The Alif al-Wiqāyah or 'Alif of protection' is an orthographical device in Classical Arabic where a <ʔ> is placed after a <w> in 3pl.m. verbs, e.g.
ḏahabū ذهبوا <ḏhbwʔ> 'they went', yaḏhabū يذهبوا <yḏhbwʔ> 'they go (subjunctive/apocopate)'.
It may also be used for final-weak verbs that end in aw, e.g. ramawرموا <rmwʔ> 'they threw'
In Classical Arabic these appear to be the only uses, but even a cursory look at the Quran, will immediately show that this use is much more widespread in the Quranic Consonantal Text.
Besides the uses where we also find it in Classical Arabic, we find for example:
- For the construct nominative plural -ū
Q37:37 ḏāʔiqū (a)lʕaḏābi (a)lʔalīmi ذايقوا العذاب العليم <ḏʔyqwʔ ʔlʕḏʔb ʔlʔlym> 'the tasters of painful punishment'
- For the plural imperative
Q8:25 wa-(i)ttaqū واتقوا <wʔtqwʔ> 'and fear!'
- For the 2pl.m. apocopate/subjunctive
Q2:24 tafʕalūتفعلوا <tfʕlwʔ>
- For word-final aʔu for imperfective verbs with their ʔ as the third root consonant:
Q10:4 yabdaʔu يبدوا <ybdwʔ> 'he begins'
- For the nominative of aʔ-final nouns (besides spellings <ʔ>)
Q38:67 nabaʔunنبوا <nbwʔ> 'news'
Q23:24 al-malaʔu الملوا <ʔlmlwʔ> 'the chiefs'
- For the nominative indefinite of (i)mrVʔ 'man':
Q4:176 imruʔunامروا <ʔmrwʔ> 'man'
This is quite an unusual collection of words, and it seems quite clear that there is no grammatical definition that ties these words together, like is the case in Classical Arabic.
However, let us now examine what words never have an Alif al-Wiqāyah:
- Stems that end in a consonantal w:
Q9:114 ʕaduwwunعدو <ʕdw> 'enemy'; Q63:4 al-ʕaduwwuالعدو <ʔlʕdw>; Q29:64 lahwunلهو <lhw> 'amusement'; Q7:199 al-ʕafwaالعفو <ʔlʕfw> but also Q(passim) huwaهو <hw> 'he'
- Stems that end in ūʔ and uʔ and awʔ:
Q(passim) sūʔ(a/i/u(n))سو <sw> 'evil'; Q52:24 luʔluʔun لولو <lwlw> 'pearl'; Q(passim) sawʔ- سو<sw>'evil'
- The particle ḏū'owner of' in the nominative. This however was not the case in older Quran documents where the spelling <ḏwʔ> is quite regular:
Q(passim) ḏūذو <ḏw>
- The monosyllabic words ʔaw and law are always spelled او <ʔw> and لو <lw>.
- A few cases of 3pl.m. endings written with just <w>
Q25:21 ʕatawعتو <ʕtw> 'they became insolent' (besides the expected عتوا <ʕtwʔ> in Q7:77, 166; Q51:44).
Q34:5 saʕawسعو <sʕw> 'they strive' (besides expected سعوا <sʕwʔ> in Q22:51)
- 3pl.m. verbs that end in ʔū:
Q12:16 ǧāʔū جاو <ǧʔw> 'they came'; Q2:61 bāʔūباو <bʔw> 'and they drew on themselves'; Q2:226 fāʔū فاو <fʔw> 'they return'
Making sense of all of this
It is quite clear from the above examples that no morphological principle is governing <w> versus <wʔ> spellings. Instead, it seems possible to assume that there's a phonetic principle at work here. The orthography seems to make a distinction between whether the <w> denotes a vowel or a consonant:
- Word-finalū is always written with <wʔ>
- Word-final w is always written <w>.
The only exception to the first rule is verbs of the type ǧāʔū. The reasons for this are rather complex, but have to do with an orthographical device <ʔw> which can denote /ʔu/ or /ʔū/ in words where the intervocalic <ʔ> was not lost (also occasionally word-initially hence ʔulāʔika as اوليك <ʔwlyk> and saʔurīkum as ساوريكم <sʔwrykm>). It seems that this orthographic device was considered to be a single grapheme <ʔw>, and did not so much denote the vowel ū, as the glottal stop, when it is followed by ū (kind of similar to the three Ugaritic glottal stop signs, depending on what vowel was next to it).
As for the second rule, it appears to be broken by the verbs that end in aw, this exception is however quite easily alleviated by assuming that (word-final) aw had monophthongized to ō, or that resulting diphthong was simply felt more of a combination of two vowels, than as a combination of vowel+consonant. Possible ambiguity that spawned from this might explain why ʕataw and saʕaw are both written without Alif al-Wiqāyah once, where the speaker hesitated between /ʕatau/ (and/or /ʕatō/) and /ʕataw/ as a phonemic representation (accompanied, with perhaps a free variation difference in pronunciation).
