Thread by @PhDniX

The famous 12th century scholar al-Zamaḫšarī was not only known for his popular exegesis al-Kaššāf, but also worked as a grammarian. The other day I was translating a little section on the demonstrative pronouns from his grammar al-Mufaṣṣal fī al-Naḥw. A short thread:

The section I was looking at can be found on page 55f. of Jens peter Broch's edition from 1879! It is a very nice and concise description of the demonstrative pronouns.

What is striking when you go through it is how little prescriptive it is, and how utterly un-Quranic it is.

Here the base pronominal forms are introduced. Striking is the enormous amount of options for the feminine singular. Moreover what is striking is that the iconic hā- prefix is missing. Unprefixed, while often dutifully mentioned are mostly absent in actual written Arabic.

The exception to this, of course, are the pre-Islamic Nabataean Arabic inscriptions, which use ḏā and tī.

"I, PN son of P.N. built this Martyrion" = انا [...] بنيت ذا المرطول

"This is the funerary monument of Mar al-Qays son of Amr" = تى نفس مرالقيس بر عمرو

In the Quran, the distal deictics (optionally) agree with the addressee by conjugating the final -ka to gender and number (-ki, -kumā, -kum, -kunna). This is a typical feature of the Quran, occasionally seen in poetry; I don't think I've ever seen it in Classical Arabic prose.

Zamaḫšarī seems to imply that the -ka can be attached to any base to create non-proximal deictics listing forms such as ḏāka, tākā, tīka, ḏīka. None of these forms exist in Quranic Arabic. ḏāka is quite frequent in Arabic poetry, and occurs sometimes in Classical prose.

Addressee agreement on demonstrative pronouns in general is a rather rare feature. The extremely exotic Yemeni dialect of Jebel Razih has it (as @ProfJCEWatson showed). Siwi Berber has it too (as @lameensouag showed).

Al-Zamaḫšarī suggests a difference between ḏāka and ḏālika based on distance. I've never seen it unambiguously used this way.

This is a departure from early grammarians. To al-Farrāʾ, for example, the -li-ka forms are simply Hijazi forms, while the -ka forms are Tamīmī forms.

Al-Zamaḫšarī himself seemed bothered by his purely theoretical -li- form tā-li-ka, as he explicitly mentions this is "rare" (read: purely theoretical). ti-l-ka is the actual form.

Nothing in his description prevents forms like ḏilka, but not mentioned.

Now finally he mentions the possibility of adding the hā- which is for to call attention, suggestion some kind of presentative function ḏā l-kalb "this dog" and hāḏā l-kalb "this here dog". These may even be applied to medial forms such as hāḏāka and hātīka.

Here what al-Zamaḫšarī describes diverges contrasts sharply with reality.

In the Quran hā- has no special valence, it is simply part of the proximal deictics: hāḏā, hāḏihī. I have never seen it attached to forms like hāḏāka and hātīka anywhere.

I wouldn't be surprised to see them show up in poetry, but in prose such forms are unthinkable.

Of the the theoretical forms hātā, hātī, hāḏī only hāḏī gets quoted with some regularity in poetry. The typical prose/Quranic form hāḏihī goes completely unmentioned.

Finally a quick description of hunā. The form he cites, hunālika is quite typically Quranic (as is hāhunā). In Arabic prose we usually see hunā and hunāka although the Quranic forms occur sometimes.

The text is actually surprisingly well-structured and clearly written, very different from the rather messy (and utterly spread out) discussion of the same topic by Sībawayh. If we were to put this description into a table we get the following.

The descriptions of the Arab grammarians (not just al-Zamaḫšarī) tend to be rather maximalist, trying to capture as many theoretical and attested forms as possible in a single description (whose actual usage reality might not quite line up with what is described).

But for a linguist it is worthwhile pulling these corpora apart, as regular patterns appear that suggest that it is worthwhile to make that differentiation.

The Quranic system looks as follows:

In the poetry, there are clearly more options available, as the meter requires it. I honestly don't read enough poetry to have any intuition what the duals look like, so I've left them out, put some question marks with other parts I'm not sure about.

The Classical Arabic prose system is quite close to the Quranic system, but with some deviations. This is the system al-Zamaḫšarī uses himself. What is interesting though is that al-Zamaḫšarī at no point describes that these are the forms you are "supposed" to use.

In fact, the theoretically derivable hāḏihī doesn't even get mentioned at all!

This is typical not just for al-Zamaḫšarī, but for all Classical Arabic grammarians. Large amounts of variation get mentioned, much of which is virtually or actually unattested.

If one were to read al-Zamaḫšarī (or Sībawayh, al-Mubarrad, Ibn Jinnī, Ibn al-Sarrāj etc.) one would never come away knowing how to write standard Classical Arabic themselves. It is never prescribed.

The idea that the grammarians standardized Classical Arabic is questionable.

At the same time, there can be no doubt that by al-Zamaḫšarī's time what comes to be considered the standard form of Classical Arabic had clearly crystallized to quite a large extent. The norms for Classical prose were understood. But who actually established those norms?!