Question about Arabic

[alice]

It seems reasonable to assume that Saudi, Aswad, and Sudan are all formed from the same triliteral S-W-D. But if that is indeed the case, what is the pattern for Sudan? The first two are clearly CaCCi and aCCaC, but Sudan requires CCCan, and I don't think Arabic permits three consonants in a row. What am I missing?

[Mecislau]

First of all, "Saudi" is a different root entirely—it's *sʕd, not *swd.


Your question ultimately boils down once again to the fact that the concept of "triconsonantal roots" isn't 100% accurate when it comes to Semitic languages. It's a convenient concept for many roots, but it's a gross oversimplification; there are many roots that don't fit a triconsonantal mould at all, but centuries of Arabic and Hebrew grammarians have tried to pigeonhole all Semitic roots into this model, even the ones that don't properly belong there.

To answer your question, the root in "Sudan" and "aswad" is, diachronically-speaking, *sūd. Not triconsonantal, not even discontiguous, just a plain monosyllabic root. That's why you see "Sudan". "Aswad" is an extension of the root *sūd, which is a trick used my many Semitic languages to modify roots that don't behave triconsonantally to be able to fit into patterns requiring three consonant, in this case *ʔaCCaC.

tl;dr: *sūd came first

[alice]

Well, I never claimed to be an expert on Arabic! Enlightened, thank you.

[Khvaragh]

It seems reasonable to assume that Saudi, Aswad, and Sudan are all formed from the same triliteral S-W-D. But if that is indeed the case, what is the pattern for Sudan? The first two are clearly CaCCi and aCCaC, but Sudan requires CCCan, and I don't think Arabic permits three consonants in a row. What am I missing?

Mecislau is correct; the root s-w-d was probably originally a biconsonantal root with a /u:/ vowel, which was later extended via analogous processes (this is also probably why similar roots in Hebrew, Arabic, and Aramaic differ only in where the vowel becomes a consonantal radical, showing the analogue was operating at different times), transforming the biconsonantal into the triconsonantal by analyzing /u:/ as /w/; i.e. probably something like /su:d/ > verbal root */sawada/ "he/it became black" > /sa:da/ (this is an uncommon or nonexistent verb, btw; usually it's understood as another verb, /sa:da/ "to be master of s.t.; the usual form is in fa``ala; /sawwada/ "he blackened," or if`alla /iswadda/ "he became black"). Synchronically, the reason it's /su:da:n/ is because the root is probably in a fu`la:n wazn, used usually for masdars, i.e. /Gufra:n/ "forgiveness" from G-f-r, /Gafara/ "forgiving." Arabic makes uses of certain morphophonemic changes to simplify disallowed vowel sequences, i.e. */bu:jd_?\un/>/bi:d_?\un/ "white (m. pl.)." The most common, of course, is the change of /awa~aja/ to /a:/, i.e. */qawala/ > /qa:la/ "he said." Gemination blocks this change. Sequences of /ij/ and /uw/ are subject to this as well, i.e. /ij/>/i:/, /uw/> /u:/, so /suwda:n/ > /su:da:n/

[Ser]

transforming the biconsonantal into the triconsonantal by analyzing /u:/ as /w/; i.e. probably something like /su:d/

But here you're contradicting Mecislau in that you're saying the root was indeed reanalyzed as a triconsonantal root s-w-d-

I like this way of thinking that Mecislau presents, it kinda shows how come in certain morphological processes, /u/s and /u:/s show up whenever they can (as in the verbal conjugation of form I "hollow" verbs with an alledged -w-). Then when they enter in patterns such as ?aCCaC or CaCCa:?, it just happens the vowel shortens as a process of becoming non-syllabic: /?asu_^ad/, /sau_^da:?/, to obey Arabic's phonotactics and in analogy with other roots which seem to be more truly consonantal. Then this might also help the question of the vowel found in the stem of perfect and imperfect verbs, which is entirely lexical too (CaCVCa; and if V = a, then yaCVC-, where both Vs are entirely lexical).

On the other hand, if this is true, then such roots show a pretty high number of cases of analogy and obedience to the language's phonotactics (which are pretty limiting: no contiguous syllabic vowels; after applying enchaînement, all syllables start with a consonant; no clusters of three consonants, and all existing clusters must be inside the word or between words, but not initially or finally; etc. —Then again, I'm talking about prescriptivist Standard Arabic, and yes, with full pronunciations, not sure how much of that isn't just an elegant conlang...). And these morphophonological processes make it tempting to say such /u:/s in roots have been reanalyzed as consonants, making most roots triconsonantal or biconsonantal, s-w-d, q-w-l, etc. On the other hand, Standard Arabic could simply be special in that case. (Or as I like to say, too much of a conlang...)

But then again, let's look at other natlangs. In Spanish, there's three stems you can find to do with rain: llov-, lluev-, lluv-, e.g. "llover" 'to rain', "llueve" 'it rains', "lluvia" 'rain'. The -o- and the -ue- stems alternate in semi-predictable ways, but the -u- isn't predictable. Can "llover~llueve" and "lluvia", from a synchronic point of view, be analyzed as having the same root? And what root would that be? Something like "ll-o/u/ue-v"? How does this affect our view of Standard Arabic?

