in W,30 CPP, BL31); ‮لااتينهم‬‎ /la-ʔatiyanna-hum/ ‘I will surely come to them’ (Q7:17 in S, SM1a, K); ‮لااصلبنكم‬‎ /la-ʔuṣallibanna-kum/ ‘I will surely crucify you’ (Q7:124 in CPP,32); ‮لاازيدكنكم‬‎ /la-ʔazīdanna-kum/ ‘I will surely increase you’ (Q14:7 K33); ‮لااعذبنه‬‎ /la-ʔuʕaḏḏibanna-hū/ ‘I will surely punish him’ (Q27:21 in W, T).

Besides these extra places in the Quran where we attest such spellings, there are also some disagreements among the reading traditions about whether certain phrases should be read with or asseverative la- that seem to stem from this spelling practice. For example, Qunbul ʕan Ibn Kaṯīr reads Q10:16 ‮لو شا الله ما تلوته ولاادريكم به‬‎ law šāʔa ḷḷāhu mā talawtu-hū ʕalaykumū wa-la-ʔadrā-kumū bi-hī “and if Allah had willed it he would have not have recited it to you, and he would have made it known to you” rather than reading ‮ولاادريكم به‬‎ as wa-lā ʔadrā-kum bi-hī “nor would he have made it known to you” (Ibn al-Ǧazarī, § 3247).34 Qunbul ʕan Ibn Kaṯīr also reads ‮لااقسم بيوم القيمه‬‎ (Q75:1) as la-ʔuqsimu bi-yawmi l-qiyāmah ‘I definitely swear by the day of resurrection’, while the rest reads lā ʔuqsimu bi-yawmi l-qiyāmah ‘No! I swear by the day of resurrection’ (Ibn al-Ǧazarī, § 3247).35

Sidky (2021, 181) points out that it was already noticed early on that this surprising early orthography could yield significant ambiguities as both the asseverative and negative indicative would end up being spelled exactly the same, which al-Farrāʔ criticitices as being “of the terrible spelling practices of those of old” (wa-huwa min sūʔi hiǧāʔi l-ʔawwalīna) (Al-Farrāʔ Maʕānī, I, 295 f.).

The spelling as ‮لا‬‎ is not just restricted to cases of the asseverative particle before a 1sg. verb, but can occur before any word that starts with a hamzah; Al-Dānī (Muqniʕ, 36) reports the spelling ‮لااوضعوا‬‎ ‘they were active’ (Q9:47), which is indeed attested in early manuscripts (GK; BL; Rampur Raza). And likewise, for the asseverative particle combined with the preposition ‮الى‬‎ we find the spelling ‮لاالى‬‎ (Q3:158: S, W,36 Q47, GK,37 CPP; Q37:68: W, Arabe 333d), as pointed out by Diem (1979, § 26). A close examination of early manuscripts will likely uncover even more cases.

A.2.5 The Prepositions ʕalā, ḥattā and ladā Are Often Spelled ‮حتا، علا، لدا‬‎

It is common in early copies of the Kufic C style to write the prepositions ʕalā and ḥattā as ‮علا‬‎ and ‮حتا‬‎ rather than the now standard ‮على‬‎ and ‮حتى‬‎ (Cellard 2015, 208–213), manuscripts of this type appear to always be of Medinan regionality (Cellard 2015, 168–186; see also van Putten 2019a, see especially 356, note 122). This alternate spelling is also found once in the CE for ladā: ‮لدى‬‎ (Q40:18) and ‮لدا‬‎ (Q12:25). These three words are exactly the words with ʔalif maqṣūrah that reading traditions that have III-w ʔimālah (see § normally read as /ā/, despite their spelling (Ibn al-Ǧazarī, § 1973),38 and ḥattā is explicitly mentioned as being an exception by Sībawayh (IV, 135). It seems to be the case that these words in Quranic Arabic were probably pronounced /ḥattā/, /ʕalā/ and /ladā/, despite their spelling. The spelling with ‮ى‬‎ for these words should probably be considered historical spellings, rather than reflecting the pronunciation of Quranic Arabic (van Putten 2017a, 62).39

The most likely explanation for this exceptional spelling of /ā/ with ‮ى‬‎ can probably be explained through their respective etymologies. It seems likely that in the history of Quranic Arabic these prepositions were *ʔilay, *ʕalay, *laday, and *ḥattay.40 When these would be combined with a noun starting with the definite article, it would create a *aya triphthong which would then contract to and get subsequently shortened to a in a closed syllable, e.g. *ʕalay al-raǧuli > *ʕalē l-raǧuli > ʕala l-raǧul. From this realization ʕala before definite articles, one could easily get a preposition ʕalā through backformation.

A.2.6 Words Starting with /l/ Preceded by the Definite Article.

The definite article when it precedes a word that starts with ‮ل‬‎, is sometimes written with only a single lām. This is regular for ‮اليل‬‎ ‘night’, and was probably original for ‮الولوا‬‎ ‘the pearls’, which is written with only a single lām in two rather early manuscripts (see B.6). All forms of the relative pronoun in the Quran are spelled with a single lām, rather than the Classical Arabic practice which only maintains this spelling for the singular and masculine plural forms, whereas all other forms write it with two lāms (see A.4.5).

The vast majority of the words whose stem starts with lām however, are written with two lāms, most notably, of course, ‮الله‬‎ /aḷḷāh/. The fact that even before the lām the definite article is usually spelled morphologically rather than phonetically (unlike Nabataean Arabic) was one of the reasons for Van Putten (2019b, 15) to suggest that the definite article was probably assimilated in Quranic Arabic, as it is in Classical Arabic, and that at the very least it cannot tell us that it was unassimilated as it is in the Damascus Psalm fragment.

A.2.7 Historical Hamzah Spelling with ‮اى‬‎

Šayʔ ‘thing’, in early Quranic manuscripts, is written both ‮شى‬‎ and ‮شاى‬‎, apparently haphazardly but with a clear preference to spelling it with ʔalif. In the Cairo edition ‮لشاى‬‎ is attested in Q18:23. There is no special significance to this position in early manuscripts, where the spelling may occur elsewhere, and some manuscripts spell it ‮لشى‬‎ in Q18:23 as well (e.g. SM1a).41 I side with Diem (1980, § 127–128) that this is likely a historical spelling. There are many cases where an original next to a *y or in a position where it would become a y is spelled with the orthographic ‮اى‬‎. Other cases of this found in the Cairo edition are:

‮الملايه/مالايهم‬‎ /al-malayi-h(um)/ ‘his/their chiefs’ (Q7:103; Q10:75, 83; Q11:97; Q23:46; Q28:32; Q43:46)
‮جاى‬‎ /ǧīy/ ‘it was brought’ (Q39:69; Q89:23)
‮من نباى المرسلين‬‎ /min nabay(i) al-mursalīn/ ‘of the tidings of the messengers’ (Q6:34)
‮افاين‬‎ /a-fa-(y)in/ ‘but if not …’ (Q3:144; Q21:34)
‮مايه، ماتين‬‎ /miyah, miyatayn/ ‘one/two hundred’ (Q2:259 (2×), 261; Q8:65 (2×), 66 (2×); Q18:25; Q24:2; Q37:147).
‮السواى‬‎ /as-sūwē/ ‘the evil’ (Q30:10)
‮يايس، تايسوا‬‎ /yayas, tayasū/ ‘he despairs/(do not) despair’ (Q12:87 (2×); Q13:31).42

In early Quranic manuscripts, the verb ‮شا‬‎ ‘to want’ and ‮جا‬‎ ‘to come’ in the suffix conjugation also occasionally employs this spelling:

‮شايت‬‎ ‘you want’ (SM1a, Q18:77; T, Q24:62) ‮شاىتم‬‎ ‘you (pl.) want’ (Arabe 331, Q2:223; DAM 01–21.3, Q7:161). ‮جايت‬‎ ‘you came’ (T, Q19:27)

Several other examples have been identified by Puin (2011, 164).

‮سايل‬‎ /sīl/ ‘it was asked’ (Q2:108, in S, DAM 01–28.1)
‮ساى‬‎ /sīy/ ‘he was distressed’ (Q11:77, in S)
‮رايي‬‎ /rūyā-y/ ‘my vision’ (Q12:100 in CPP; W; SM1a; Q12:43 in W, SM1a, GK)
‮بايس‬‎ /bayyis/ ‘wretched’ (Q7:165 in A6140a43)

An apparent application of this same spelling practice is found in the spelling of classical as-sayyiʔah. While this is spelled ‮السييه‬‎ in the CE, occasionally in early manuscripts we find ‮الساييه‬‎, with the etymological ʔalif seemingly before the wrong consonant in terms of the seat of the hamzah.

Q7:95 ‮الساييه‬‎ /as-sayyiyah/ (CPP; BL)

A similar process is found on word-boundaries. Whenever a word is preceded by bi- or li- and the consonant after the ʔalif is a yāʔ, a second yāʔ is written. In the CE this only occurs three times ‮باييم‬‎ /bi-(y)āyyām/ ‘in the days of …’ (Q14:5), ‮باييد‬‎ /bi-(y)ayd/ ‘with strength’ (Q51:47), ‮باييكم‬‎ /bi-(y)ayyi-kum/ ‘which of you’ (Q68:6). But this practice is much more widespread in early Quranic manuscripts than it is in the CE. The spelling is especially common in the phrase ‮باييت‬‎ ‘with the signs/verses of …’ which is subjected to a rigorous study by Déroche (2014, 47). Also, the singular ‮باييه‬‎ ‘with the sign/verse of …’ is usually spelled in this manner. There are, at least, occasionally cases where the same spelling is employed after the prefix li- (van Putten 2018, 111).

bi-ʔayyi is invariably spelled as ‮بايى‬‎ in early manuscripts, as can be easily seen in the oft-repeated ‮فبايى‬‎ in Q55 (see Arabe 331, W, SM1a, Top etc.). In other places in the Quran, we likewise find the same spelling regularly: ‮فبايى‬‎ (Q7:185; Q45:6; Q53:55; Q77:50), ‮بايى‬‎ (Q31:34; Q81:9) and ‮لايى‬‎ (Q77:12).

Van Putten (2018, 109 f.) suggested that these spellings are hybrid spellings that represent both the original etymological ʔalif, but use the yāʔ to point out that these forms were now pronounced as /bi-yāyāt/, /bi-yayyi/ etc. Some evidence for this reading is furthermore found in the Quranic reading traditions. Al-ʔaṣbahānī ʕan Warš ʕan Nāfiʕ is said to have pronounced every instance of fa-bi-ʔayyi as [fabiyayyi], and there is some disagreement within his transmission whether bi-ʔayyi-kum (Q68:6) and bi-ʔayyi (Q31:34) are read this way (Ibn al-Ǧazarī, § 1492–1493). There is however no mention of Q81:9, nor li-ʔayyi Q77:12, and nothing on bi-ʔāyāt, bi-ʔaydin or bi-ʔayyām.

While Van Putten (2018) does not comment on this, it is very striking that whenever such a spelling takes place across word boundaries, it invariably involves a word that starts with ‮ا‬‎ and has ‮ى‬‎ later in the word. This is unlikely to be a coincidence. Perhaps the loss of word-initial did not undergo the same developments as word-internal *ʔ (something suggested by other parts of its orthographic behaviour as well) and a secondary sound-law takes place where > y /i_ā̆y. The rather specific context in which the sound law takes place, however, is ad hoc and another solution may be thought of, but it seems that any explanation must account for the fact that this spelling practice seems to be conditioned by a y later in the word.44

A.2.8 The Spelling of dāwūd as ‮دواد‬‎ and ruʔūs as ‮رواس‬‎

A rather puzzling spelling practice which may be related to the marking of historical hamzah when it stands next to ū can be found in the name dāwūd. In the CE this name is spelled ‮داود‬‎, but in early manuscripts it is frequently spelled ‮دواد‬‎ (Marx and Jocham 2015, 29 ff.). It seems possible that this reversed order of the ʔalif and wāw represents an alternative pronunciation of the name: /duwād/ or /duʔād/,45 similar to the appearance of the spelling ‮ابرهم‬‎ and ‮ابرهيم‬‎ for ʔibrāhām and ʔibrāhīm (van Putten 2020b). Unlike the spellings of ‮ابرهم, ابرهيم‬‎ no clear pattern arises of the two spellings being used in fixed places across manuscripts. In general, manuscripts either have the ‮دواد‬‎ or ‮داود‬‎ spelling (see B.17).

Such an explanation is less obvious for the representation of ruʔūs however. It likewise occurs spelled as ‮رواسهم‬‎ in two manuscripts (Q14:43, Is. 1615 I, CA1). Here we can hardly argue that the pronunciation was ruwās or ruʔās. It seems then that, for whatever reason the ‮وا‬‎ sequence may reflect a historical sequence *ʔū, with the etymological position of the transposed, perhaps because it was no longer pronounced, much like what we saw with the historical spelling ‮اى‬‎ for words that involved yāʔ, where the ʔalif also often does not stand in the right place, e.g. ‮جاى‬‎ < *ǧīʔa. If this is also the case for ‮دواد‬‎, the spelling may still represent /dāwūd/, which would then have come from an earlier *dāʔūd.46 Note however, other manuscripts occasionally attest an etymological hamzah spelling for words with similar syllable structure, where the ʔalif does stand in the etymologically correct position, e.g. ‮براوسكم‬‎ (Q5:6, ms.or.fol. 4313), likewise yaʔūsan is spelled with a historical spelling ‮ياوسا‬‎ (Q17:83, Or. 2165) (see B.18).

A.2.9 Plene Spelling of Short u

Occasionally, the Quran attests examples where what was likely short u is spelled with a ‮و‬‎. This is well attested in forms of the plural demonstrative element *ʔul- such as ‮اوليك‬‎ /ulāyik/ ‘those’, ‮اولا‬‎ /ulāʔ/ ‘those’; ‮اولوا‬‎ /ulū/ ‘those of (masculine, nominative)’; ‮اولى‬‎ /ulī/ ‘those of (masculine, genitive/accusative)’; ‮اولات‬‎ /ulāt/ ‘those of (feminine)’. For words of this type, this spelling practice continues to be the regular spelling all throughout the written history of Arabic.47 As it is unusual to find short vowels being spelled with a long vowel sign in the QCT, as this is not at all the norm, one might consider the possibility that these forms in Quranic Arabic originally had long vowels, as per their spelling. This option appears to be supported by Rabin (1951, 153), who however does not comment on it explicitly and also says that Classical Arabic has ʔūlāʔi, which to my knowledge only occurs with a short vowel (Fischer 2002, § 7, n. 7). However, the short vowel is in better agreement with the cognates of this plural morpheme in other Semitic languages, which all universally point to a short vowel. The spelling of short u with ‮و‬‎ is attested once in a context other than the plural pronoun base ʔul-, namely, ‮ساوريكم‬‎ /sa-urī-kum/ ‘I will show you’ (Q7:145; Q21:37).

A.2.10 Defective Spelling of Word-Final Long Vowels before ʔalif al-waṣl

While long vowels before ʔalif al-waṣl are generally spelled plene, it appears that these long vowels were shortened, at least, before the ʔalif al-waṣl of the definite article, and this shortening is on occasion expressed in the orthography, as pointed out by Nöldeke et al. (2013, 409).

