There's something weird about the Classical Arabic word for the torah, the Jewish holy book. The Hebrew word is tōra spelled תורה <twrh>, and its a derivation from the verb yara, which in Hebrew has a root *yry, but as we see the w reappearing when y is not initial, this points to the well-attested initial *w > y shift found in hebrew (cp. Hebr. yɛlɛḏ 'boy' Ar. walad), so the original root must have been *wry.
The Arabic word, which is certainly a loanword from either Hebrew, or indirectly from Aramaic, is tawrāh توراة <twrʔḧ> (or in the Classical Pronunciation tawrātun). This is not too surprising and seems to be a good approximation of the original Hebrew word. The problem however, is its spelling in the Qurʔān. In the Qurʔān the word is attested several times, but every time it is spelled as: تورىة <twryḧ>. The additional vocalisation adds a 'dagger alif' on top of this yāʔ to point out that you should read it as a long ā and not as an, as you might expect ī, ay, aya or iya: تَوْرَىٰة, but the spelling nevertheless remains highly surprising.
There are other nouns with final -āh/-āt which also contain unexpected semi-vowels in the Qurʔānic text, where Classical Arabic spells it with an alif. Many of these fall in the realm of religious vocabulary, and may often be considered loanwords from Aramaic:
ṣalāh صلوة <ṣlwḧ> 'prayer', compare Aramaic ṣlō, ṣlōṯ-ā <ṣlw, ṣlwtʔ> 'id.'
zakāh زكوة <zkwḧ> 'alms, religious tax', compare Aramaic zāḵū, zaḵūṯ-ā <zkw, zkwtʔ> 'merit, victory'
ḥayāh حيوة <ḥywḧ> 'life', compare Aramaic ḥayyū, ḥayyūṯ-ā <ḥyw, ḥywtʔ> 'id.'
miškāh مشكوة <mškwḧ> 'niche (of a lamp)', I could not find the Aramaic origin of this word, but Fischer (2002 :7) considers it an Aramaic word, it probably is my knowledge of Aramaic is limited.
While the 'Aramaic origin' explanation is attractive for words that have obvious religious connotations, there are also clear examples of native Arabic words with the same spelling:
ġadāh غدوة <ġdwḧ> 'morning', a synonym of the word ġudwah, which in QCT is written the same. This certainly is a native word to Arabic, I have not found any Aramaic cognates. Instead this form seems to come from a Proto-Arabic *ġadaw-at-u(n). The QCT orthography reflects the original form.
naǧāh نجوة <nǧwḧ> 'salvation', while semantically clearly part of the religious vocabulary, but does not seem to have an Aramaic cognate either. < Proto-Arabic *nagaw-at-u(n). It also clearly has a masculine counterpart naǧan 'id.' < Proto-Arabic *nagaw-un.
A word not usually mentioned but probably part of the Aramaic religious vocabulary that influencd Arabic is:
manāh <mnwḧ>'fate, Manāt, one of the three Pre-Islamic goddesses'. A goddess that was obviously revered by the Nabataeans, who were presumably a mostly Arabic-speaking kingdom, although they mostly wrote in Aramaic.
Despite its obvious Nabataean origin, the word itself might not be Aramaic, but an Arabic name of a Goddess, as the Nabataeans seem to be mostly Arabic-speaking. The closest Aramaic word for this Goddess is <mnh> mənāh, mənāṯ-ā 'part, portion', developing into 'that is what is allotted to you' > 'fate'. It seems more likely that the name of the Goddess comes from the Arabic linguistic stratum in Nabataean-Aramaic.
The famous JSNab 17, one of the few Nabataean-Arabic inscriptions, has the name of this goddess in the name <ʕbdmnwtw> /*ʕabd manōtū (or manātū )/ 'servant of Manāt'.
JSNab 17 is accompanied by a Thamudic D inscription, which contains the same personal names. But in the Thamudic D script the name is spelled <ʕbdmnt>. Thamudic D does not use matres lectionis and would only write w if it had a consonantal value. This shows that mnwt was read with a vowel where Arabic today has ā. It may either have been ō or already ā (although that would be irregular from an Aramaic orthographical form).
In Safaitic, which does not write non-consonantal w, and not even w in word-internal diphthongs, however attests one inscription with the w being written consonantally in the name <ʔrśmnwt> (MISSB 1) /*ʔarś manawat?/.
While it is not clear how the QCT would have pronounced this name, that it was originally something like Proto-Arabic *manaw-at-u(n), is clear.
Besides words that write <wḧ> for final āh in QCT, there are also two words besides tawrāh that use <yḧ> for final āh. Both are clearly of Arabic origin, and certainly originally contained a y:
tuqāh تقية <tqyḧ> 'precaution', a feminine of tuqan 'devoutness, piousness' from the verb taqā 'to fear (god)'. On a Proto-Arabic level these forms point to *tuqay-un for the masculine and *tuqay-at-u(n) for the feminine. This shows that the sequence *aya, seemingly still preserved in the QCT orthography lead to a ā in Classical Arabic.
muzǧāh مزجية <mzǧỳḧ> 'being of little value' a passive participle of the causative verb ʔazǧā 'to shove, push', through a rather interesting semantic development 'one who is shoved, pushed' > 'being worthless/of little value'. It is the feminine counterpart to muzǧan. This certainly comes from a Proto-Arabic form *muzǧay-at-u(n).
