In Classical Arabic orthography the glottal stop [ʔ] is represented with the so-called hamza ء a letter which is often placed on top, or below other consonants to denote the presence of this glottal stop. Its diacritical nature, is because the QCT orthography did not have a hamza to denoted the presense of the glottal stop, and because the QCT had to be kept intact when the hamza was added to the Qurʔān, it was placed, on top of, below or besides the consonant that it followed or preceded.
Historically, the sign that would denote the glottal stop in the Semitic scripts was the aleph, which is the ancestor of Greek Alpha, Α, α, Latin A, a. This sign shows up in Hebrew and Aramaic as א and in Arabic as the alif ا. This sign has lost most of its consonantal value in Aramaic, and subsequently Arabic (which inherited its script from Aramaic) due to phonological developments of the language. In Aramaic, the vowel denotes a long ā in word-final position. In Classical Arabic, it also denoted a logn ā in word-final position, and in modern Classical Orthography also in medial position.
The use of alif as a sign for word-medial ā is a Arabic innovation, and it appears to be an orthographic convention which was not fully established at the time that Qurʔān was composed. Hence a word like samāwāt- 'skies, heavens' is consistently spelled سموت <smwt> in the Qurʔān, and never as سماوات <smʔwʔt> which is the spelling in Classical Arabic. In the modern QCT we find some examples of word-medial ā being written with alif, such as the extremely common qāla <qʔl> 'he said'.
Older manuscripts of the Qurʔān, even the ones that date from very shortly after the death of the prophet Muhammad, are strikingly close to the modern QCT and show just how well the original text was preserved. However, it is exactly the cases of word-medial ā where the QCT of such manuscript deviate from the modern QCT. For example,
It is exactly in these positions, however that old Qurʔān manuscripts may deviate from the modern QCT. If we look at the text in the Birmingham manuscript, probably the oldest dated piece of Qurʔānic text we find a clear example of this.
Q20:10 in its current form reads:
ʔiḏ raʔā nāran fa-qāla li-ʔahli-hi (u)mkuṯū ʔinnī ʔānastu nāran laʕallī ʔātīkum minhā bi-qabasin ʔaw ʔaǧidu ʕalā (a)n-nāri hudan
QCT <ʔḏ rʔ nʔrʔ fqʔl lʔhlh ʔmkṯwʔ ʔny ʔnst nʔrʔ lʕly ʔtykm mnhʔ bqbs ʔw ʔǧd ʕly ʔlnʔr hdy>
BQCT <ʔḏ rʔ nʔrʔ fql lʔhlh ʔmkṯwʔ ʔny ʔnst nʔrʔ lʕly ʔtykm mnhʔ bqbs ʔw ʔǧd ʕly ʔlnʔr hdy>
Notice, however, that at the same time, the word nār which appears three times in this verse, is consistently written with an alif. And this seems to be typical for the orthography of the QCT, word-medial alif for ā is inconsistently written, strongly depending on the word.
A similar variant of qāla is found in the Sanaa manuscript in Q7:38:
qāla (u)dxulū fī ʔumamin qad xalat min qablikum mina (a)lǧinni wa-(a)lʔinsi fī (a)n-nāri [...]
QCT <qʔl ʔdxlwʔ fy ʔmm qd xlt mn qblkm mn ʔlǧn wʔlʔns fy ʔlnʔr [...]>
SQCT <ql ʔdxlwʔ fy ʔmm qd xlt mn qblykm mn ʔlǧn wʔlʔns fy ʔlnʔr [...]?
[Edit 20/1/2016: I've now also found a few instances in the Cairo Qurʔān standard text with ql for qāla: Q21:112, Q43:24, Q23:112, Q23:114]
It is clear that this is an orthographic practice still in development. And one wonders where this practice came from. There are several possibilities:
- The Arabic scribes simply generalized the word-final rule to a word-internal position.
- Arabic underwent a sound shift which causes a previously consonantal writing to now represent an ā
While the first option is certainly possible, it is worth considering option 2. Although the consonantal value of the alif appears to be diminished, it should be noted that words which certainly had a glottal stop at least originally, are always written with it. Examples are:
raʔs 'head' راس <rʔs>
saʔala 'to ask' سال <sʔl>
In many modern dialects the sequences /aʔ/ and /aʔa/ become ā. It is easy to imagine that in a dialect where this development has taken place, the difference between qāla and sāla, traditionally written <ql> and <sʔl> would become very difficult to maintain. Hence the 'uncertainty' between the spellings arising.
In early Islamic (non-Qurʔānic) texts, we find clear examples of this shift having taken place, causing saʔala to be often written سل <sl> 'he asked', presumably *sāla. Such forms appear already in the first century of Islam, and it seems safe to say that there were certainly varieties of Arabic that had undergone the aʔ(a) > ā shift already at the time the Qurʔān was composed.