The vocative ʔayyuhā is usually spelled ‮ايها‬‎ in the QCT, but a spelling ‮ايه‬‎ occasionally occurs. In all cases this happens in front of a noun with the definite article, ‮ايه المومنون‬‎ ‘O believers’ (Q24:31), ‮يايه الساحر‬‎ ‘O sorcerer!’ (Q43:49) and ‮ايه الثقلان‬‎ ‘O two dependents’ (Q55:31).48 To my knowledge no other cases of defective final ā.

For defective ū we find: ‮يدع الانسان‬‎ ‘man supplicates’ (Q17:11), ‮ويمح الله‬‎ ‘and God eliminates’ (Q42:24), ‮يدع الداع‬‎ ‘the caller calls’ (Q54:6), ‮سندع الزبانيه‬‎ ‘we will call the angels of Hell’ (Q96:18), ‮صلح المومنين‬‎ ‘the righteous ones of the believers’ (Q66:4).

Most common, however, is the shortening of ī before ʔalif al-waṣl: ‮سوف يوت الله‬‎ ‘God will bring’ (Q4:146), ‮ننج المومنين‬‎ ‘we save the believers’ (Q10:103), ‮بالواد المقدس‬‎ ‘in the sacred valley’ (Q20:12; Q79:16), ‮لهاد الذين‬‎ ‘verily a guide of those who’ (Q22:54), ‮واد النمل‬‎ ‘the valley of ants’ (Q27:18), ‮اتين الله‬‎ ‘God has given me’ (Q27:36),49 ‮شطى الواد الايمن‬‎ ‘the rightmost side of the valley’ (Q28:30), ‮العمى ٮهد ‬‎ ‘guide of the blind’ (Q30:53),50 ‮يردن الرحمن‬‎ ‘The beneficent intends for me’ (Q36:23), ‮صال الجحيم‬‎ ‘the burning one in hell’ (Q37:163), ‮يعباد الذين‬‎ ‘O my slaves who …’ (Q39:10), ‮الجوار المنشيت‬‎ ‘the elevated ships’ (Q55:24),‮51‬‎ ‮الجوار الكنس‬‎ ‘running, disappearing’ (Q81:16).

A special case is ‮يقص الحق‬‎ (Q6:57) which is variously read as yaqḍi l-ḥaqq and yaquṣṣu l-ḥaqq (Ibn al-Ǧazarī, § 3029). Only in the former option are we dealing with the shortening of ī before ʔalif al-waṣl. However, I agree with Sadeghi (2013) that the second reading is probably original.

A.3 Phonology

A.3.1 Consonants

The consonantal system as can be reconstructed for Quranic Arabic based on the QCT has been illustrated in Van Putten (2019b). The table below reproduces the reconstructed phonological system, when the sign used in the transcription does not correspond to the IPA realization, the IPA realization is written behind it. In some cases, I have simplified technically ambiguous realizations of sounds to the most likely realization based on modern taǧwīd realizations. For a more detailed discussion of other likely realizations, I refer the reader to Van Putten (2019b).

The ˭ sign, which usually marks ‘unaspirated’ is here to be understood as the glottis standing in prephonation state, i.e. a somewhat closed glottis which stops strong turbulent airflow (which leads to aspiration in stops and “voicelessness” in fricatives). For the stops this auditorily presents itself as unaspirated stops, and for the which is likewise maǧhūr, as a fricative without turbulent airflow (van Putten 2019b, 7–12).









t [tʰ], d, ṭ [tˁ˭]

k [kʰ], ǧ [ɟ], q [q˭]




[θ], [ð], [ðˁ]

ḍ [ɮˁ]

x [χ], ġ [ʁ]

[ħ], ʕ


s, z, ṣ [sˁ˭]

š [ʃ]







y [j], w

A.3.2 The Loss of the Hamzah

As has long been recognized, the orthography of the Quran seems to reflect a dialect that has lost the hamzah in most environments. In an earlier article, I have shown that rhyme confirms that this is not purely an orthographic idiosyncrasy, but is an accurate reflection of how the Quran was pronounced (van Putten 2018).52 The table below illustrates the main phonological developments caused by the loss of hamzah, along with examples.



*Cʔv > Cv

*yasʔalu > ‮يسل‬‎ /yasal/ (Q70:10)

*al-ʔafʔidati > ‮الافده‬‎ /al-afidah/ (Q104:7)

*ǧuzʔun > ‮جز‬‎ /ǧuz/ (Q15:44)

*vʔC > v̄C

*yaʔkulu > ‮ياكل‬‎ /yākul/ (Q10:24)

*yuʔminu > ‮يومن‬‎ /yūmin/ (Q2:232)

*aḏ-ḏiʔbu > ‮الذيب‬‎ /aḏ-ḏīb/ (Q12:13)

*Uʔ > i/uWW

*barīʔun > ‮بري‬‎ /bariyy/ or /barī/53 (Q6:19)

*sūʔun > ‮سو‬‎ /suww/ (Q3:174)

Final *āʔ remains unchanged

*as-samāʔi > ‮السما‬‎ /as-samāʔ/ (Q14:24)

*ʔinšāʔan > ‮انشا‬‎ /inšāʔā/ (Q56:35)

*aʔa > ā

*saʔala > ‮سال‬‎ /sāl/ (Q70:1)

*aʔi/u > aWi/u?

*baʔisa > ‮بيس‬‎ /bayis/ (Q11:99)

*naqraʔu-hū > ‮نقروه‬‎ /naqrawu-h/ (Q17:93)

*i/uʔā̆ > i/uWā

*muʔaḏḏinun > ‮موذن‬‎ /muwaḏḏin/ (Q7:44)

*fiʔatun > ‮فيه‬‎ /fiyah/ (Q3:13)

*i/uʔU > U

*mustahziʔūn > ‮مستهزون‬‎ /mustahzūn/ (Q2:14)

*al-mustahziʔīna > ‮المستهزين‬‎ /al-mustahzīn/ (Q15:95)

*ruʔūsakum > ‮روسكم‬‎ /rūsakum/ (Q2:196)

*aʔU > aW

*yaṭaʔūna > ‮يطون‬‎ /yaṭawn/ (Q9:120)

*baʔīsin > ‮بيس‬‎ /bays/54 (Q7:165)

*āʔi/u > āWi/u

*sāʔilun > ‮سايل‬‎ /sāyil/ (Q70:1)

*duʕāʔu-kum > ‮دعاوكم‬‎ (Q25:77)

A.3.3 Vowels

As for the vowel system of Quranic Arabic, it shares with Classical Arabic the short vowel system a, i, u and likewise shares the long vowels ā, ī and ū. However, the Classical Arabic ā corresponds to ā, ē and ō. The table below represents the phonemic system of Quranic Arabic that can be reconstructed from the QCT.















Besides the Classical Arabic long vowels /ā/, /ī/ and /ū/, Quranic Arabic had a fourth phonemic vowel which was written with a yāʔ and likely pronounced as /ē/, e.g. ‮هدى‬‎ /hadē/ ‘he guided’ (passim). It is clear from the rhyme that this was a separate sound from final /ā/ written with ʔalif, § 5.8, as they do not cross-rhyme. This fourth vowel /ē/ should not be seen as a variant of ā, which its Arabic name ʔalif maqṣūrah bi-ṣūrat al-yāʔ ‘the shortened ʔalif with the shape of the yāʔ’ might suggest, nor should its pronunciation ʔimālat al-ʔalif naḥw al-yāʔ ‘The leaning of ā in the direction of ī’ be understood as a historical process, which was not the concern of the Arab grammarians. Instead, these are purely descriptive terms. Van Putten (2017a) has shown that not only is the ʔalif maqṣūrah bi-ṣūrat al-yāʔ in the Quran pronounced differently, it also has a different historical background from the ʔalif maqṣūrah bi-ṣūrat al-ʔalif and is fully phonemic so that it cannot be understood as an allophone. The table below gives an overview of some of the instances of Quranic Arabic /ē/ and the etymological origins from which it develops. It likewise shows that the outcome of the original triphthongs containing *y is orthographically distinct from those that contain *w and original .55


Quranic Arabic


Classical Arabic






‘he guided’










‘a reminder’





‘he led him’





‘a precaution’





‘he invoked’





‘he called him’

Verbs with final /ē/ in early Quranic manuscripts, dissimilate to /ā/, written as ‮ـا‬‎ or defectively when the pronominal suffix -nī/-ni follows (‮ـنى‬‎ or ‮ـن‬‎). This same development happens when the 1sg. suffix -ya (‮ـى‬‎) follows a noun that ends in /ē/. Van Putten (forthcoming) has argued that, since this spelling difference is phonetically conditioned, we are likely dealing with a regular dissimilation of ē to ā in the vicinity of ī or y. This difference in spelling has mostly been lost in the CE, where these verbs and nouns are treated exactly the same before the 1sg. suffixes as before any of the other pronominal suffixes. For a full overview see Van Putten (forthcoming) but, one finds for example Q7:143: ‮ترانى‬‎ (W; SM1a; GK; BL; CPP; 330g; DAM29 ‮ترنى‬‎) where the CE has ‮ترينى‬‎.

A small group of nouns in Quranic Arabic are written with a final ‮ـوه‬‎. These are ‮صلوه‬‎ ‘prayer’ (passim), ‮زكوه‬‎ ‘alms’ (passim), ‮حيوه‬‎ ‘life’ (passim), ‮منوه‬‎ ‘Manāt’ (Q53:20), ‮بالغدوه‬‎ ‘in the morning’ (Q6:52; Q18:28); ‮النجوه‬‎ ‘the salvation’ (Q40:41) and ‮كمشكوه‬‎ ‘like a niche’ (Q24:35). While these words are often explained as representing an orthographic innovative way of writing word-internal /ā/, based on Aramaic spellings of some of these words, Al-Jallad (2017c) shows that this explanation is not very convincing. It is clear that all the words of Arabic origin in this list originally had a sequence *awat which monophthongized to /-ōh/ (see also § 5.3).

Another word that may have had the phoneme /ō/ is ‮الربوا‬‎ /ar-ribō/ ‘usury’ (Q2:275 (3×), 276, 278; Q3:130; Q4:161). The spelling with wāw ʔalif while in Classical Arabic ending up as ā, similar to the /ō/ of /ṣalōh/ becoming ṣalāh, ṣalāt may suggest that this word was /ar-ribō/. The etymology of this word is rather unclear (Rabin 1951, 109, § v), and current accounts of the phoneme /ō/ do not predict native words to have /ō/ in word-final position (Al-Jallad 2017c; van Putten 2017a). There is also no forthcoming explanation why the indefinite form of this noun apparently shifts this /ō/ to /ā/, as it is spelled ‮ربا‬‎ (Q30:39). Some nouns that etymologically end in a stem *-āʔ may have shifted to *-āʔu to /ō/ in the nominative, as is discussed in more detail in A.4.11.

A.3.4 Loss of Final Short Vowels and tanwīn

From the internal rhyme found in the Quran, it seems clear that what are considered the pausal pronunciations of final short vowels and tanwīn are in fact also the pronunciation in verse internal position as well (van Putten and Stokes 2018). Hence the developments that have taken place are the following, *u, *i, *a, *un and *in are lost word-finally, whereas *an has shifted to ā. Case and mood vowels appear to have been retained in construct, however. This reconstruction seems to be further confirmed by the Quranic orthography which indeed lacks any sign of regular tanwīn that we would have otherwise expected.

While throughout most of the corpus the generalization of this reduced case/mood system is borne out, there are a couple of Sūrahs that appear to tell a different story, at least in pausal position. In several final short *-a appears to have been lengthened. In some cases, this appears in the QCT, and is further confirmed by the rhyme, e.g.: ‮الظنونا‬‎ /aẓ-ẓunūnā/ ‘the assumptions’ (Q33:10), ‮الرسولا‬‎ /ar-rasūlā/ ‘the messenger’ (Q33:66) and ‮السبيلا‬‎ /as-sabīlā/ ‘the way’ (Q33:67). To this we may add as well the diptotic plural with an apparent diptotic accusative ending: ‮قواريرا‬‎ /qawārīrā/ ‘crystal clear’ (Q76:15, 16)56 and ‮سلسلا‬‎ /salāsilā/ ‘chains’ (Q76:4).

In other cases, the spelling is not changed, but the rhyme that such words appear in make it clear that they are to be read with final /-ā/, thus every other case of the accusative of as-sabīl in rhyme is spelled ‮السبيل‬‎ but certainly rhymed /as-sabīlā/ (Q4:44, Q25:17, Q33:4). Likewise, two subjunctives seem to rhyme with final /-ā/ ‮ان ازيد‬‎ /an azīdā/ ‘that I should add’ (Q74:15) and ‮لن يحور‬‎ /lan yaḥūrā/ ‘he will not return’ (Q84:14). Note that these are isolated exceptions, and both the definite accusative and the subjunctive occurs hundreds of times in rhyme where they are not pronounced as /-ā/. How to understand the exceptional status of these rhymes (which mostly concentrate and Q33 and Q76) requires further research.57

A.3.5 Assimilation Across Vowels

A major feature of Quranic Arabic that distinguishes it quite clearly from later Classical norms is its assimilation of identical and coronal consonants across vowels, while some of these ambiguous cases lead to disagreement between the Quranic readers, there is not a single reading that shows no signs of this assimilation at all.

For assimilation across vowels where the consonants are identical, it mostly concers with the first-person clitics -nī and -nā. The table below illustrates the examples. In some of these cases there is a disagreement between the regional codices, where one of the codices has an unassimilated form where the other does, in such cases I have given the abbreviated code (S = Syria, M = Medina, B = Basra, K = Kufa, C = Mecca) of the regional codex that has the minority variant. The unmarked version is then the variant that occurs in all other codices (see Sidky 2021; Cook 2004).



Quranic recitation

‮اتحجونى‬‎ (Q6:80)


ʔa-tuḥāǧǧūn-nī, ʔa-tuḥāǧǧū-nī58

‮تامرونى‬‎ (Q39:64)


taʔmurūn-nī/-niya, tāmurūn-nī, taʔmurū-niya, tāmurū-niya59

‮تامروننى‬‎ (Q39:64, S)



‮تامنا‬‎ (Q12:11)


taʔmanʷ-nā, tāmanʷ-nā, tāman-nā60

‮مكنى‬‎ (Q18:95)



‮مكننى‬‎ (Q18:95, C)



‮فنعما‬‎ (Q2:271)


fa-naʕim-mā, fa-niʕim-mā, fa-niʕm-mā, fa-niʕĭm-mā62

‮نعما‬‎ (Q4:58)


naʕim-mā, niʕim-mā, niʕm-mā, niʕĭm-mā

‮انى‬‎ (Q2:30 & passim)



‮انى‬‎ (Q2:47 & passim)



‮انا‬‎ (Q2:14 & passim)



‮انا‬‎ (Q4:66 & passim)



‮لياتينى‬‎ (Q27:21)


la-yaʔtiyan-nī, la-yātiyan-nī63

‮لياتيننى‬‎ (Q27:21, C)



‮ترينى‬‎ (Q23:93)



One might be tempted to understand such assimilation taking place as evidence that in Quranic Arabic the intervening short vowels of these stems had been lost, even before clitics. Interpreted in this way, these would not be examples of assimilation across short vowels. For early Christian Arabic, which shows similar cases, e.g. ‮اخزنى‬‎ /ʔaxzan-nī/ ‘he grieved me’, ‮امكنى‬‎ /ʔamkan-nī/ ‘it was possible for me’, ‮يدينى‬‎ /yadīn-nī/ ‘you judge me’. Blau (1967, § 35.4; § 41.4) indeed interprets these as evidence for that.

