Lyrics and language preservation?

The Berber-speaking oasis of Siwa in western Egypt, where I did doctoral fieldwork, has a rather extensive poetic tradition embodied in song lyrics. Practically every Siwi I spoke to quoted me some lyrics at some point, and songs in Siwi apparently remain a key element of parties. I included a few of these in my book about Siwi, among them four nicely arranged lines referring to Shali, the main town of Siwa:

Quṛ, ya lmendi, quṛ!	Dry, O wheat, dry!
Baba nnek yexsa Cali.	Your owner misses Shali.
Nan edderb n Cali,	People of the Shali road,
Sellemm-i af elɣali.	Give the beloved my greetings.

It wasn't until later that I finally received a copy of Bricchetti-Robetti's (1889) article "Sul dialetto di Siuwah", and not until this month that I finally got around to reading it carefully. When I did, I was surprised to find this poem transcribed practically word for word:

qor aimindi qor	قور ايمند قور
babenik jiksa - sciali	ببنك يخس - شالى
nani derbj ensciali	تندرب انشالى
salamuet - afelrhali	سلموت افلغالى

Apparently, these lyrics have been passed on orally for more than 120 years, with only minor changes.

There are many ways in which Siwa is different from Tabelbala, the Algerian oasis where I did the other half of my doctoral fieldwork. Linguistically, one that struck me early on was the variability of Tabelbala's language, Korandje, compared to Siwi. In Siwa, there was some interesting variation even within the speech of single individuals (1st sg. -ɣ- vs. -ʕ-, negative copula qačči/'ačči/ɣačči), but it hardly seemed possible to speak of dialects. In Tabelbala, not only did different villages take pains to distinguish themselves by different ways of speaking, but neighbours and cousins often showed substantial differences in pronunciation and even sometimes vocabulary. And whereas Siwis rarely seemed at a loss for words, in Tabelbala even the oldest speakers routinely had trouble finding a word, or disagreed on its meaning once they had remembered it.

Another striking difference is the low profile of Korandje poetry, if it exists at all. Whereas in Siwa I could hardly stop people from telling me lyrics, in Tabelbala my utmost efforts barely dredged up a few ditties which the speakers themselves considered absurdly simple. The poetry that men cared about and appreciated was in dialectal Arabic, and even that was far less prominent than in Siwa. (Some older women reportedly sing Korandje poetry in honour of the Prophet at regular Sufi gatherings, but I was unable to hear any of that; given its subject matter, I suspect the language used would be heavily influenced by Arabic.)

One possibility I'm tempted to consider is that these two facts are causally linked. In Siwa, songs are heard and sung in groups, and the best lyrics are widely circulated and - apparently - remembered for many decades; their rhythm and rhyme makes major rewording impractical. Logically, this should keep less frequently used vocabulary in circulation in much the same way as a written literary tradition, or a national broadcasting service. Without songs, for instance, would Siwi have kept a Berber word for "gazelle" (izem), an animal rarely if ever seen in the oasis today, but to which the beloved is constantly compared? In Korandje, on the other hand, the standardising force of songs and poetry is practically absent, and it's not obvious that anything else in their verbal arts (already sadly atrophied by television) compensates for it.

Does this reflect your experience, or contradict it? How do poetic traditions (or lack of them) in societies you're familiar with seem to affect the prospects for their languages?