Thread by @PhDniX

In my book "Quranic Arabic" I argue that if you look closely at the Quranic rasm you can deduce that the text has been composed in Hijazi Arabic (and later classicized into more mixed forms in the reading traditions). Can we identify dialects in poetry?

https://twitter.com/Quranic_Islam/status/1781866855990235365

I think this is possible to some extent, yes. And so far this has really not been done at all. Most of the time people assume complete linguistic uniformity in the poetry, and don't really explore it further.
But there are a number of rather complex issues to contend with:

As @Quranic_Islam already identified, there are some philological problems that get in the way in poetry that aren't there for the Quran: I would not trust a hamzah being written in a written down poem. This might be classicization. So it's hard to test for this Hijazi isogloss.

To learn anything at all, we would have to examine cases that affect the metre.
yaʾkulu and yākulu are metrically identical,
but yasʾal (HH) and yasal (LH) are metrically distinct, so you can see the loss of hamzah there.
Similarly saʾaltu (LHL) sāltu (HH)

Syllable structure, on the other hand is something we have *excellent* access to in poetry. You can easily show of a poem has wa-huwa (Hijazi) or wa-hwa (Najdi). There is however a problem: poetry, even if composed by a Hijazi cannot help but be unhijazi...

Why? Well because Hijazi syllable structure constantly creates sequences that do not fit in any poetic metre. Even something as basic as rusulunā "our messengers" has the impossible LLL sequence, and would therefore always take on the Najdi form ruslunā (HLH).

Vowels, like with the Quran, are extremely difficult to deduce. But, Poetry is rhymed, and the rhyme is sometimes multi-syllabic so we can at least learn something about vowel quality in the rhyme position.

We can learn if a certain poem has a distinction between /ē/ and /ā/.

Morphology that is metrically distinct can be discerned as well. Hijazi ḏālika (HLL) and hāḏihī (HLH) are clearly distinct from Najdi ḏāka (HL) and hāḏī (HH).
The problem is: poets noticed this too, and happily latch onto both forms whenever it is metrically convenient.

So poems will frequently have both in a single poem, sometimes even lines apart. It is worth stressing how different this is from the Quran, where ḏāka and hāḏī are totally unheard of. The Quran consistently has the Hijazi form, poetry frequently mixes its forms.

So the structure of poetry makes it conducive to including more forms that are not strictly native to the poet's dialect. So if one were to do any type of study like this, it would have to be a more statistical based approach:

Poet A is more likely to do X than poet B, and this makes sense because feature X is part of tribe C which Poet A belongs to.

For anything non-impressionistic to come out of this, you need to have a lot of data, dutifully collected and tagged in a way that is sensible.

You probably also need to be good at making sense of pre-Islamic poetry, which I'm absolutely not!

A couple of years ago @azforeman presented a great paper highlighting some of the linguistic variation, and his current research also focuses on the language of poetry...

I don't know if his approach will specifically try to address these questions. But I'm sure we'll come closer to having an insight into these questions once the PhD thesis is done. :-) Perhaps he's interested in following up with some of his own thoughts on the matter.