It is worth making several more observations here however: even when the previous consonant is completely unvocalized in the reading traditions, identical consonants following each other may be written twice, thus the jussive ‮يدرككم‬‎ (Q4:78) recited as yudrik-kum ‘(death) will overtake you’ is written with two kāfs, ‮يوجهه‬‎ (Q16:76) recited as yuwaǧǧih-hu and ‮يكرههن‬‎ (Q24:33) recited as yukrih-hunna are written with two hāʔs. But assimilation written out may also happen as found in the jussive ‮تفتنى‬‎ (Q9:49) recited as taftin-nī. As such, the fact that e.g. ‮يدعوننى‬‎ (Q12:33) ‘they call me’ is written with two nūns does not necessarily prove the pronunciation /yadʕūna-nī/, it could just as well stand for /yadʕūn-nī/ with morphophonological spelling. However, in light of the fact that nouns followed by pronominal clitics appear to have kept their final short vowels (A.3.4), it seems reasonable to assume that this is the case for verbs too. The examples given above are therefore not evidence for the lack of final short vowels, but rather examples of assimilation across vowels, a phenomenon of which there are many more examples in Quranic Arabic where we cannot propose the absence of an intervening vowel as we will see below.

Assimilation of identical consonants across a vowel also rarely occurs in the jussives of geminated verbs like yamdud. These forms are far outnumbered by cases where the metathesis did not take place, but it is worth mentioning all the cases here. If the same word also occurs elsewhere unassimilated, I have included them in this table as well. When regional variants play a role letter codes are given once again.



Quranic recitation

‮يرتدد‬‎ (Q2:217)



‮يرتدد‬‎ (Q5:54, SM)



‮يرتد‬‎ (Q5:54)



‮يرتد‬‎ (Q27:40)



‮يشاقق‬‎ (Q4:115)



‮يشاق‬‎ (Q59:4)



‮يضار‬‎ (Q2:282)


yuḍārra, yuḍār65

‮تضار‬‎ (Q2:233)


tuḍārra, tuḍārru, tuḍār66

Another place where the QCT irregularly has assimilation across short vowels is in the tD- and tL-stems, where the ta- prefix may be optionally assimilated to the following coronal consonant. This may happen both in the suffix conjugation and in the prefix conjugation, although in the latter it is much more common. The seven cases of this assimilation in the prefix conjugation are the following.



Reading Traditions

‮فادرتم‬‎ (Q2:72)


fa-ddāraʔtum(ū), fa-ddārātum(ū)

‮فاطهروا‬‎ (Q5:6)



‮اداركوا‬‎ (Q7:38)



‮اثاقلتم‬‎ (Q9:38)



‮وازينت‬‎ (Q10:24)



‮اطيرنا‬‎ (Q27:47)



‮ادرك‬‎ (Q27:66)



For the prefix conjugation there are many more examples, but are cause for some disagreement between the readers. When a ya- prefix precedes an assimilated tD/tL-stem, all readers are in agreement that the prefix assimilates, but when it stands before a ta- prefix, some rather see it as the haplological avoidance of the sequence ta-ta- > ta-, similar to ‮ولا تفرقو‬‎ /wa-lā tafarraqū/ ‘do not become disunited’ (Q3:103) but ‮ولا تتفرقوا‬‎ /wa-lā tatafarraqū/ ‘id.’ (Q42:13). As both haplology avoidance and assimilation occur in the QCT, it is not possible to be certain in those cases whether we are dealing with assimilation or haplology. The table below gives several illustrative examples of the problem using the common verb taḏakkara as the basis for examples.



Reading Traditions

‮يذكر‬‎ (Q2:269)



‮يتذكر‬‎ (Q13:19)



‮تتذكرون‬‎ (Q6:80)



‮تذكرون‬‎ (Q6:152)

/taḏḏakkarūna/, /taḏakkarūna/

taḏakkarūna, taḏḏakkarūna

There is no way to decide what the intended pronunciation was of a second person, or third person feminine prefix conjugation verb when the next consonant can be assimilated, and the Quranic readings do not seem to retain a historical memory of it, and rather have complex generalized rules. For example, the Kufans always read ta- followed by an assimilatable consonant assuming haplology (thus taḏakkarūna), whereas the other readers always assume assimilation (thus taḏḏakkarūna) (Ibn al-Ǧazarī § 3084).

Other coronal consonants may occur assimilated as well, are unattested in the prefix conjugation:



Reading Traditions

‮يشقق‬‎ (Q2:74)



‮يطوف‬‎ (Q2:158)



‮يصدقوا‬‎ (Q4:92)



‮يضرعون‬‎ (Q7:94)



‮يدبروا‬‎ (Q23:68)



‮يسمعون‬‎ (Q37:8)



‮يزكى‬‎ (Q80:3)


yazzakkā, yazzakkē, yazzakkǟ

As was the case with the assimilation of identical consonants across vowels, this type of assimilation is also cause for some disagreement between the regional codices, for Q7:3 the Syrian codex has ‮ٮتذكرون‬‎ recited by Ibn ʕāmir as yataḏakkarūna whereas the non-Syrian codices have ‮تذكرون‬‎ variously recited as taḏakkarūna or taḏḏakkarūna. In light of the non-Syrian reading, the more natural reading of Q7:3 in the Syrian codex is probably tataḏakkarūna, a reading that is indeed reported for Ibn ʕāmir (as a non-canonical transmission) and for ʔabū al-Dardāʔ (Ibn Xālawayh muxtaṣar, 42; Ibn Muǧāhid, 278). ʔabū al-Dardāʔ was one of the companions of the prophet who indeed died in Damascus, and was one of Ibn ʕāmir’s teachers (Ibn al-Ǧazarī, § 627), it is therefore no surprise that he would have the expected Syrian variant.

The assimilation across vowels of ta- to a following coronal may even happen if the ta- is preceded by a consonant. This seems to occur in the Ct-stem istaṭāʕa ‘to be able’, but may, depending on the interpretation, also occur with the Gt-stems ihtadā and ixtaṣama, for a discussion on the interpretation of the Gt cases see § 5.10. The table below illustrates some examples.



Reading Traditions

‮استطاعوا‬‎ (Q18:97)



‮اسطاعوا‬‎ (Q18:97)


isṭāʕū, isṭṭāʕū67

‮لم تستطع‬‎ (Q18:78)

/lam tastaṭiʕ/

lam tastaṭiʕ

‮لم تسطع‬‎ (Q18:82)68

/lam tasṭṭiʕ/

lam tasṭiʕ

‮يهتدى‬‎ (Q10:108)



‮يهدى‬‎ (Q10:35)

/yahddī/ (or /yahdī/?)

yahiddī, yahaddī, yahăddī, yahddī, yahdī69

‮تعتدوا‬‎ (Q2:190)



‮تعدوا‬‎ (Q4:154)

/taʕddū/ (or /taʕdū/?)

taʕddū, taʕaddū, taʕăddū, taʕdū70

‮يختصمون‬‎ (Q26:96)



‮يخصمون‬‎ (Q36:49)

/yaxṣṣimūn/ or /yaxṣimūn/

yaxiṣṣimūna, yaxaṣṣimūna, yaxăṣṣimūna, yaxṣṣimūna, yaxṣimūna71

These examples should make it clear that syncope of short vowels between two identical consonants, and assimilation of t to coronals across a vowel happens quite frequently all throughout the QCT. It seems to have always been optional, and for almost every single example of such a phenomenon there are examples where the assimilation did not take place as well. Its distribution does not present an obvious explanation for this variation. The fact that both ‮لم تستطع‬‎ and ‮لم تسطع‬‎ occur only several verses apart, and ‮استطاعوا‬‎ and ‮اسطاعوا‬‎ even occur in the same verse, give us little reason to suggest that this is due to multiple authors or scribes for different parts of the Quran. It seems that we must conclude that such assimilations across vowels were a free variant option in the language of the Quran, which infrequently occurred regardless of environment.

The freedom between different assimilated and unassimilated forms is in fact so close, that almost perfectly parallel verses may occur both with assimilated and unassimilated forms, for example:

Q6:42 wa-laqad ʔarsalnā ʔilā ʔumamin min qabli-ka fa-ʔaxaḏnā-hum bi-l-baʔsāʔi wa-ḍ-ḍarrāʔi laʕallahum yataḍarraʕūna

We have sent already unto peoples that were before thee, and we visited them with tribulation and adversity, so that perhaps they might grow humble.

Q7:94 wa-mā ʔarsalnā fī qaryatin min nabiyyin ʔillā ʔaxaḏnā ʔahla-hā bi-l-baʔsāʔi wa-ḍ-ḍarrāʔi laʕallahum yaḍḍarraʕūna

And we sent no prophet unto any town except to visit its people with tribulation and adversity, so that perhaps they might grow humble.

Another case of this process is attested in the non-canonical readings, and is well-attested in vocalized Quranic manuscripts for the verb ‮يخصفان‬‎ (Q7:22; Q20:121) ‘they covered (themselves)’, which is read by the canonical readers as yaxṣifāni but is attested vocalized as yaxaṣṣifāni, yaxiṣṣifāni in kufic manuscripts (e.g. Q7:22 in Arabe 334j; Q20:121 in Arab 325j, 347a), which is clearly an assimilated form of the Gt-stem yaxtaṣifāni. These forms are attested in the Šāḏḏ literature too, e.g. Ibn Xālawayh (muxtaṣar, 42, 90).

A.3.6 Pausal Shortening of

Quranic Arabic has two realizations of word-final , it can either be written with a ‮ـى‬‎ or with no mater at all. This concerns any type of word-final yāʔ: (1) Final of definite III-y nouns, e.g. ‮الواد‬‎ ‘the valley’ (Q89:9); (2) 1sg. Object pronoun -nī, e.g. ‮فاتقون‬‎ ‘fear me’ (Q2:41); (3) 1sg. possessive pronoun , e.g. ‮دين‬‎ ‘my religion’ (Q109:6); (4) The of imperfect verbs (only once): ‮يسر‬‎ ‘it passes’ (Q89:4).

Van Putten & Stokes (2018, 156–158) showed that these shortened forms are overwhelmingly favoured in pausal positions. While long forms rarely occur in pause, and shortened forms only occasionally occur outside of pause. Thus, pause seems to be quite clearly the origin for the shortening. The fact that these forms stand in rhyme where the vowel is entirely unpronounced, suggests that the was not shortened, but dropped altogether, which would mean the pausal form of the 1sg. possessive marker was zero-marked.

A.3.7 *sayyiʔāt as ‮سيات‬‎ Reflecting /sayyāt/

Original *sayyiʔāt- ‘evil deeds’ in the CE is regularly spelled as ‮سيات‬‎, ‮السيات‬‎ seemingly with an ʔalif in the position of the .72 In early manuscripts this spelling is not always regular. It is outside the scope of the current work to examine this spelling in every single manuscript. Instead, below I have listed the spellings for every single occurrence of the word in the CPP. It becomes clear that the spelling ‮سييت‬‎, more in line with the normal orthographic practices of the QCT, occurs besides ‮سيات‬‎.

‮سياتنا‬‎ (Q3:193), ‮السيات‬‎ (Q10:27)
‮سييتهم‬‎ (Q3:195; Q25:70), ‮السييت‬‎ (Q4:18; Q7:153, 168; Q42:25; Q45:21, 33), ‮سييتكم‬‎ (Q4:31; Q5:12; Q8:29), ‮سييته‬‎ (Q65:5)

As with most other CayyiC adjectives, sayyiʔ has a contracted by-form sayʔ in Classical Arabic (Lane, 1491a, see also al-Farrāʔ Luġāt, 30), cf. ḍayyiq-, ḍayq- ‘narrow’ (Lane, 1868b), mayyit-, mayt- ‘dead’ (Lane, 2800b) and layyin-, layn- ‘soft’ (Lisān, 4117b). It seems then that the spelling ‮سيات‬‎ should be considered the outcome of this contracted form, i.e. /sayyāt/ or /sayāt/, whereas the spelling ‮سييت‬‎ represents the uncontracted form /sayyiyāt/. The ʔalif then is not a sign for the hamzah, but rather the result of the regular rule for the plene spelling of the plural feminine ending which occurs if the word would otherwise consist of only three letters (see Appendix A.2.1).

Whether the QCT indeed originally showed free variation between the contracted or uncontracted form, or whether the CE is correct in only showing the contracted form is a question that cannot be addressed in the current work.

A.3.8 A Case of N-Assimilation?

While by no means regular, there are two examples in the QCT where the sequence of two nūns is simplified to just a single nūn, namely: Q21:88 ‮نجى‬‎ ‘we save’, Q12:110 ‮فنجى‬‎ ‘so we save’ (cf. Q10:103 ‮ننج‬‎ ‘we save’) which are read as nuǧǧī/nunǧī and fa-nuǧǧiya/fa-nunǧī respectively.73 If this is not simply a writing error that has been propagated from the archetype, this should probably be understood as an isolated case of an assimilated n to the following ǧ.

A.3.9 The Genitilic Adjective Ending

In most modern dialects, the gentilic adjective ending (Nisbah) is for the masculine and -iyya(h) for the feminine. While the masculine form has become a fairly common place transcription of the Classical Arabic gentilic adjective, normatively it is to be pronounced as -iyy even in pausal pronunciation. The Quranic rhyme suggests that the simplification of -iyy to took place in Quranic Arabic as well, whereas the indefinite accusative remained -iyyā.

Q20:85, 87 ‮السامرى‬‎ /as-sāmirī/ ‘The Samaritan’ rhymes with Q20:86 ‮موعدى‬‎ /mawʕid-ī/ ‘promise to me’, Q20:88 ‮فنسى‬‎ /fa-nasī/ ‘so he has forgotten’ and Q20:95 ‮يسمرى‬‎ /yā-sāmirī/ ‘O Samaritan!’ rhymes with Q20:94 ‮قولى‬‎ /qawl-ī/ ‘my word’ and Q20:95 ‮نفسى‬‎ /nafs-ī/ ‘my soul’.

Q19:16 ‮مكانا شرقيا‬‎ /makānā šarqiyyā/ ‘an eastern location’ rhymes with Q19:17 ‮بشرا سشيا‬‎ /bašarā sawiyyā/ ‘an able-bodied man’.

The feminine gentilic adjective ending would presumably have been /-iyyah/, but it is unattested in rhyme position.

A.3.10 ʔalif al-waṣl

In Classical Arabic, there is a significant group of words that start with an initial vowel, which is elided when another vowel precedes it.74 These can be found in five main environments.