> verbal root */sawada/ "he/it became black" > /sa:da/ (this is an uncommon or nonexistent verb, btw; usually it's understood as another verb, /sa:da/ "to be master of s.t.; the usual form is in fa``ala; /sawwada/ "he blackened," or if`alla /iswadda/ "he became black").

I take it that, traditionally, a form I verb is usually taken as the representation of the root (heck, if not the root itself!). Is there any sensible reason why we should take that a form I verb in particular is the most basic surface form of the root, specially for cases like s-w-d where a form I verb doesn't even exist? Why assume a */sawada/ verb meaning "to be/become black" that doesn't exist? What about certain biconsonantal cases like /dam/ 'blood'?

[Khvaragh]

transforming the biconsonantal into the triconsonantal by analyzing /u:/ as /w/; i.e. probably something like /su:d/

But here you're contradicting Mecislau in that you're saying the root was indeed reanalyzed as a triconsonantal root s-w-d-

Not at all; I think you're misunderstanding my post; I'm AGREEING that the biconsonantals (which is, in itself, probably just a convenient term to characterize roots) are original. It's the triconsonantal formulation which is derived.
The typical classical Arabic grammarian analyzing a root as triconsonantal does not mean that it was, in actuallity, as so (Grammarians argued for centuries over whether the root of /ism/ was ?-s-m, s-m-y, s-m-w, s-m-?, etc., when it was most likely just /ism/ or perhaps even PS */Sim~Sam/ which was the "real" root. By saying "analyzed" I mean the grammarians introduced an artificiality into the roots in order to make them fit the paradigm which they constructed.

I like this way of thinking that Mecislau presents, it kinda shows how come in certain morphological processes, /u/s and /u:/s show up whenever they can (as in the verbal conjugation of form I "hollow" verbs with an alledged -w-). Then when they enter in patterns such as ?aCCaC or CaCCa:?, it just happens the vowel shortens as a process of becoming non-syllabic: /?asu_^ad/, /sau_^da:?/, to obey Arabic's phonotactics and in analogy with other roots which seem to be more truly consonantal. Then this might also help the question of the vowel found in the stem of perfect and imperfect verbs, which is entirely lexical too (CaCVCa; and if V = a, then yaCVC-, where both Vs are entirely lexical).

I don't see how this is different from what I said; the morphophomenic rules of Arabic disallow certain vowel sequences, so when one might result from "plugging in" a root, whether "real" or "artificially analyzed," it results in the reduction of the disallowed form into an allowed one.

On the other hand, if this is true, then such roots show a pretty high number of cases of analogy and obedience to the language's phonotactics (which are pretty limiting: no contiguous syllabic vowels; after applying enchaînement, all syllables start with a consonant; no clusters of three consonants, and all existing clusters must be inside the word or between words, but not initially or finally; etc. —Then again, I'm talking about prescriptivist Standard Arabic, and yes, with full pronunciations, not sure how much of that isn't just an elegant conlang...). And these morphophonological processes make it tempting to say such /u:/s in roots have been reanalyzed as consonants, making most roots triconsonantal or biconsonantal, s-w-d, q-w-l, etc. On the other hand, Standard Arabic could simply be special in that case. (Or as I like to say, too much of a conlang...)

It's quite tempting to say that the Classical language was, in fact, something like a conlang. Many of the derivational templates are highly artificial, and rather Cartesian-looking, which leads me to think that when naḥw became a science in the Arabic world, it also led to the "correction" into a more orderly (in the minds of the grammarians) form. Keep in mind that naḥw was considered a branch of `ulūm al-dīn (religious sciences), so finding the underlying "perfections of the language" was something of a religious calling - and most grammarians were religious scholars. Arguably, the entire discipline came into existence to clarify the meanings of difficult lexemes and syntax in the Qur'an and Hadith corpus. This situation probably also has its roots in Pre-Islamic poetry, since the early poets were the closest to "standardizers" of grammatical usage before the advent of naḥw; for example, the alif maqsura is a vestige of divergence in dialectal pronunciations. A similar situation existed with Sanskrit in India, though arguably Sanskrit is more irregular due to more sound changes piled up in comparison to Arabic's relationship to Proto-Semitic, but I digress. In other words, I think we can quite easily conclude that the grammatical tradition of Arabic is indeed one of the most prescriptivist of any modern language, and there are a huge number of artificialities in the classical language (btw. this is also suggested by Kaye in his Arabic article in Comrie's "The World's Major Languages," p. 665).

> verbal root */sawada/ "he/it became black" > /sa:da/ (this is an uncommon or nonexistent verb, btw; usually it's understood as another verb, /sa:da/ "to be master of s.t.; the usual form is in fa``ala; /sawwada/ "he blackened," or if`alla /iswadda/ "he became black").

I take it that, traditionally, a form I verb is usually taken as the representation of the root (heck, if not the root itself!). Is there any sensible reason why we should take that a form I verb in particular is the most basic surface form of the root, specially for cases like s-w-d where a form I verb doesn't even exist? Why assume a */sawada/ verb meaning "to be/become black" that doesn't exist?