The suggestion that the w is due to orthographic conservatism of the original Aramaic orthography, while possible in some case, certainly does not apply for all cases. Moreover, the cases where ā is written with a yāʔ can never be attributed to Aramaic influence.
So how do we deal with these assorted facts?
It is clear that whatever eventually became āh, comes from an earlier *awat- or *ayat-. It is also clear that already in Pre-Islamic times there were non-consonantal pronunciation, at least of *manawat-. This suggests a monophthongization like:
*awat- > *ōt, *ayat- > *ēt
In fact, this is exactly the pronunciation you get for *ayat sequences in the Warsh reading tradition, e.g. Q3:93 has the word tawrāh in it, which is read (a)t-tawrētu. The dot below the <r> indicates that the next vowel is to be read as an ē (Note: Warsh does not read *awa as ō, but simply as ā).
However, the QCT orthography indicates a shift of final t to h, this need to be accounted for in some way.
We might imagine that ōt, ēt shifted to ōh, ēh and that both vowels eventually shifted to āh. This is essentially possible, but rather ad hoc. Why would the rather rarely occurring vowels ō and ē have that effect on final t, while neither āt nor ūt or īt have that effect on final t.
It therefore seems more attractive to make these sequences take part in the common -at > -ah shift as found in the regular feminine ending.
*awat, *ayat > *awah, *ayah > *ōh, *ēh
The sequences aya, awa are not only found in these nouns. Also final-weak verbs such as ramā رمى <rmy> 'to throw' and daʕā دعا <dʕʔ> 'to call' originally come from *ramaya and *ramawa. The sound law would then predict ramē and ramō. The former is unproblematic, this is the Warsh pronunciation, and the spelling has final <y>.
However for daʕā, we have a problem in the QCT orthography. Final original awa sequences area always written as ā with a final alif. We can solve this in several ways:
1. Assume that word final vowels were lost, and that aw# > ā but awa > ō.
*daʕawa > *daʕaw > daʕā
This solves the problem, but will, as we'll see in a later blog post, cause problems. The Warsh reading tradition seems to have different reflexes for *aya# and *ay# sequences, so if the final -a was lost, the most obvious solution to the Warsh ē would be lost.
There are ways around this, and we will get back to this solution in the upcoming blog about the diphthongs.
2. Assume that the -ōh words are loanwords from a Nabataean-Arabic which did have awa > ō, while the dialect of the QCT did not.
This is actually an attractive solution as it explains why all <-wḧ> nouns are part of the religious vocabulary. Even the words that are clearly not of Aramaic origin, like manāh 'fate, the Goddess Manāt' and naǧāh 'salvation' are obviously part of the religious vocabulary. The only word that seems to fall outside of this set of words is ġadāh 'morning', which is identical to ġudwah 'morning' in the QCT orthography.
We may imagine that the reading tradition here is inaccurate in the vocalisation of the word, and that the QCT originally was intended to read ġudwah instead of ġadāh.
This however leaves us with the problem that there's not a single noun with the *CaCaw-at- pattern in QCT besides these words. Considering the paucity of this pattern in general, this need not be a problem but is worth noting.
While perhaps Nabataean-Arabic (not Aramaic) spelling or pronunciation of religious words may explain the <wḧ> feminines, it cannot explain any of the <yḧ> spellings. Therefore, tawrāh, somewhat surprisingly, cannot be a direct loanword from Aramaic or Hebrew. In Hebrew the word is spelled <twrh> without a yod. Surprisingly, the Arabic yāʔ is etymologically correct as it comes from a root *wry. How did Arabic orthography manage this feat of etymological magic? An option might be an Aramaic source. In Aramaic the Torah is called <ʔwrʔh, ʔwrytʔ> ʔōrāˀā, ʔōrāyt-ā. Were Arabic speakers somehow able to derive the root from that word and form their own native *tawrayah? That solution would only make sense if tafʕal(-ah) was a native noun formation, which seems to be absent from Fischer (2002). So that solution is not quite complete. The Arabic word somehow looks like a mixture between the Aramaic and the Hebrew form, but this is hardly a satisfying solution. Whatever the language was that was the donor of the term was pronouncing it as *tawrayah, or *tawrēh rather than tawrāh, as there simply is no reason to write a long ā with yāʔ within Arabic orthography.
The final question we may ask is whether the QCT orthography represents the actual language of the QCT, or whether it is purely an orthographical standard. This question is difficult to answer, and depends on how much weight you want to place on the reading traditions of the Qurʔānic text. The Warsh reading tradition clearly distinguishes ā from ē in these words (although not the hypothesized ō), which suggests that it may have been a legitimate pronunciation of the word. In future blog posts, though, I will show that the reading traditions differ radically from what I think can reasonably be assumed to be the original language of the QCT.
Using it as evidence would therefore be inconsistent within my model. However, others who might disagree with my reconstruction of the language of the Qurʔān, might have less trouble accepting the reading traditions.
Fischer, W. 2002. A Grammar of Classical Arabic, translated from German by Jonathan Rodgers, 3rd rev. ed. New Haven & London: Yale University Press
[Edit A spelling that still needs an explanation is ar-ribā <ʔlrbwʔ>.]