  1. The definite article: (a)l-bašar

  2. A small group of nouns such as (i)sm ‘name’, (i)bn ‘son’, (i)mruʔ ‘man’, (i)mraʔah ‘woman’.

  3. Imperative verbs, (i)fʕal, (u)ktub

  4. Gt-, N- and Ct-stem verbs: (i)ftaʕala, (i)nfaʕala, (i)stafʕala

  5. Assimilated tD- and tL-verbs: (i)ḏḏakara, (i)ṯṯāqala

From the QCT, it is not at all clear that such an elision takes place in Quranic Arabic, as the prothetic vowel is spelled morphophonologically, so even when a particle precedes that would cause the ʔalif al-waṣl to be elided, is still written. From the orthography it is therefore equally possible that the ʔalif was actually pronounced in such cases.

From the Damascus Psalm fragment, we learn that it need not be the case that all contexts of the ʔalif al-waṣl are equal in this regard. There the ʔalif al-waṣl of the definite article is elided in much the same way as in Classical Arabic e.g. οελναρ /wa-l-nār/ ‘and the fire’ (v. 21), βιλλαυ /bi-ḷḷāh/ (v. 22), φιλ•β[…] /fi l-b[ariyyah]/ ‘in the wilderness’ (v. 52), φιλ•βαχερ /fi l-bašar/ ‘among men’ (v. 60), λιλ•σεβ• /li-l-sab(y)/ ‘into captivity’. However, the Gt- and N-stems seem to have a true hamzat al-qaṭʕ, e.g. οα•αβ•τε•λεῦ /wa-ʔabtalaw/ ‘they tempted’ (v. 56) and φα•ανκα•λε•β(ο)υ• /fa-ʔanqalabū/ ‘and they turned their backs’ (v. 57) (for the analysis of the Damascus Psalm Fragment see Al-Jallad 2020b, 79 ff.). As already pointed out by Al-Jallad (2020b, 51, 60), it is therefore not a given that the QCT orthography represented a linguistic situation identical to Classical Arabic rather than the situation identical to that of the Damascus Psalm Fragment. In this section we will examine each of the five environments, and considered the evidence for the elision of the ʔalif al-waṣl in each of them.

In the QCT, it is regular to drop the ʔalif al-waṣl of the definite article when la- or li- precedes, e.g. ‮الحمد لله‬‎ /al-ḥamd li-llāh/ ‘praise be to God’ (Q1:2), ‮هدى للمتقين‬‎ /hudē li-l-muttaqīn/ ‘a guidance to the god fearing’ (Q2:2), ‮وانه للحق من ربك‬‎ /wa-inna-h la-l-ḥaqq min rabbi-k/ ‘for this is indeed the truth from your lord’ (Q2:149). In early manuscripts this behaviour is quite frequent, although never regular, when bi- precedes the definite article. This is especially common in the phrase ‮بالحق‬‎ /bi-l-ḥaqq/ ‘with the truth’ (Cellard 2018, 8), although not exclusively, e.g. ‮بلمعروڢ‬‎ /bi-l-maʕrūf/ ‘what is fair’ (Cellard 2018, ٤٧-٤٨, l. 5), ‮بلامس‬‎ /bi-l-ams/ ‘yesterday’ (Cellard 2018, ٧٧-٧٨, l. 6). Very rarely the preposition ka- has the same effect, e.g. ‮كلمهل‬‎ /ka-l-muhl/ ‘like molten brass’ (Cellard 2018, ١٩٩-٢٠٠, l. 10). If wa- or fa- or precede, the ʔalif al-waṣl is always written.

The only possible example that may be cited of an example where the ʔalif al-waṣl of the definite article is perhaps left unwritten is the phrase ‮ولدار الاخره‬‎ (Q12:109; Q16:30). This is recited as a construct phrase as wa-la-dāru l-(ʔ)āxirati, however ‮وللدار الاخره‬‎ (Q6:32), recited as wa-la-d-dāru l-(ʔ)āxiratu,75 suggests that this might not be a construct phrase with the asseverative particle la- in front of it, but rather the single lām represents the definite article, i.e. /wa-d-dār al-āxirah/ (Nöldeke et al. 2013, 397, fn. 56).

Despite the frequent morphophonological spelling then, it seems clear that indeed the vowel of the definite article was elided if a particle preceded. This is further confirmed by the fact that, occasionally, word-final long vowels are spelled defectively when they immediately precede a definite article, e.g. ‮سوف يوت الله المومنين اجرا عظيما‬‎ /sawf yūti (< yūtī) llāh al-mūminīn aǧrā ʕaẓīmā/ ‘Allah will bring the believers a great reward’ (Q4:146), ‮صلح المومنين‬‎ /ṣāliḥu (< ṣāliḥū) l-mūminīn/ ‘the righteous among the believers’ (Q66:4) and ‮ايه اللذين امنوا‬‎ /ayyuha (< ayyuhā) l-mūminūn/ ‘O believers!’ (Q24:31) (see A.2.10).

There is very little direct evidence that the ʔalif al-waṣl on words such as imraʔah and imruʔ was elided. However, the basmalah formula is written ‮بسم الله‬‎ /bi-smi llāh/ and never ‮باسم الله‬‎. This is a strong indication of the elision of this ʔalif al-waṣl. Outside of the basmalah, bi-smi occurs occasionally with the morphophonological spelling as well, though this is cause for some disagreement among early Quranic manuscripts, e.g. ‮باسم ربك‬‎ /bi-smi rabbi-ka/ (Q56:74) (see B.19). The CE attests ‮يبنوم‬‎ /ya-bna-wumm/ (Q20:94) ‘O son of my mother!’, which would be a good example of the elided ʔalif al-waṣl before ‮ابن‬‎, however this word is consistently spelled ‮يابنوم‬‎ in early Quranic manuscripts (see B.20).

While several I-ʔ verbs have irregular biradical imperatives such as kul ‘eat!’ and xuḏ ‘take!’, most verbs are treated as regular triradical verbs, with the loss of the hamzah in Quranic Arabic, however, these develop a special allomorphy, where the unprefixed imperative have an initial long vowel /ī/ whereas when they are prefixed by wa- or fa- these merged into /wā-/ and /fā-/. This behaviour can only be understood if we assume that such imperatives in an early stage of the language indeed had a non-phonemic initial i- in absolute initial position, *(i)ʔti > /īt/ but *fa-ʔti > /fāt/.



‮ايت‬‎ /īt/ (Q10:15)

‮فات‬‎ /fāt/ (Q2:258)


‮ايتوا‬‎ /ītū/ (Q20:64)

‮واتوا‬‎ /wātū/ (Q2:189)

‮فاتوا‬‎ /fātū/ (Q2:23)

‘come (pl.)!’

‮فاتيا‬‎ /fātiyā/ (Q26:16)

‘come (du.)!’

‮ايذن‬‎ /īḏan/ (Q9:49)


‮فاذنوا‬‎ /fāḏanū/ (Q2:279)

‘be informed (pl.)!’

‮وامر‬‎ /wāmur/ (Q7:145)


‮فاوا‬‎ /fāwū/ (Q18:16)

‘retreat (pl.)!’

While this behaviour clearly proves that such verbs had an ʔalif al-waṣl historically, it is not entirely clear that this is the case synchronically. Verbs of this type do not have the same morphological behaviour as in Classical Arabic. For example: ‮قل الذين لا يرجون لقانا ايت بقران غير هذا او بدله‬‎ can really only be understood as /qāl allaḏīn lā yarǧūn liqāʔa-nā īt bi-qurān ġayr hāḏā baddil-(u)h/ “Those who do not expect to meet us say: ‘bring a recital other than this or change it’ ”. Had the Classical pronunciation /liqāʔa-na ʔti/ or with loss of hamzah /liqāʔanāti/76 been intended, we would not expect ‮ايت‬‎ to have been spelled with the yāʔ. This behaviour clearly cannot be attributed to pausal spelling, as had that been the case, we would expect the form with wa- or fa- in front of it to also be written with the yāʔ, i.e. **‮فايت‬‎ ‘come/bring!’.77

Synchronically, it therefore seems that verbs of this type had a ʔalif al-qaṭʕ when there was not a direct proclitic in front of it. It seems possible that ‮واغفر‬‎ ‘and forgive!’ (Q2:285), ‮وانظر‬‎ ‘and see!’ (Q2:259), ‮واعلم‬‎ ‘and know!’ (Q2:26) are read as /wa-ġfir/, /wa-nẓur/ and /wa-ʕlam/ respectively, which would assume some amount of morphophonological spelling (something that is clear for the definite article as well) but alternatively /wa-iġfir/, /wa-unẓur/ and /wa-aʕlam/ cannot be excluded. The fact that, unlike the definite article, we never find phonetic spellings without the prothetic ʔalif may be interpreted as an indication that these indeed had ʔalif al-qaṭʕ.

When li- and la- precede the definite article, they always trigger an elided spelling of the ʔalif al-waṣl. This is not the case when la- precedes the ʔalif al-waṣl of derived verbs of the N-, Gt- or Ct-stem, which may suggest that, similar to the Damascus psalm fragment, these derived verbs indeed had a prefix ʔa- rather than ʔalif al-waṣl, e.g. ‮لاختلفتم‬‎ /la-ʔaxtalaftum/ ‘you would have differed’ (Q8:42), ‮لانفضوا‬‎ /la-ʔanqaḍḍū/ ‘they would have dispersed’ (Q3:159), ‮لاستكثرت‬‎ /la-ʔastakṯart/ ‘I would have multiplied’ (Q7:188).

There is one case against the presence of an ʔalif al-qaṭʕ in the Gt-stem. ‮لتخذت‬‎ ‘you would have taken’ (Q18:77) is recited as la-ttaxaḏtā, la-ttaxatta by most readers, despite the absence of the ʔalif al-waṣl in the QCT. The reading of Ibn Kaṯīr, ʔabū ʕamr and Yaʕqūb is la-taxiḏta, la-taxitta, which would not imply the elision of the ʔalif al-waṣl (Ibn al-Ǧazarī, § 3525). However, it is quite clear that this is the inferior reading. Ittaxaḏa is an irregular Gt-stem. Instead of the expected **iʔtaxaḏa, Quranic Arabic treats it as a derivation of a I-w verb. The G-stem taxiḏa is transparently an analogical backformation from ittaxaḏa. As the G-stem of ittaxaḏa is just the original ʔaxaḏa everywhere else in the Quran, e.g. ‮اخذتهم‬‎ /ʔaxaḏtu-hum/ (Q22:44), it is difficult to accept the sudden use of taxiḏa in this place only. Thus, the more natural reading of ‮لتخذت‬‎ is indeed /la-ttaxaḏt/, which suggests that the ʔalif al-waṣl was unpronounced, in line with Classical Arabic, and different from the Arabic of the Damascus psalm fragment. There is however a question whether the spelling ‮لتخذت‬‎ is in fact archetypical to the UT. While a good number of manuscripts indeed exhibit this spelling, several quite ancient manuscripts point to the expected spelling ‮لاتخذت‬‎, such as Saray Medina 1a (corrected to ‮لتخذت‬‎ by a later hand) (see B.21). If the spelling with the elided ʔalif al-waṣl is not original to the UT, then it once again becomes quite likely that the initial cluster was preceded by an ʔalif al-qaṭʕ instead.

Finally, the tD- and tL-stems as with the derive N-, Gt-and Ct-stems always write the ʔalif al-waṣl with an ʔalif, regardless of whether it is preceded by a proclitic or not. However, the only proclitics that occur before it are wa- (‮وازينت‬‎ /wa-(v)zzayyanat/ ‘and is embellished’, Q10:24) and fa- (‮فادرتم‬‎ /fa-(v)ddārātum/ ‘so they disputed’, Q2:72). These same proclitics also do not cause the elision of the ʔalif al-waṣl of the definite article in the orthography, which in proncuniation it was almost certainly unpronounced. As such, it is not readily possible to determine whether stems like these retained their epenthetic initial syllable if a clitic precedes.

A.3.11 An Isolated Case of Word-Initial *wu > ʔu

The Arab grammarians record the possibility of shifting word-initial *wu and *wū to ʔu and ʔū, e.g. wulida > ʔulida, and in wuǧūh > ʔuǧūh (Sībawayh IV, 331). This rule has made its way, not entirely regularly, into the textbook Classical Arabic as well (Fischer 2002, § 36b). While most of the time, this shift does not occur in the QCT, e.g. ‮ولد‬‎ (Q19:15) /wulid/ ‘he was born’ and ‮وجوه‬‎ (Q3:106) /wuǧūh/ ‘faces’, there is a single occurrence of this development, namely, ‮اقتت‬‎ /ʔuqqitat/ ‘the time has come’ (Q77:11)78 transparently from the root √wqt.79

A.4 Morphology

A.4.1 Independent Pronouns

Almost the complete paradigm of the independent pronouns is attested in the QCT, only the second person feminine plural is unattested.






/hū/, /huww/?







/hī/, /hiyy/?


















From the fact that the masculine plurals are spelled ‮هم‬‎ and ‮انتم‬‎ rather than ‮هموا‬‎ and ‮انتموا‬‎ make it obvious that Quranic Arabic did not employ the long forms of the plural pronouns, unlike some of the Hijazi reading traditions (§ 3.6.5).

The reconstruction of the phonetics of the third person singular pronouns requires some discussion. In the ʕarabiyyah these pronouns are consistently huwa and hiya, unless they stand in an environment where they may syncopate to wa-hwa and fa-hya (§ From a Semitic perspective, the ʕarabiyyah forms are surprising, the Hebrew forms ‮הוא‬‎ and ‮היא‬‎ are best understood as reflexes of Proto-West-Semitic *hūʔa and *hīʔa (Suchard 2019, 211). Both the loss of length and the loss of the in the ʕarabiyyah are irregular. Many modern dialects of Arabic have forms such as huwwa and huwwe (besides , ) (Fischer and Jastrow 1980, 80) which do not appear to be reflexes of *huwa and *hiya but rather of *hūʔa-h and *hīʔa-h, i.e. the Proto-West-Semitic pronouns followed by the -h pronominal extension also found in the Hebrew second person masculine pronoun ʔattå < *ʔanta-h, and in the third person pronouns as well in the dead sea scrolls ‮היאה ,הואה‬‎ (Suchard 2019, 210). For a discussion on these stem extensions see Al-Jallad (2014b).

The expected reflex of Classical Arabic *huwa in Quranic Arabic, after the loss of final short vowels, would be **hū. As we saw in A.2.3, word-final is usually written with an ʔalif al-wiqāyah, and therefore the expected spelling of our hypothetical **hū would be ‮هوا‬‎. Instead, we regularly find ‮هو‬‎, which would be the expected spelling for the reflex of *hūʔa > huww. On this basis we might want to posit the third person pronouns as *hūʔa > /huww/ and *hīʔa > /hiyy/ for Quranic Arabic. However, the fact that the pausal form ‮هيه‬‎ (Q101:10) rhymes as /hiyah/, seems to suggest that Quranic Arabic indeed goes back to a form closer to the one we find in Classical Arabic instead, which would make a reading as /hū/ and /hī/ more attractive, in which case the spelling of ‮هو‬‎ is irregular.80

A.4.2 Clitic Pronouns

The pronominal system of the Quranic reading traditions shows a large amount of variation, most of which is not continued in Classical Arabic (van Putten and Sidky forthcoming). As final short vowels are lost in Quranic Arabic, some of this variation present in the reading tradition was presumably not expressed at all. It is unclear to what extent there was vowel harmony between the case vowel and the following pronominal suffix in the masculine plural clitics, but reports of grammarians suggest that it was typical of the Hijaz to not have vowel harmony. This leads me to tentatively suggest that Quranic Arabic lacked vowel harmony as well, although there is no independent way to confirm this.