No, of course there's no reason to, nor am I assuming a theoretical */sawada/ existed - because it probably didn't. But historically, this is exactly what was done, and this is how the lexicographic works tended to be organized, mostly because the 3rd person masculine singular fa`ala form had (theoretically) the least "extra" derivational morphology. I should emphasize that I am describing the methodology of the grammarians, not my own.

As such, my use of a hypothetical */sawada/ is a purely theoretical construct to highlight how this was done, but this is the sort of methodology that grammarians used, and treated it as real. If I did not make it clear, then the progression of /su:d/ > /sawada/ > /sa:da/ is intended not to show real life (beyond the original root, I suppose), but rather how the roots were analyzed/artificialized, and as such, why many still think of this and other hollow roots as having an original consonantal radical, rather than a simple vowel (It's possible that there were some actual roots with a "real" medial approximant, but it's difficult to tell; it's also sometimes thought that Qur'anic /u:/ and /i:/ were actually /uw/ and /ij/ to begin with). For the grammarians, it was more elegant to understand them this way.

EDIT: Quick correction; I switched the positions of the schools by accident!
By the way, this method of describing roots is a feature of the Medieval Kufan school, which understood the verb as the "root," while the Basran school argued for the masdar (the truth is probably somewhere in between, but more to the side of the Basran school in the case of primitive roots). Sibawayhi's al-Kitab was supposedly aligned with the Basran school, but the verbal view is the dominant one in Arabic linguistics up until very recently (G. Flügel proposes in Die grammatischen Schulen der Araber that the normative view on Arabic grammar in medieval times is the result of a fusion of these school in Baghdad after the rise of the Abbasids; later scholars have commented that "schools" is more of a shorthand for "certain scholars from X place" btw.).

What about certain biconsonantal cases like /dam/ 'blood'?

In the view of the grammarians, yes. This can be seen in contrived "roots" like /dam:a/ "to coat, smear, color" and even better, /damija/, "to bleed." Also, very likely /fa:ha/ "to speak, utter" via /fam/ "mouth," moreso one of the construct forms: /fu:/, /fi:/, /fa:/, even more apparent via the plural; /afwa:h/. The final /h/ is likely itself an artificiality to extend the root - the same kind of use is found in Hebrew, which uses final -/h/ as a cover symbol for vowel/approximant final roots. The verb, not the primitive root/morpheme, is being taken as the origin, even though these verbs are in all likelihood denominatives of the primitive morpheme.

[Cathbad]

I take it that, traditionally, a form I verb is usually taken as the representation of the root (heck, if not the root itself!). Is there any sensible reason why we should take that a form I verb in particular is the most basic surface form of the root, specially for cases like s-w-d where a form I verb doesn't even exist? Why assume a */sawada/ verb meaning "to be/become black" that doesn't exist? What about certain biconsonantal cases like /dam/ 'blood'?

In fact, in cases like saada, using the form I verb actually detracts from analyzing the root properly. I'm thinking it's just because, for most roots, the third person masculine singular Form I is the "simplest" form available that still makes sense within the language - the three consonants of the root, each with a short fatHa. But of course, Arabic is not an engelang (well, at least not entirely, so it doesn't turn out to be that simple...

EDIT: ... see Khvaragh's post just above. If I hadn't been writing this for half an hour I would have been quicker, but he beat me to it alas.

[Cathbad]

By the way, this method of describing roots is a feature of the Medieval Basran school, which understood the verb as the "root," while the Kufan school argued for the masdar (the truth is probably somewhere in between, but more to the side of the Kufan school in the case of primitive roots).

!!!

This topic interests me very much. Do you have any books or articles on this you can point me to very quickly? (also by PM if it's easier)

[Mecislau]

Also, very likely /fa:ha/ "to speak, utter" via /fam/ "mouth," moreso one of the construct forms: /fu:/, /fi:/, /fa:/, even more apparent via the plural; /afwa:h/. The final /h/ is likely itself an artificiality to extend the root - the same kind of use is found in Hebrew, which uses final -/h/ as a cover symbol for vowel/approximant final roots. The verb, not the primitive root/morpheme, is being taken as the origin, even though these verbs are in all likelihood denominatives of the primitive morpheme.

In the case of "mouth", yes, the final /h/ was perhaps a non-original component added analogically to extend the root, although this particular case has some additional issues that make it more complex. The addition of final -h is seen in a few cases in various Semitic languages, but it is not a particularly common means of extending roots, and it may well be that it's actually a remnant of some older morphology that's now lost; this is hinted at by some of the patterns you sometimes see with words containing this -h-, such as both "father" and "brother" in a number of Semitic languages, where the semantic similarity makes you wonder whether there was in fact a specific function for this 'morpheme' at some point (cf. PIE *-ter- that appears in all sorts of IE kinship terms: father, mother, brother, sister, etc). So, in the case of "mouth" at least, this /h/ may actually have been an original component of the root rather than an extension of it.