Contrary to what is often said about the Qurʔānic orthography, there is real reason to think that the glottal stop was retained in the language of the Qurʔān, albeit in limited environments. In the environment given about, that is, after a, the QCT always has a written alif in every word with an etymological glottal stop:
- baʔs 'might, wrath' باس <bʔs>
- daʔb 'plight' داب <dʔb>
- daʔaban 'as usual' دابا <dʔbʔ>
- raʔs 'head' راس <rʔs>
- raʔfah 'compassion' رافة <rʔfḧ>
- raʔā 'to see' راى <rʔỳ>
- saʔala 'to ask' سال <sʔl>
- šaʔn 'matter' شان <šʔn>
- ḍaʔn 'sheep' ضان <ḍʔn>
- kaʔs 'cup' كاس <kʔs>
naʔā 'to become remote' نا <nʔ>, N.B. In Classical Arabic this word is spelled ناى <nʔỳ>, but it is treated in the Qurʔān as a final w verb rather than a final y verb, e.g. Q6:26 wa-yanʔawna وينون <wynwn>. One might expect a spelling ناا <nʔʔ> instead of نا <nʔ> in this verb, but it's easy to imagine that the sequence اا <ʔʔ> was avoided for orthographical reasons.
In other environments though, the alif is consistently lost. This is easy to show with the verbs with a medial alif, as here we can find verbal forms and nominal derivations of the root. Let's take saʔala as an example.
suʔila 'he was asked' سيل <syl>. The spelling quite clearly points to *suyila through a development Vʔi > Vyi
tusʔalu 'you will be asked' تسل <tsl> Post-consonantal ʔ is lost CʔV > CV. N.B. This different from Classical Arabic spelling, which would spell it <tsʔl>.
sl 'ask!' سل <sl>, grammar would predict (i)sʔal. but even in Classical Arabic verbs such as these can irregularly lose the glottal stop yielding sal سل <sl>. (i)sʔal اسال <ʔsʔl> is apparently also attested (Fischer 2002: 129).
suʔāl 'demand' سوال <swʔl> points to *suwāl through uʔā > uwā.
suʔl 'request' سول <swl> points to *sūl through uʔ > ū
sāʔil 'asking' سايل <sʔyl> points to *sāyil through Vʔi > Vyi
masʔūl 'questioned' مسول <mswl> points to *masūl through CʔV > CV.
This root lacks some environments where specific developments have taken place in the QCT, this is iʔ > ī as found in ḏiʔb 'wolf' ذيب <ḏyb>
Another example lacking in these examples is the sequence Vʔū, as found in the plural ruʔūs 'heads', and badaʔū 'they began' which, while written with a glide in the Classical Arabic orthography which suggests Vʔū > Vwū, it is written with only a single waw in QCT روس <rws> and بدو <bdw>, this seems to suggest a reading rūs and badū. Although one might argue that the sequence of two w graphs were simply avoided in the orthography.
We find the same development in front of an ī: baʔīs 'wretched' written in Classical orthography as بئيس <byys>, but in QCT as بيس <bys> = bīs ?
The rules formulated here can be summarized as follows:
- *aʔ(a) is retained.
- *CʔV > CV
- *uʔVV > uwVV (where VV stand for long vowel), also the plural ruʔūs 'heads' رووس <rwws> should be considered the result of this rule.
- *Vʔi > Vyi
- And perhaps: *VʔVV > VV
The question then becomes: To what extent does this writing represent orthographic convention of Arabic at the time, and to what extent does it actually reflect the language of the Qurʔān.
I am inclined to think that this represents an actual situation. As mentioned above, there are examples of early Islamic writing (almost) contemporary with the QCT, and when these dialects lose the distinction of aʔ and ā chaos in the orthography ensues. And while the spelling of ā is certainly chaotic in the QCT, the spelling of aʔ and aʔa never is.
Moreover, the QCT deviates from the later Classical Norm by consistently not spelling the glottal stop in post-consonantal position. I believe this should be interpreted as an actual reality of the language. Consistently omitting the consonant while they pronounced it, while in other environments its consistently written seems highly improbable.
References
Fischer, W. 2002. A Grammar of Classical Arabic, translated from German by Jonathan Rodgers, 3rd rev. ed. New Haven & London: Yale University Press
Good stuff. The glottal stop in kaʔs may well be secondary, as it isn't attested elsewhere: Hebrew כוס with no aleph; Akkadian /kās-/ where *kaʔs- may have yielded **/kēs-/. Neither cognate completely precludes it, though.
Posted by: Benjamin | 01/13/2016 at 04:52 PM
Interesting. There are a bunch of dialects that retain the Glottal stop in Yemen (always Yemen...), looking in Behnstedt's dialect dictionary/atlas the same dialect has raʔs pl. rūs but kās pl. kīsān (Dialect of Ṣanʕāʔ).
So apparently even within Arabic, there's little evidence for a glottal stop there.
Posted by: PhoeniX | 01/15/2016 at 10:09 AM
"However, the recitation of the Prophet peace be upon him was similar to what is known as the recitation of Warsh from Nafi’. The accent of the tribe of Quraysh did not have the “hamza” letter when it came in the middle of the word. For example, they would say “moomin – مومن” not “mo’min – مؤمن” (believer)." - Mohamed Ghilan.
Although I believe Warsh does have post-consonantal hamza.
Posted by: Lameen | 01/15/2016 at 10:27 PM
Both kās and kaʔs would have yielded the same plural, kuʔūs; perhaps the latter is to be explained as a back-formation from the plural.
Posted by: Lameen | 01/15/2016 at 11:15 PM
Would the plural really become kuʔūs? I would probably sooner expect kuwūs then, but not sure if Arabic allows w in between two back vowels.
The only medial weak stem with a CuCūC plural I can currently think of is ṭāʔir pl. ṭuyūr, but that's a y as the medial consonant.
But yeah If you're right that the ʔ would show up there, the back formation solution would make sense.
Posted by: PhoeniX | 01/15/2016 at 11:25 PM