Lengthened forms of the singular pronouns -hū and -hī were certainly absent, as we would expect those to have been written as ‮ـهوا‬‎ and ‮ـهى‬‎. The same is true for the lengthened pronominal forms -humū, -himī, himū and -kumū which would be expected to be written ‮ـهمى ,ـهموا‬‎ and ‮ـكموا‬‎. The long form of the second person plural pronoun only occurs four times before other clitic pronouns (Q8:44; Q11:28; Q15:22; Q47:37). The table below illustrates the probable reconstruction of the pronominal suffix paradigm.



























/-kumū-/ (before pronouns)




1 (verbal)








1 (nominal)



/-ī/, /v̄-y/, /ē/,


Special mention needs to be made of the 1sg. pronoun which has several different allomorphs. Due to pausal shortening of final *-ī both the verbal /-nī/ and nominal /-ī/ also occur as /-n/ and /-∅/ respectively (see A.3.6). After long vowels, the 1sg. nominal suffix is /-y/. Finally, there likely was a special vocative 1sg. marker that shows up in expressions of woe, e.g. ‮ياسفى‬‎ /yā-ʔasaf-ē/ ‘O my sorrow!’ (Q12:84), ‮يحسرتى‬‎ /yā-ḥasrat-ē/ ‘O my regret!’ (Q39:56), and ‮يوليتى‬‎ /yā-waylat-ē/ ‘Woe is me!’ (Q5:31; Q11:72; Q25:28). While technically the spelling with ‮ـى‬‎ could be read as as well,81 the normal 1sg. ending, this is unlikely to be the intended reading here. Vocatives throughout the Quran consistently have the short pausal 1sg. ending, e.g. ‮يقوم‬‎ /yā-qawm-∅/ ‘O my people!’, ‮يابت‬‎ /yā-abat-∅/ ‘O my father!’, ‮يرب‬‎ /yā-rabb-∅/ ‘O my lord!’.82 Had the vocatives of woe had the normal 1sg. ending, we would have expected it to have been shortened as well. Moreover, in the canonical Quranic reading traditions this vocative 1sg. is indeed consistently read as -ē/-ǟ/-ā, as expected (Ibn al-Ǧazarī, § 1973, § 2041–2042).

The first singular possessive clitic occurs a few times in pausal position (all in Q69) with a final h, clearly confirmed by the rhyme to represent a reading /-iyah/: ‮كتبيه‬‎ /kitāb-iyah/ ‘my book’ (Q69:19, 25); ‮حسابيه‬‎ /ḥisāb-iyah/ ‘my reckoning’ (Q69:20; Q69:26); ‮ماليه‬‎ /māl-iyah/ ‘my property’ (Q69:28) ‮سلطنيه‬‎ /sulṭān-iyah/ ‘my authority’ (Q69:29). Elsewhere in the Quran the pausal 1sg. /-∅/ is used in verse final position.

A.4.3 Verbal Endings

The suffix conjugation of the perfective verb appears to have been identical to the pausal pronunciations of Classical Arabic. The 1p suffix /-nā/ is always spelled defectively in the QCT when it is followed by a pronominal clitic. This is presumably defective spelling, and does not indicate an actual shortening of the suffix to /-na/ in that context.










/-ū/, /-aw/

















/-tumū-/ (before clitic pronouns)









‮ـنا، ـنـ‬‎


The third person masculine plural ending ‮ـوا‬‎ /-aw/ would be the form that occurs in verbs that end in ʔalif maqṣūrah. This is indistinguishable from /-ū/ in the orthography of the QCT, but it seems reasonable to assume that Quranic Arabic retained this distinction. The third person masculine plural ending ‮ـو‬‎ /-w/, never followed by an ʔalif al-wiqāyah occurs on hollow roots with hamzah as final root consonant such as ‮جاو‬‎ /ǧāw/ ‘they came’ (e.g. Q3:184) and also ‮راو‬‎ /rāw/ ‘they saw’ (e.g. Q7:149).83

The prefix conjugation has two different sets of ending, depending on whether it represent the imperfective, or the subjunctive/jussive. Invariably the imperfective form is longer, and those forms are given in between brackets when necessary. The vowel of the prefix appears to have occurred in two forms either with an a (used for the G-, tD-, tL-, Gt-, N- and Ct-stems) and u (used for the D-, L- and C-stems). In Quranic Arabic there was no alternation in the prefix vowel between a and i as reported for some eastern dialects (see § 4.7).






/ya-/, /yu-/


/ya-, /yu/…/-ā(n)/


/ya-/, /yu-/…/-ū(n), -aw(n)/



/ta-/, /tu-/


/ta-/, /tu-/…/-ā(n)/


/ya-/, /yu-/…/-n/



/ta-/, /tu-/


/ta-/, /tu-/…/-ā(n)/


/ta-/, /tu-/…/-ū(n), -aw(n)/



/ta-/, /tu-/…/ī(n), -ay(n)/


/ta-/, /tu-/…/-ā(n)/


/ta-/, /tu-/…/-n/



/a-/, /u-/


/na-/, /nu-/

A.4.4 Demonstrative Pronouns

The near deixis demonstrative pronouns of Quranic Arabic have much less variation than is reported for Classical Arabic. It is seemingly a Hijazi innovation to always prefix the deictic pronouns with hā- (see § 4.5), save for certain specific archaic constructions, where traces of the ancient forms without ḏā are retained (see below).

Near deixis





‮هذا‬‎ /hāḏā/

‮هذن‬‎ /hāḏān/

‮هولا‬‎ /hāwulāʔ/ or /hawlāʔ/


‮هذه‬‎ /hāḏih/

‮هاتين‬‎ /hātayn/84

In Classical Arabic, the dual of the near deixis inflects for case, as a dual noun would, i.e. nom. hāḏāni gen./acc. hāḏayni. There is no evidence that this is the case in Quranic Arabic. The masculine dual occurs twice, once at Q22:19 ‮هذان خصمان‬‎ /hāḏān xaṣmān/ ‘these are two enemies’, with nominative function, and the other is the famous verse Q20:63 ‮ان هذن لاسحرن‬‎ /in(n) hāḏāni la-sāḥirān/ ‘indeed, these are two magicians’, where it functions as an accusative, where Classical Arabic would require hāḏāyni. However, as this is the only attestation of the near deixis dual pronoun in an accusative position, there is no reason to believe that this dual inflected for case.

The feminine dual is only attested in the gen./acc. and has the expected form ‮هاتين‬‎. This could either mean that at an earlier stage of Quranic Arabic, it did inflect for case and the masculine and feminine generalized different case forms, or that Q20:63 really is an error.

The far deixis in Quranic Arabic is marked by the deictic pronominal base, followed by a typically Hijazi element -l(i)- in the singular followed by the second person pronoun suffix, which can agree with the addressee.

Far Deixis





‮ذلك‬‎ /ḏāli-k/ (2sg.)

‮ذنك‬‎ /ḏāni-k/ (2sg)

‮اوليك‬‎ /ulāyi-k/ (2sg.)

‮ذلكما‬‎ /ḏāli-kumā/ (2du.)

‮ذلكم‬‎ /ḏāli-kum/ (2pl.m.)

‮اوليكم‬‎ /ulāyi-kum/ (2pl.m.)

‮ذلكن‬‎ /ḏāli-kunn/ (2pl.f.)


‮تلك‬‎ /til-k/ (2sg.)

‮تلكما‬‎ /til-kumā/ (2du.)

‮تلكم‬‎ /til-kum/ (2pl.m.)

While ‮ذلك‬‎ and ‮تلك‬‎ can clearly be used in environments where the addressee is plural, the other forms seem to always be explicitly used in addressee agreement. Fischer (2002, § 275.2) suggests that the addressee agreement in pre-classical Arabic no longer holds. This may be true for the poetry where these forms occur, but the system is evidently productive in the Quran.85

The locative deictics follow the same pattern as the pronominal deictics, where the near deixis always has the prefix hā- and the far deixis always has the -li- stem extension. There is no evidence for addressee agreement for the locative deictic.

Near deixis

Far deixis


‮ههنا‬‎ /hāhunā/

‮هنالك‬‎ /hunāli-k/

The Arab grammarians report forms of the short demonstrative without prefix hā- as a possible forms, this use of ‮ذا‬‎ /ḏā/ has fallen out of use in Quranic Arabic. In the QCT it is only attested after the interrogatives ‮من‬‎ /man/ ‘who’ and ‮ما‬‎ /mā/ ‘what’. The long interrogative /man ḏā/ is only used in the cleft construction ‮من ذا الذى‬‎ /man ḏā allaḏī/ ‘who is it that …?’ (Q2:245, 255; Q3:160; Q33:17; Q57:11). ‮ما ذا‬‎ /mā ḏā/ (passim)86 shows no obvious difference in meaning or syntax from /mā/.87 The long deictic can also be combined with ‮من‬‎ /man/: ‮امن هذا الذى‬‎ /ʔam-man hāḏā allaḏī/ ‘or who is it that …?’ (Q67:20, 21).

Classical Arabic has a construction of independent pronouns followed by the deictic elements with a presentative function. In such cases, the deictic lacks the hā- prefix but it may stand in front of the independent pronoun e.g. hā-ʔana ḏā ‘here I am!’, ʔanta ḏā ‘here you are’, hā-naḥnu ʔulāʔi ‘here we are!’ (Fischer 2002, § 279). Quranic Arabic attests this construction twice, both times with plural pronouns: ‮هانتم اولا تحبنهم‬‎ /hā-antum ulāʔ tuḥibbūna-hum/ ‘Here you are loving them’ (Q3:119) ‮هم اولا على اثرى‬‎ /hum ulāʔ ʕalā aṯar-ī/ ‘Here they are on my track’ (Q20:84).

Such constructions may also have the hā- prefix on the demonstrative after the pronoun, and the prefix may also occur on both: ‮انتم هولا تقتلون انفسكم‬‎ /antum hāwulāʔ taqtulūn anfusa-kum/ ‘Here you are killing one another’ (Q2:85); ‮هانتم هولا حججتم فيما لكم به علم‬‎ /hā-antum hāwulāʔ ḥāǧaǧtum fī-mā la-kum bi-h ʕilm/ ‘Here you are, having argued about that of which you have knowledge’ (Q3:66), see also Q4:109 and Q47:38.

A.4.5 Relative Pronouns

The relative pronouns, unlike the Classical Arabic spelling, is spelled with a single lām in Quranic Arabic in all its forms.88





‮الذى‬‎ /allaḏī/

nom. ‮الذان‬‎ /allaḏān/

obl. ‮الذين‬‎ /allaḏayn/

‮الذين‬‎ /allaḏīn/


‮التى‬‎ /allatī/

‮التى‬‎ /allātī/

‮الى، الاى‬‎ /allāy/

While Classical Arabic allows for two forms of the feminine plural relative pronoun, the form besides allātī is normally allawātī. Such a form does not occur in the Quran. Instead, a pronoun spelled variously in early manuscripts as ‮الاى‬‎ or ‮الى‬‎, presumably /allāy/, is used, with no discernable difference in function.89 Where the other pronominal forms are quite clearly the definite article al- + a particle la followed by a demonstrative element, the origin of the /-āy/ of /allāy/ is not entirely clear.

A.4.6 The Relative Possessive Demonstrative

The relative possessive demonstrative which created constructions like “those of X” inflect for case and gender. For the plural two competing stems occur, the /ulū~ī/ and /ḏawī/.





‮ذوا، ذى، ذا‬‎ /ḏū, ḏī, ḏā/

‮ذوا، ذوى‬‎ /ḏawā, ḏaway/

‮اولوا/اولا‬‎‮،‬‎90 ‮اولى‬‎ /ulū, ulī/

‮ذوي‬‎ /ḏawī/ (gen.) Q2:177


‮ذات‬‎ /ḏāt/

‮ذواتا، ذواتى‬‎ /ḏawātā, ḏawātay/

‮اولت‬‎ /ulāt/

A.4.7 Short Compound Interrogatives with mā

Prepositional compounds with occur several times in the Quran in short forms, where the interrogative is only written as a single mīm. All of these occur besides the long form. Whether the lack of an ʔalif should be understood as them ending in a short /ma/, or ending in /m/ cannot be deduced from the QCT, and is dependent on the relative chronology of these shortened forms in Quranic Arabic. It is worth noting that these shortened forms predominantly occur when the combination of preposition + is interrogative in function, only Q86:5 appears to have a relative function with the short ‮مم‬‎.

‮لم‬‎ /li-m(a)/ (Q3:183; Q4:77; Q5:18; Q7:164; Q9:43; Q19:42; Q20:125; Q27:46; Q41:21; Q61:2, 5; Q66:1)

‮لما‬‎ /li-mā/ (e.g. Q2:41)

‮فيم‬‎ /fī-m(a)/ (Q4:97; Q79:43)

‮فى ما‬‎ /fī mā/ (e.g. Q2:240)

‮بم‬‎ /bi-m(a)/ (Q15:54; Q27:35)

‮بما‬‎ /bi-mā/ (e.g. Q2:4)

‮مم‬‎ /mim-m(a)/ (Q86:5)

‮من ما‬‎ /min mā/ (e.g. Q30:28), ‮مما‬‎ /mim-mā/ (Q2:23)

‮عم‬‎ /ʕam-m(a)/ (Q78:1)

‮عن ما‬‎ /ʕan mā/ (e.g. Q7:166), ‮عما‬‎ /ʕan-mā/ (e.g. Q2:74)

‮كم‬‎ ‘how much?’, which in Classical Arabic is invariably read as kam, may also be considered the result of this historical shortening of -mā in compound interrogatives, with lexical specialization. Historically, it seems to derive from *ka-mah literally ‘like what?’, as can be seen in Semitic comparanda such as Hebrew kammå ‘how much?’ (with irregular gemination also found in låmmå ‘why?’) and Aramaic kəmā, kammā ‘how much?’ (Brockelmann 1908, 326). The fact that the form ends up as kam in the Classical language and not as kama might be an indication that the shortened pronoun was indeed pronounced /-m/ in Quranic Arabic, rather than /-ma/.

The semantic development of ka-mā ‘like what?’ → ka-ma ‘how much?’ also finds a parallel in another interrogative with the same meaning, namely ka-ʔayyin, likewise ‘like’ + ‘what?’, as attested in the Quran in the phrase ‮كاين من‬‎ ‘how much of!’ (Q3:146; Q12:105; Q22:45, 48; Q29:60; Q47:13; Q65:8), with fossilized nunation written out (see van Putten and Stokes 2018, 170). In Classical Arabic ka-ʔayyin can even have the interrogative function of kam ‘how much?’ (Lane, 134a) rather than only serving as an expression wonder.