Words with the glottal stop
These rules help us place the loss of glottal stop in these cases in a new light. Let us first example words that have stem-final ūʔ-:
While there is no direct evidence (e.g. from the rhyme) that ūʔv̆ shifted to ūwv̆~uwwv̆ in the language of the QCT, this outcome seems the most logical outcome of this sequence after the loss of the /ʔ/ in this position. As such sūʔ- not receiving an alif al-wiqāyah can be understood as it represented /suww/, and like ʕaduww has a consonantal /w/ as final letter. Note that the outcome of this sequence is the same for uʔv̆, luʔluʔ-(un) must have yielded /lūluw-u(n)/, and the fact that it is not spelled <lwlwʔ> (as we find in Early Islamic Arabic), is a fairly good indication that there were word-final case vowels, causing the final /w/ to be consonantal rather than vocalic */lūlū/.
As for nouns and verbs that ended in *aʔu they appear to have lost the glottal stop, yielding word-final aw = /au/ or /ō/, yielding the vocalic spelling <wʔ>. The spellings with simply <ʔ> for the nouns, must then perhaps be explained as spellings of the pausal pronunciation *almalaʔ#> /almalā/.
This, finally also suggests that the word for 'man' in the indefinite form was *(i)mraʔ-, rather than the strange stem that harmonizes with the case vowel as is suggested by the Arab grammarians: imruʔun, ʔimriʔin, imraʔan. imruʔun would be expected to yield /imruwu/ and be spelled **<ʔmrw> rather than the attested <ʔmrwʔ> (cf. <lwlw> above). Instead we must posit a nominative /imrau/ or /imrō/.
The origins of this orthographic practice
So to conclude, I believe that word-final <w> denotes a consonantal /w/, whereas <wʔ> denotes a word-final vowel /ū/ or /ō/. But accepting that there are word-final short vowels one could alternatively formulate a slightly different explanation: /w/ followed by a vowel is spelled <w>, while absolute final /w/ (if [ū] is analysed as /uw/) is spelled <wʔ>.
Whatever explanation you prefer, it is clear that in the orthography of the QCT there's a purely phonetic motivation for the writing of the Alif al-Wiqāyah, rather than a morphological motivation as we find in Classical Arabic.
What is unclear, however, is where this orthographic practice came from. In Nabatean Aramaic (to which Arabic script and orthography can ultimately be traced), word-final ū was simply spelled with <w>, even in nouns of Arabic origin which in the Nabatean Arabic dialect had a final ū, (from which Arabic still have the spelling عمرو <ʕmrw> for the name ʕamr(un)), such words are never spelled with <wʔ>. The spelling must therefore be a post-Nabatean Aramaic innovation.
Being able to distinguish <w> as a vowel or as a consonant, is a useful distinction, but makes one wonder, why, if this was a fully conscious and 'artificial' innovation the same thing was not done with <y>, which just like <w> can potentially stand for a consonant /y/ or a vowel /ī, ē/ in word-final position.
It is true that in the writing style of, especially early Arabic it is often difficult to decide whether the <w> belongs to the end of a word, or is wa-'and' at the beginning of the word, and it has been suggested that the Alif al-Wiqāyah functions as a sort of word-divider to indicate that it is not the proclitic wa-. But it is difficult to understand why the orthography would make a distinction between consonantal and vocalic <w> in such a case, since both have the potential to be misunderstood in such a way.
Nabatean Arabic, appears to have retained the *ʔ in many more environments than Arabic of the QCT, so for example the name <dʔbw> is clearly ḏiʔbū, where QCT has <ḏyb> /ḏīb(u)/. One might imagine that a word like sūʔ which would be spelled *<swʔ>, might end up being pronounced sū, which gave rise to the spelling <wʔ> for word-final ū, this then could have expanded to places where ū did not go back to an old *ūʔ sequence. However, if that is indeed the path along which this orthographic device developed, it must have happened in a different dialect than the dialect of the QCT, as sūʔ presumably ended up as /suww(u)/ in the language of the QCT.
It is not outside the realm of possibility that the Arabic orthography that is employed for the QCT, was originally used to write a dialect different from the language of the QCT. As I have mentioned in earlier posts, the Arabic orthography must almost certainly have developed from a dialect that 1. did not have nunation 2. probably did not have final short vowels. While the indications are quite marginal, I think it is most efficient to assume that the language of the QCT also did not have nunation, but did have final short vowels.
The highly hypothetical situation therefore is, at least theoretically, possible. But we will have to find concrete evidence for such a dialect (and would we recognize it as such when we find it?).