A.4.8 Noun Inflection

Van Putten & Stokes (2018) have argued that Quranic Arabic had a reduced case system where only triptotic nouns distinguished the indefinite accusative with /-ā/ but otherwise lost inflect, except in construct. Case was retained in the Dual and Sound masculine plural. The paradigms of nouns can be reconstructed as follows:





‮كتب‬‎ /kitāb/ (Q2:89)

‮الكتب‬‎ /al-kitāb/ (Q2:2)

‮كتب‬‎ /kitābu/ (Q11:17)


‮كتب‬‎ /kitāb/ (Q20:52)

‮الكتب‬‎ /al-kitāb/ (Q2:85)

‮كتب‬‎ /kitābi/ (Q5:44)


‮كتبا‬‎ /kitābā/ (Q3:145)

‮الكتب‬‎ /al-kitāb/ (Q2:44)

‮كتب‬‎ /kitāba/ (Q4:24)






‮رجلان‬‎ /raǧulān/ (Q5:23)

‮الولدان‬‎ /al-wālidān/ (Q4:7)

‮رسولا‬‎ /rasūlā/ (Q20:47)


‮شهرين‬‎ /šahrayn/ (Q4:92)

‮الولدين‬‎ /al-wālidayn/ (Q4:135)

‮ابنى‬‎ /ibnay/ (Q5:27)


‮رجلين‬‎ /raǧulayn/ (Q16:76)

‮الذكرين‬‎ /aḏ-ḏakarayn/ (Q6:143)

‮ابويكم‬‎ /abaway-kum/ (Q7:27)






‮بنون‬‎ /banūn/ (Q26:88)

‮البنون‬‎ /al-banūn/ (Q18:46)

‮بنوا‬‎ /banū/ (10:90)


‮بنين‬‎ /banīn/ (Q17:6)

‮البنين‬‎ /al-banīn/ (Q37:153)

‮بني‬‎ /banī/ (Q17:4)

Sound masculine plural





‮مغٰنم‬‎ /maġānim/ (Q4:94)

‮القوٰعد‬‎ /al-qawāʕid/ (Q24:60)

‮مفٰتح‬‎ /mafātiḥu/ (Q6:59)91


‮مغٰنم‬‎ /maġānim/ (Q48:15)

‮القوٰعد‬‎ /al-qawāʕid/ (Q2:127)

‮مسٰكن‬‎ /masākini/ (Q14:45)


‮مغٰنم‬‎ /maġānim/ (Q48:19)

‮القوٰعد‬‎ /al-qawāʕid/ (Q16:26)

‮مسٰجد‬‎ /masāǧida/ (Q2:114)






‮رحمه‬‎ /raḥmah/ (Q2:157)

‮الرحمه‬‎ /ar-raḥmah/ (Q57:13)

‮رحمت‬‎ /raḥmatu/ (Q11:73)


‮رحمه‬‎ /raḥmah/ (Q2:159)

‮الرحمه‬‎ /ar-raḥmah/ (Q17:24)

‮رحمه‬‎ /raḥmati/ (Q15:56) ‮رحمت‬‎ /raḥmati/ (Q19:2)


‮رحمه‬‎ /raḥmah/ (Q3:8)

‮الرحمه‬‎ /ar-raḥmah/ (Q6:12)

‮رحمه‬‎ /raḥmata/ (Q39:9)

‮رحمت‬‎ /raḥmata/ (Q2:218)

Feminine singular





‮بينت‬‎ /bayyināt/ (Q3:97)

‮البينت‬‎ /al-bayyināt/ (Q2:209)

‮جنات‬‎ /ǧannātu/ (Q13:23)


‮بينت‬‎ /bayyināt/ (Q2:99)

‮البينت‬‎ /al-bayyināt/ (Q2:87)

‮جنات‬‎ /ǧannāti/ (Q5:65)

Sound Feminine plural

A.4.9 III-w and III-y Nouns with Preceding a Vowel.

Nouns that end in stem-final *-ay- and *-aw-, unlike Classical Arabic, appear to be distinct in Quranic Arabic, where the former collapsed to /ē/ and the latter to /ā/ (§ 5.8). The tables below give paradigm for both types of nouns.





‮هدى‬‎ /hudē/ (Q2:2)

‮الهدى‬‎ /al-hudē/ (Q2:120)

‮هديهم‬‎ /hudē-hum/ (Q2:272)


‮هدى‬‎ /hudē/ (Q2:5)

‮الهدى‬‎ /al-hudē/ (Q17:94)

‮هديهم‬‎ /hudē-hum/ (Q16:37)


‮هدى‬‎ /hudē/ (Q17:2)

‮الهدى‬‎ /al-hudē/ (Q20:47)

‮هدى الله‬‎ /hudē llāhi/ (Q6:71)

Words that end in /ā/ are rarer, and thus a full paradigm cannot be recovered.





‮عصاك‬‎ /ʕaṣā-k/ (Q7:117)


‮الصفا‬‎ /aṣ-ṣafā/ (Q2:158)

‮بعصاك‬‎ /bi-ʕaṣā-k/ (Q26:63)


‮عصاه‬‎ /ʕaṣā-h/(Q7:107)

A.4.10 III-w/y and III-ʔ Nouns

Final weak nouns whose stem ends in historical *-iy- such *wādiy- ‘valley, river’ have some amount of variation due to the appearance of shortened forms of the stem-final . The defective spelling of the definite form is especially common in pause, and seems to be the result of a process of pausal shortening of final ī that we find throughout the Quran (see A.3.6). The short spellings in construct are presumably simply context spellings of the shortening of the long vowel before the CC cluster of the following definite article.





‮قاض‬‎ /qāḍ/ (Q20:72)

‮الزانى‬‎ /az-zānī/ (Q24:2)

‮المهتد‬‎ /al-muhtad#/ (Q17:97)

‮اتى‬‎ /ʔātī/ (Q19:93)

‮لهاد‬‎ /la-hādi/ (Q22:54)


‮بواد‬‎ /bi-wād/ (Q14:37)

‮الداع‬‎ /ad-dāʕ/(Q2:186)

‮ٮهدى‬‎ /bi-hādī/ (Q27:81)92

‮ٮهد‬‎ /bi-hādi/ (Q30:53)


‮واديا‬‎ /wādiyā/ (Q9:121)

‮الداعى‬‎ /ad-dāʕī/ (Q20:108)

‮عليها‬‎ /ʕāliya-hā/ (Q11:82)

As in Classical Arabic, final weak plurals that are in origin diptotic have a slightly different form in the indefinite accusative form, lacking the final /-ā/. Here again we find shortened forms in the definite forms (besides long forms) although they do not occur in obvious pausal positions.





‮غواش‬‎ /ġawāš/ (Q7:41)

‮الجوار‬‎ /al-ǧawār/ (Q55:24)


‮ليال‬‎ /layāl/ (Q69:7)

‮المثانى‬‎ /al-maṯānī/ (Q15:87)

‮كالجواب‬‎ /ka-l-ǧawāb/ (Q34:13)

‮موليكم‬‎ /mawālī-kum/ (Q33:5)


‮مولى‬‎ /mawālī/ (Q4:33)

‮المولى‬‎ /al-mawālī/ (Q19:5)

Nouns which end in an original stem-final *-iʔ- are barely attested, but when they appear, they seem to behave identically to final weak nouns, although pausal forms with shortening are unattested.





‮البارى‬‎ /al-bārī/ (Q59:24)


‮باريكم‬‎ /bārī-kum/(Q2:54)


‮خاسيا‬‎ /xāsiyā/ (Q67:4)

‮شانيك‬‎ /šāniy(a)-k/(Q108:3)

One other noun that has a hamzah-final stem is ‮المنشيت‬‎ (Q55:24). This word is spelled in the CE as ‮المنشات‬‎, but this is clearly not original to the UT, as all early manuscripts retain the spelling ‮المنشيت‬‎ (see B.22). This word is read by the majority of the readers as a passive participle of ʔanšaʔa, i.e. munšaʔāt ‘(sails) raised’, whereas Ḥamzah reads it as an active participle munšiʔāt ‘raising (its sails)’ (Ibn al-Ǧazarī, § 4316). The rasm is only consistent with Ḥamzah’s reading, pointing to /munšiyāt/. If the majority reading is indeed intended, it means that the adjective munšaʔ has merged completely with III-y adjectives, and must be understood as coming from a paradigm */munšē/ */munšawn/; */munšēh/ /munšayāt/.

The noun which in Classical Arabic would be sayyiʔ is consistently spelled ‮السيا‬‎ in early Quranic manuscripts (van Putten 2018, 115). This is similar to verbs ending in the same sequence: ‮هيا‬‎ hayyiʔ (Q18:10) and ‮يهيا‬‎ yuhayyiʔ (Q18:16). The reasons for this are unknown. It is tempting to see this as a historical hamzah spelling.

A.4.11 Nouns in *-āʔ in Construct

In the discussion of the ʔalif al-wiqāyah above, we already saw that nouns ending in -āʔ in the construct nominative sometimes are spelled not with final ʔalif, as is the normal spelling, but rather with ʔalif+wāw (most notably with ǧazāʔ- spelled as ‮جزاو‬‎) and one time as wāw+ʔalif, ‮ابنوا‬‎ (see B.11 and B.14). Also, the genitive is occasionally expressed with a glide yāʔ in construct. This seems to be reconstructible for the following words in the UT: ‮تلقٰى‬‎ ‘the accord of’ (Q10:15), ‮اناى‬‎ ‘the hours of’ (Q20:130) and perhaps als ‮ايتاى‬‎ ‘the giving of’ (Q16:90) (see B.23).

When nouns of this type are followed by a pronominal clitic, they always reflect the case vowel with wāw in the nominative and yāʔ in the genitive in the CE. But this is a quirk of the CE, and examination of early Quranic manuscripts reveals that both spellings with and without the glides are attested (van Putten and Stokes 2018, 172–176). While previously, Van Putten & Stokes (2018, 159, 160 f.) have interpreted this as evidence that case vowels in construct could optionally be lost, I now believe that a more natural interpretation of this data is to see this as related to the special status of this word-final hamzah after /ā/.

From Quranic rhyme it is clear that the hamzah was retained in this position, thus ‮الدعا‬‎ (Q3:38) clearly rhymes with other words that end in /āG/, which suggests a pronunciation /ad-duʕāʔ/. Moreover, ‮انشا‬‎ (Q56:35) stands in an /āGā/ rhyme, thus suggesting that the indefinite accusative was pronounced with final /āʔ-ā/, i.e. /inšāʔā/.

Presumably those forms that lack the glides are cases where the stem-final hamzah (spelled with the ʔalif) was retained. While those that show a glide have optional elision of the hamzah in this non-word-final position. The paradigm of nouns of this type must therefore be something along these lines as shown in the table below.






‮فجزا‬‎ (Q5:95)


‮السما‬‎ (Q25:25)


‮جزاو‬‎ (Q5:29)


‮جزا‬‎ (Q2:85)


‮جزاوهم‬‎ (Q17:98)


‮فجزاه‬‎ (Q4:93)



‮جزا‬‎ (Q5:38)


‮الجزا‬‎ (Q53:41)


‮دعا‬‎ (Q24:63)

/duʕāʔa/, /duʕā(.a)/

‮دعاكم‬‎ (Q35:14)




‮دعا‬‎ (Q41:51)


‮الدعا‬‎ (Q3:38)


‮اناى‬‎ (Q20:13)


‮دعا‬‎ (Q41:49)


‮دعايهم‬‎ (Q46:5)


‮بدعاك‬‎ (Q19:4)


The noun ʔawliyāʔ ‘allies; protectors’ is of exceptional status. While it is a noun that historically end in -āʔ, when the noun stands in construct the glide for the case vowel never appears, not when it stands in construct with a noun, nor when a pronominal suffix follows. This idiosyncrasy is not retained in the CE, but can be reconstructed for the UT, see B.24 (see also Nöldeke et al. 2013, 422). It thus seems that this noun has merged with nouns that end in -yā such as ‮الدنيا‬‎ /ad-dunyā/ ‘world’ (passim), ‮الحوايا‬‎ /al-ḥawāyā/ ‘intestines’ (Q6:146) and ‮خطيكم‬‎ /xaṭāyā-kum/ ‘your sins’ (Q2:58).






‮اوليا‬‎ (Q46:32)


‮اوليا‬‎ (Q5:51)





‮اوليا‬‎ (Q3:28)


‮اوليا‬‎ (Q4:76)


‮اوليٰه‬‎ (Q3:175)



‮اوليا‬‎ (Q11:20)


‮اوليٰهم‬‎ (Q6:121)


This shift of category seems to be unique to this noun, ‮ادعيٰيهم‬‎/ʔadʕiyāyi-hum/ ‘adopted sons’ (Q33:37) is consistently spelled with the glide for the genitive in early manuscripts (see B.25).

A.4.12 Confusion between Subjunctive and Apocopate

There is one example in the QCT where we find confusion between the subjunctive and the apocopate. The following verse uses an apocopate stem, in a clearly subjunctive context:

‮رب لولا اخرتنى الى اجل قريب فاصدق واكن من الصلحين‬‎ (Q63:10)

/rabb-∅, lawlā axxarta-nī ilā ajal qarīb fa-aṣṣaddaq wa-akun min aṣ-ṣāliḥīn/

My lord, if only you would delay me for a brief term so I would give charity and be among the righteous93

A.4.13 Partial Merger of III-ʔ Verbs and III-y/w Verbs

In Classical Arabic grammar III-w/y verbs and III-ʔ are kept clearly distinct. This is, as far as we can tell from the defective spelling, not the case in Quranic Arabic, where we see a certain amount of merger of the two stem types. This merger is certainly less complete than it is in the modern dialects, but nevertheless we can deduce mergers from the QCT that did not take place in Classical Arabic.

G-stems of III-ʔ verbs are still clearly distinct from III-y and III-w verbs, e.g. ‮قرات‬‎ /qarāt/ ‘you recited’ (Q16:98) vs. ‮نجوت‬‎ /naǧawt/ ‘you fled’ (Q28:25) and ‮قضيت‬‎ /qaḍayt/ ‘you decided’ (Q4:65), and even in derived stems there are clear examples where they are distinct, e.g. ‮نبات‬‎ /nabbāt/, or /nabbaʔat/ ‘she informed’ (Q66:3), ‮نباتكما‬‎ /nabbātu-kumā/ ‘I informed you’ (Q12:37); ‮اخطاتم‬‎ /axṭātum/ ‘you have sinned’ (Q33:5); ‮امتلات‬‎ /imtalāt/ ‘you filled’ (Q50:30).

In the imperfect stem and nominal derivations, however, these verbs merge to a large extent throughout the whole paradigm. With the loss of the ʔ, word-final yielded , merging in most places with word-final of final weak roots. This can be clearly seen in some of the derived stems of final glottal stop roots that in the imperfect plural forms as well as the participial plural forms have merged with the III-y/w verbs.

‮يستهزى‬‎ /yastahzī/ (Q2:15) < *yastahziʔu
‮مستهزون‬‎ /mustahzūn/ (Q2:14) < *mustahziʔūna
‮يستهزون‬‎ /yastahzūn/ (Q6:5, etc.) < *yastahziʔūna
‮تستهزون‬‎ /tastahzūn/ (Q9:65) < *tastahziʔūna
‮والصبون‬‎ /wa-ṣ-ṣābūn/ (Q5:69) < *aṣ-ṣābiʔūna
‮والصبين‬‎ /wa-ṣ-ṣābīn/ (Q2:62; Q22:17) < *aṣ-ṣābiʔīna
‮اتنبون‬‎ /ʔa-tunabbūn/ (Q10:18) < tunabbiʔūna
‮فمالون‬‎ /fa-mālūn/ (Q37:66; Q56:53) < *fa-māliʔūna
‮يطفوا‬‎ /yuṭfū/ (Q9:32; Q61:8) < *yuṭfiʔū
‮ليواطوا‬‎ /li-yuwāṭū/ (Q9:37) < *yuwāṭiʔū
‮الخطون‬‎ /al-xāṭūn/ (Q69:37) < *al-xāṭiʔūna
‮(لـ)ـخٰطين‬‎ /(la-)xāṭīn/ (Q12:29, 91, 97; Q28:8) < *xāṭiʔīna
‮متكين‬‎ /muttakīn/ (Q18:31, etc.) < *muttakiʔīna
‮خاسين‬‎ /xāsīn/ (Q2:65; Q7:166) < *xāsiʔīna

This merger has led to some amount of disagreement whether certain verbs are III-y or III-ʔ among the canonical readers, see § 6.5.5 for a discussion.

Words ending in *aʔūna are technically ambiguous in terms of their interpretation, due to the tendency to not write double wāw sequences for representing /wū/ (see A.2.2). It however stands to reason that these would have merged to /-awn/, e.g.

‮يطون‬‎ /yaṭawn/ (Q9:120) < *yaṭaʔūna
‮يقرون‬‎ /yaqrawn/ (Q10:94; Q17:71) < *yaqraʔūna
‮يدرون‬‎ /yadrawn/ (Q13:22; Q28:54) < *yadraʔūna
‮مبرون‬‎ /mubarraʔūn/ (Q24:26) < *mubarraʔūna
‮مرجون‬‎ /murǧawn/ (Q9:106) < *murǧaʔūna

In the ʕarabiyyah, the apocopate and imperative would be places where III-ʔ and III-w/y verbs would remain distinct, even if one were to pronounce them with the loss of hamzah. The imperative of ‮صلى‬‎ ṣallā ‘to bless’ would be ‮صل‬‎ ṣalli ‘bless!’, whereas the imperative of ‮نبأ‬‎ nabbaʔa ‘to inform’ would be ‮نبئ‬‎ nabbiʔ which with dropping of the hamzah should yield nabbī.

In the QCT we see that a merger between the two stem types is under way, no doubt due to their complete merger in the imperfective and subjunctive stems. The table below illustrates the examples of apocopates and imperatives of historically III-ʔ verbs and how they appear in the QCT.


Classical Arabic

‮ارجه‬‎ (Q7:111; Q26:36)



‮نبنا‬‎ (Q12:36)94



‮نبى‬‎ (Q15:49)



‮نبيهم‬‎ (Q15:51; Q54:28)



‮انبيهم‬‎ (Q2:33)95



One final verb could perhaps be added here, namely ‮ننسها‬‎ (Q2:106), which is either read nunsi-hā, an apocopate of ʔansā ‘to cause to be forgotten’ or nansaʔ-hā from nasaʔa ‘to cause to be delayed’ (Ibn al-Ǧazarī, § 2720). If the latter reading is correct, this would be yet another hamzated apocopate that appears to function as a final weak verb. But *aʔ usually does not show this merger, e.g. ‮لم ينبا‬‎ /lam yunabbā/ ‘he was not informed’ (Q53:36), ‮اقرا‬‎ /iqrā/ ‘recite!’ (Q17:14; Q96:1, 3). Semantically, nunsi-hā seems like a better fit in this verse: ‮ما ننسخ من ايه او ننسها نات بخير منها او مثلها‬‎ “whatever we abrogate from a verse or cause it to be forgotten we bring one better or equal to it.”, and thus I think it is better taken as the regular outcome of a final weak verb.

Finally, the verb hayyaʔa ‘to make ready’ is consistently spelled with a final ʔalif in early Quranic manuscripts: hayyiʔ ‮هيا‬‎ (Q18:10); yuhayyiʔ ‮يهيا‬‎ (Q18:16). This spelling should be reconstructed for the Uthmanic archetype, but its interpretation is not very clear, for a suggestion and other words with such spellings, see Van Putten (2018, 115).

A.4.14 Pausal Imperatives/Apocopates of III-y/w Verbs Iqtadih, yatasannah

III-w/y apocopates and imperatives throughout the Quran are consistently without any reflex of the final radical, thus we see, e.g. ‮يرم‬‎ /yarmi/ ‘throws’ (Q4:112), ‮يدع‬‎ /yadʕ(u)/ ‘invokes’ (Q23:117), ‮يلق‬‎ /yalq(a)/ ‘meets’ (Q25:68); ‮ايت‬‎ /īt(i)/ ‘come!’ (Q10:15), ‮ادع‬‎ /udʕ(u)/ ‘invoke!’ (Q2:68).

However, the only two times that an imperative and apocopate occur in pause, these stems are suffixed with a final hāʔ: ‮فبهديهم اقتده‬‎ /fa-bi-hudē-hum iqtadih#/ “so follow after their guidance.” (Q6:90),96 which is followed by the ‮قلے‬‎ pausal sign in the CE, which indicates an optional pause, with a preference towards pausing.97 The other case is found in ‮قل بل لبثت مايه عام فانظر الى طعامك وشرابك لم يتسنه‬‎ /qāl bal labiṯt miyah ʕām fa-nẓur ilā ṭaʕāmi-k wa-šarābi-k lam yatsannah#/ “He said: Nay, you have remained for a hundred years, look at your food and your drink; it did not age.” (Q2:259), which is followed by the ‮صلے‬‎ pausal sign in the CE, which indicates an optional pause, with a preference towards continuing.98 Based on these two examples it seems that in Quranic Arabic imperatives and apocopates received /h/ in pause.

It is worth noting that the fact that this hāʔ only shows up in pausal position, is yet another piece of evidence that ‘pausal spelling’ is not a governing principle in Quranic orthography. Had that been the case, all apocopates and imperatives should have received a final h, not just the one that stand in a pausal position.

A.4.15 Partial Merger of the I-ʔ and I-w Verbs in Derived Stems

Due to the loss of the hamzah (see § 5.2) D- and L-stems of verbs with a ʔ as their initial consonant merge with D- and L-stems of verbs with w as their initial consonant, e.g. *yuʔaxxiru-hum > ‮يوخرهم‬‎ /yuwaxxiru-hum/ ‘he gives respite to them’ (Q14:42); *yuʔāxiḏu > ‮يواخذ‬‎ /yuwāxiḏ/ ‘he would punish’ (Q35:45). Such verbs usually remain distinct in the perfect where you get forms like *ʔaxxara > ‮اخر‬‎ /ʔaxxar/ ‘left behind’ (Q75:13). The partial merger of these verb types is no doubt the origin of the pseudocorrect use of hamzah in muʔṣadah for mūṣadah (§ 6.4.2).

A more pervasive merged with I-w is found in the Gt-stem of the verb ʔaxaḏa, which is treated as a I-w in the QCT. This idiosyncrasy also finds its way into Classical Arabic, e.g. ‮اتخذ‬‎ /ittaxaḏ/ ‘he took’ (e.g. Q18:4). Other Gt stems of I-ʔ verbs continue to behave distinctly from I-w verbs, e.g. ‮لا ياتل‬‎ /lā yātal/ ‘may they not swear’ < *yaʔtali (Q24:22).

A.4.16 /yak/ besides /yakun/

The verb ‮كان‬‎ /kān/ ‘to be’, has an anomalous form in the apocopate. Besides the regular stem form ‮ـكن‬‎ which is identical to that of Classical Arabic, quite often we find the form ‮ـك‬‎. Van Putten & Stokes (2018, 168–170) argue that this is best understood as the regular outcome of this verb in Quranic Arabic. As word-final nunation and case vowels were lost, the word final *-un of *yakun would also regularly be lost, yielding /yak/. The long form is then an analogically restored version of the apocopate.

A.4.17 *raʔaya ‘to See’ and *naʔaya ‘to Be Distant’ as ‮را‬‎ and ‮نا‬‎

The regular spelling of the verbs raʔā ‘to see’ and naʔā ‘to be distant’, both historically final weak verbs with a medial hamzah, is ‮را‬‎ and ‮نا‬‎ respectively in the QCT. Their orthographic behaviour suggests that they have merged with hollow roots with a final hamzah, e.g. ‮جا‬‎ /ǧāʔ/, ‮جات‬‎ /ǧāt/ ‮جاو‬‎ /ǧāw/ ‘to come’, at least in the 3rd person masculine singular and plural forms, hence we find spellings ‮را‬‎ /rāʔ/ ‮راو‬‎ /rāw/. The spelling ‮را‬‎ occurs twenty times in the Quran, and only Sūrat al-Naǧm attests the form ‮راى‬‎ (Q53:11, 18), which at least in the first verse seems to be the use of a dialectal form /raʔē/ to accommodate the rhyme. The exact interpretation of the unusual behaviour in this Sūrah, however, should not distract us from the fact that the regular Quranic form is ‮را‬‎, which is not likely to have been a spelling for /raʔē/.

How exactly ‮را‬‎ and ‮نا‬‎ took on the shapes that they have is not entirely obvious. One might imagine that at an earlier stage of Quranic Arabic, the *y and were regularly metathesized, *raʔaya > *rayaʔa which then regularly yielded /rāʔ/. Alternatively, one might imagine that the intervocalic hamzah had dropped yielding *raʔaya > rāya which then, similar to *samāy ‘sky’ shifted its word final y to ʔ, likewise yielding /rāʔ/.

In the former development one would expect the verb to have completely merged with verbs of the type ‮جا‬‎ /ǧāʔ/, in which case one would predict the first and second person forms to be like ‮جيت‬‎ /ǧīt/. But this does not seem to be the case. The Cairo edition attests both ‮اريت‬‎ ‘did you see?’ (e.g. Q18:63) and ‮رايت‬‎ ‘you saw’ (e.g. Q47:20).99 In Early Quranic manuscripts it is not at all uncommon to only see the spelling ‮ريت‬‎, but ‮رايت‬‎ spellings do occur. Considering these spellings, it seems that the suffixed forms were probably /rāy-t/ ‘you saw’, etc. In which case the second scenario which requires hamzah to be lost before the *āy > āʔ shift, becomes more probable. This specific behaviour with partial merger, rāʔa but raʔaytu (or rāytu), is exactly what is reported by al-Farrāʔ as being a typical Hijazi isogloss (§ 5.11).


This same feature is well-attested in early Islamic Arabic, and generally recognized to be part of Pre-Classical orthography (Blau 1967, § 9.1; 2002, 35, § 26; Hopkins 1984, § 10).


A lack of awareness of the special status of qāla has led to some confusion in epigraphic research. The extremely common formula ‮اللهم اغفر […] لمن قل/قال امين‬‎ ‘O God, forgive […] whoever says Amen’, is misread by Grohmann (1962, 148–149; Z 256, Z 257) as ‮اللهم اغفر […] لمن فا امين‬‎ ‘O God, grant pardon […] to everyone who returns, Amen’. cf. the same formula with ‮قال‬‎ (al-Kilābī 2009, nos. 78, 215) and with ‮قل‬‎ (al-Kilābī 2009, nos. 49, 90). A similar misunderstanding is found in the edition of the 31 AH gravestone inscription from Aswan, where line 4–5 ‮استغفر له اذا قرا هذا الكتب وقل امين‬‎ should be understood as “and ask (Allah) pardon for him (the deceased) when he reads this writing and says Amen”, and not how it is translated “(passer by) When reading this inscription ask pardon for him (the deceased) and say Amen!” (El-Hawary 1930, 322).


In all of these cases, the choice of spelling these defectively seems to be an attempt to accommodate the other canonical readings, which in these places disagree on the reading of this word. Some of them thus read it as qul (Ibn al-Ǧazarī, § 3640, § 3705, § 3706, § 4154).


The common defective spelling of the feminine plural ending also occurs in early Islamic inscriptions, but is misunderstood by Grohmann (1962, Z 48), who interprets ‮صلوت الله‬‎ as a singular ‘the blessing of God’ rather than ‘the blessings of God’, cf. ‮صلوات الله‬‎ (al-Rāšid 2009, 242). Likewise, Grohmann translates ‮رحمت الله وبركته عليك‬‎ ‘the Mercy and blessing of God may be upon you’ taking ‮بركته‬‎ as a singular (Grohmann 1962, Z 150, Z 171), but this formula certainly has the plural /barakāt-uh/, cf. ‮رحمت الله وبركاته عليكم‬‎ (Grohmann 1962, Z 225).


This practice is also attested in early Islamic Papyri (Hopkins 1984, § 10d), in the Ibn Wahb literary papyrus (Blau 1999, 124). Several clear cases are found in early Islamic inscriptions as well, e.g. ‮يرحمن‬‎ (al-Kilābī 2009, no. 35), ‮يرب‬‎ ‘O my lord’ (Grohmann 1962, nos. 165, 232).


The origin of this innovation in the Cairo Edition appears to come from Al-Dānī’s al-Muqniʕ who mentions that the Qurans of Medina and Iraq spell these words with only one yāʔ, a practice copied by the Cairo Edition (Al-Dānī al-Muqniʕ, 56).


Among the non-canonical readings there are moreover reports of bayʔisin, bīsin, baysin and biʔīs (Ibn Xālawayh muxtaṣar, 47).


One might also consider the reading /bāyis/ < *bāʔisin, which would be in line with the orthography ‮بايس‬‎ attested in BnF Arabe 6140a, although this could also be analysed as a case of historical hamzah spelling see A.2.7.


Ibn al-Ǧazarī (§ 1531).


Ibn al-Ǧazarī (§ 2493).


Ibn al-Ǧazarī (§ 2519). Warā-ya is attributed to Ibn Kaṯīr in a non-canonical transmission (Ibn Muǧahid, 407).


Ibn al-Ǧazarī (§ 2714).


Ibn al-Ǧazarī (§ 2715).


The only exception to this contraction is ‮عليون‬‎ /ʕilliyyūn/, ‮عليين‬‎ /ʕilliyyīn/ ‘Elyon’. As this is likely a loanword from Hebrew ʿɛlyon ‘upper part of something; epithet of God’ (Jeffery 2007, 215–216), it should not surprise us that this contraction does not take place, as it may have been borrowed at a time postdating the contraction.


I thank Ahmad Al-Jallad for suggesting this analysis to me.


‮تلوا‬‎ ‘you distort’ (Q4:135) is also read as talū by Ḥamzah and Ibn ʕāmir (Ibn al-Ǧazarī § 2962), so may not represent an example of this. The interpretation of the reading talū seems somewhat controversial. Al-Farrāʔ (Maʕānī I, 291) derives it from a root lʔy, which he claims has the same meaning as tatawallaw ‘they follow in succession’. Ibn Xālawayh (Ḥuǧǧah, 127) see it as a G-stem of the root wly.


There is significant disagreement among the readers whether to read these words with a question particle in front of both, for a discussed see Ibn al-Ǧazarī (§ 1413).


Thus, Quranic orthography is unable to make the distinction between the homophonous yadʕū ‘he calls’ and yadʕū ‘they call (subjunctive/jussive)’ which in Classical orthography is expressed as ‮يدعو‬‎ versus ‮يدعوا‬‎.


Ibn al-Ǧazarī (§ 4397) reports both variants lawaw (Nāfiʕ and Rawḥ) and lawwaw (the rest).


Ibn al-Ǧazarī (§ 948, § 950, § 1234, § 1343) makes an explicit distinction between ḥurūf al-madd (ū, ī, ā) and ḥarfay al-līn (aw, ay). This does not appear to be a distinction systematically made by the early grammarians like Sībawayh, which seems to use the terms indiscriminately, and often uses the compound term hurūf al-madd wa-l-līn. Even if it were an ancient distinction, the two terms are still clearly distinguished from uses of wāw and yāʔ were a consonant, rather than a vowel, precedes.


It is worth appreciating how the QCT aptly distinguishes this word from ‮فاوا‬‎ /fāwū/ < *fa-ʔwū ‘so retreat!’ (Q18:16), which would have been homographic had the Classical Arabic rule of the ʔalif al-wiqāyah been adhered to.


The Quranic Arabic perfect of ‘to see’ was /rāʔ/, not /raʔā/, see § 5.11 for a discussion.


The genitive seems to show similar free variation, but there is only evidence for it in construct e.g. ‮نباى المرسلين‬‎ (Q6:34) but ‮من نبا موسى‬‎ (Q28:3).


Some Arab Grammarians appear to have argued that unlike *CaCaw- stems, like ‮عصا‬‎ “stick”, originally *CiCaw- and *CuCaw- stems shifted their final root consonant to y, something that also happened in Quranic Arabic ‮الضحى‬‎ /aḍ-ḍuḥē/ ‘the forenoon’ (Q93:1) and ‮العلى‬‎ /al-ʕulē/ ‘highest (plural)’ (Q20:4). For a brief discussion see Ibn al-Ǧazarī (§ 1974).


Blau (1967, § 8.2) gives several clear examples of this same orthographic feature in early Christian Arabic, e.g. ‮لااقرن‬‎ ‘I shall admit’, ‮لااعدلن‬‎ ‘I shall return’, ‮لاابيعنك للبربر‬‎ ‘I swear I shall sell you to the Berbers’, ‮لااصنعها‬‎ ‘I shall do it’, etc.


The extra ʔalif has been removed.


The extra ʔalif has been removed.


The extra ʔalif has been added by a later hand.


The extra ʔalif has been removed.


The extra ʔalif has been added by a later hand.


The extra ʔalif has been added by a later hand.


The extra ʔalif has faded, and was perhaps removed on purpose.


The extra ʔalif has been removed.


Surprisingly, this reading is not mentioned by Ibn Muǧāhid (121), despite Qunbul being his direct teacher.


Most works mention explicitly that Q75:2 ‮ولا اقسم بالنفس اللوامه‬‎ is read as wa-lā ʔuqsimu bi-n-nafsi l-lawāmah ‘And nay! I swear by the reproaching soul’ even by Qunbul, although here too a reading wa-la-ʔuqsimu seems more natural.


The ʔalif has been removed.


The ʔalif has been removed.


Also, the preposition ‮الى‬‎ is read as /ʔilā/ rather /ʔilē/, but this word is not commonly spelled ‮الا‬‎ (but see the corrected spelling of ‮الا‬‎ to ‮الى‬‎ in Q46:5 in CA1).


Such spellings also occur in the early papyri (Hopkins 1984, § 10d, only mentioning ‮حتا‬‎) and Christian Arabic (Blau 1967, § 10.1). Considering the special position of these prepositions in the reading traditions and the grammarians, the spelling of these prepositions cannot be used as evidence that the vowel /ē/ and /ā/ have merged (pace Hopkins 1984, § 12c; Blau 2002, § 16).


On the etymology of ḥattā, see Al-Jallad (2017b).


The spelling ‮شاى‬‎ is also well-attested in the early Islamic Papyri (Hopkins 1984, § 15d).


Diem (1980, § 127) explores the possibility that this might in fact represent /yāyas/, the outcome of the metathesized root ʔayisa ‘to despair’ as attested in several modern dialects, as well as in the Classical Arabic lexicons. He suggests this is not likely, as the perfect form does not point to this metathesis. I tentatively follow this conclusion, although it could be that > ʔy was a regular metathesis, which eventually gave rise to the perfect stem being analogically remodeled towards ʔayisa. This other reading with metathesis is possible, and is in fact attested among the reading traditions, al-Bazzī ʕan Ibn Kaṯīr reads istāyasū (Q12:80), tāyasū, yāyasu (Q12:87), istāyasa (Q12:110), yāyas (Q13:31), (Ibn al-Ǧazarī § 1528). The metathesized perfect form of this verb is attested in the early Islamic papyri (Hopkins 1984, § 56).


It is possible that the original reading of this word was rather /bāyis/, something both ‮بايس‬‎ and the more generally attested ‮بيس‬‎ also supports as a reading. There does not seem to be a significantly difference in meaning between bāʔis and baʔīs.


This spelling convention on word-boundaries is not reported on for non-Quranic early Islamic Arabic, but it is at least found on the Dome of the Rock inscription which spells as ‮باييات‬‎ (dotted as such!) (Kessler 1970, 6).


Cf. the Muʕtazilī ʔaḥmad b. ʔabī Duʔād (d. 240 AH), see EI2: s.v. Aḥmad b. Abī Duʾād. Of course, we may also entertain the idea that this is a later spelling pronunciation and in fact Ibn ʔabī Dāʔūd was intended.


Which in turn could, in fact, come from *dāwūd again, due to the presence of a shift of *wu and to ʔu/ʔū well-attested in Classical Arabic, and also found in the QCT once (see A.3.11).


Puin (2011, 150) identified several early Quranic manuscripts where words of this type are occasionally spelled without the ‮و‬‎, e.g. ‮فاليك‬‎ /fa-ulāyik/, ‮الى‬‎ /ulī/.


Cf. early Christian Arabic with the same practice (Blau 1967, § 9.2).


See Van Putten (forthcoming) for a discussion on the spelling of this word and other cases of ʔalif maqṣūrah followed by the clitic -ni/nī.


Read by Ḥamzah as tahdi l-ʿumya (Ibn al-Ǧazarī: § 3896).


المنشات in the CE, but this is an idiosyncrasy of this edition. See A.4.10 for a discussion on the spelling of this adjective.


On the topic of the hamzah spelling see also Diem (1976; 1980, § 116–128).


See A.3.9.


This is the reading of Ibn Šihāb al-Zuhrī (Ibn Xālawayh muxtaṣar, 47). Some other possible interpretations of this rasm, e.g. /bayyis/ seems possible too. For a discussion see A.2.2.


The Cairo Editions contain a few exceptions to this orthographic practice. For example, iǧtabā-hu (Q16:121) is spelled ‮اجتبه‬‎; Early Quranic manuscripts, however consistently spell this ‮اجتبيه‬‎ (e.g. B, W, BL, SM). The same is true for the same word in Q68:50 (e.g. W, SM). Likewise, ʕuqbā-hā (Q91:15) spelled ‮عقبها‬‎ in the Cairo Edition, is simply found as ‮عقبيها‬‎ in early manuscripts (e.g. SM, G) (see B.27). However, while ‮مضى‬‎ /maḍē/ ‘departed’ (Q43:8) has the expected spelling in the CE, early Quranic manuscripts surprisingly seem to converge on the spelling ‮مضا‬‎ (see B.28).


The form in Q76:16 does not occur in rhyme, but is the first word of the verse, directly following the previous word spelled like this. This being said, the later Basran codices seem to change this spelling to the expected ‮قوارير‬‎. For a discussion on the reports on this spelling and its attestations in early manuscripts see Sidky (2021).


For a further discussion on these rhymes see also Van Putten & Stokes (2018, 161–163).


Ibn al-Ǧazarī (§ 3037).


Ibn al-Ǧazarī (§ 4091).


Ibn al-Ǧazarī (§ 1209).


Ibn al-Ǧazarī (§ 1208).


Ibn al-Ǧazarī (§ 2806).


Ibn al-Ǧazarī (§ 3801).


Ibn al-Ǧazarī (§ 2989).


Ibn al-Ǧazarī (§ 2774).


Ibn al-Ǧazarī (§ 2774).


Ibn al-Ǧazarī (§ 3540).


The Manuscript Ma VI 165 has ‮تستطع‬‎ for both Q18:78 and Q18:82.


Ibn al-Ǧazarī (§ 3256).


Ibn al-Ǧazarī (§ 2969).


Ibn al-Ǧazarī (§ 4010).


This spelling also appears to be common in early Christian Arabic (Blau 1967, §


Ibn al-Ǧazarī (§ 3354; § 3633), who considered Q21:88 a hapological reduction of nunaǧǧī. Ibn Muǧāhid (430), surprisingly considers it to be the passive perfect nuǧǧiya with dropped final -a. This is grammatically quite problematic considering the following noun al-muʔminīna is in the accusative.


In poetry, the ʔalif al-waṣl may sometimes be treated as a true hamzah (Nöldeke 1896, 7).


Except by the Syrian canonical reader Ibn ʕāmir who reads it wa-la-dāru l-ʔāxirati, because the Syrian Muṣḥaf spells this ‮ولدار الاخره‬‎ rather than ‮وللدار الاخره‬‎ (Ibn al-Ǧazarī § 3017; Cook 2004, 92, (S4)).


As is the recitation of Warš ʕan Nāfiʕ, ʔabū Ǧaʕfar and optionally for ʔabū ʕamr.


This is thus one of the many examples where the alleged “pausal spelling principle” is violated in Quranic orthography. See Van Putten & Stokes (2018, 152–158) for a more detailed discussion.


ʔabū ʕamr reads wuqqitat and ʔabū Ǧaʕfar reads this wuqitat, ignoring the dropping of the hamzah suggested by the rasm (Ibn al-Ǧazarī § 4494).


This phenomenon is also attested occasionally in early Christian Arabic. Blau (1967, § 83) reports ‮اُ جد‬‎ ‘was found’, ‮اُلد‬‎ ‘was born’, ‮اُقفت‬‎ ‘she was placed’ and ‮اُعظت‬‎ ‘you have been instructed’.


Al-Farrāʔ (Luġāt, 29) reports that Banū ʔasad uses and for huwa and hiya, and he cites poetry using the form. Such monosyllabic forms of the independent pronouns occur on occasion in poetry.


Indeed, some non-canonical readers would read it as such, see Ibn Xālawayh (muxtaṣar, 32).


This, incidentally, seems to suggest that in the original prosody of Quranic recitation, such epenthetic vocatives had a minor pause following them, explaining the pausal form.


See A.2.3 for the discussion of the use of the ʔalif al-wiqāyah and § 5.11 on the Quranic Arabic use of /rāʔ/ and /nāʔ/ instead of Classical raʔā and naʔā.


The plene spelling of this pronoun seems to be the common spelling in early Quranic manuscripts (see B.29).


Al-Mubarrad (III, 275) discusses the full system of addressee agreement.


This word is normally interpreted as a single word māḏā and written as such in typewritten Arabic. There is no way to distinguish ‮ما ذا‬‎ from ‮ماذا‬‎ in handwritten Arabic, as a space between unconnected letters is of the same size in between words as within it. In light of ‮من ذا‬‎ above, it seems best to interpret the form as /mā ḏā/ in Quranic Arabic. The ambiguity whether these phrases should be seen as one word or not seems to also underlie the reports that the Muṣḥafs of Ibn Masʕūd would write man ḏā as a single word ‮منذا‬‎ (Al-Farrāʔ Maʕānī, III, 132).


Sībawayh (II, 416–419) specifically discusses constructions of this type.


This is a spelling practice it shares with early Christian Arabic (Blau 1967, § 26.3.2).


It is tempting to see in ‮الى‬‎ the ubiquitous relative pronouns illi of the modern dialects, but the spelling ‮الاى‬‎ seems to preclude such an interpretation. It is, moreover, unclear how a pronoun as rare as the feminine plural relative pronoun would be likely to spread to all positions and become the dominant relative pronoun.


When ulū stands before a CC cluster, early Quranic manuscripts frequently write the demonstrative as ‮اولا‬‎. The reasons for this are not entirely clear. For a rather speculative account on this phenomenon see Puin (2011, 154). See also Sidky (2021) for a discussion of this phenomenon, but also lacking a solution. A dedicated study of this orthographic phenomenon is warranted.


The use of the plural pattern CaCāCiC for ‘keys’ is somewhat surprising. Strict Classical Arabic grammar would require the plural of miftāḥ to be mafātīḥ. The use of this pattern for stems with a long vowel in the last syllable seems to be more common in early Islamic Arabic papyri (Hopkins 1984, § 87b). Generalization of CaCāCiC over CaCāCīC is also a typical isogloss of the modern Maghrebi Arabic dialects (Fischer and Jastrow 1980, 91). The Lisān al-ʕarab (Lisān, 3337c) explains this unusual plural as corresponding to a singular *miftaḥ rather than miftāḥ, but the only evidence cited for it is the present Quranic verse, which seems to confirm it exceptional status. Note that Ibn Xālawayh (muxtaṣar, 35) cites a non-canonical reading for this verse with the singular, which would be equally acceptable to the rasm.


Q27:81 and Q30:53 are read by hamzah as tahdi l-ʿumya (Ibn al-Ǧazarī: § 3825).


It is interesting to note here that, while most reading traditions simply follow the rasm and read this word as an apocopate ʔakun, ʔabū ʕamr ignores the rasm and reads it as the Classically normative ʔakūna (Ibn al-Ǧazarī, § 4401).


This word is spelled ‮نبينا‬‎ in the CE, but in early Quranic manuscript ‮نبنا‬‎ is regular. See B.26 for an overview.


Most manuscripts have the rasm ‮انبيهم‬‎ but DAM 01–32.1 has ‮انبهم‬‎. This latter rasm is not common, but it is consistent with the reading of al-Ḥasan al-Baṣrī ʔanbi-himī (Ibn Xālawayh muxtaṣar, 4).


Ibn ʕāmir treats this final hāʔ as a pronoun, reading it iqtadi-hi or iqtadi-hī (Ibn al-Ǧazarī, § 2375). This reading is grammatically rather awkward. It is difficult to take it as a resumptive pronoun of the preceding object (bi-hudā-hum) since that object is marked with bi-, thus we would expect iqtadi bi-hī rather than iqtadi-hī̆. Ibn Muǧāhid (262) shared this sentiment and explicitly calls it a mistake (wa-hāḏā ġalaṭun) because this is a pausal hāʔ, not a pronoun.


See also Saǧāwindī (ʕilal al-Wuqūf, 333).


See also Saǧāwindī (ʕilal al-Wuqūf, 482).


There is a certain conditioned distribution between these two spellings in the Cairo edition, but this appears to be absent in early Quranic documents (see van Putten 2018, 107 f.).

  • Collapse
  • Expand