I've been reading the still unpublished Tashelhiyt texts collected by Roux.
Tashelhiyt is a funny language. In many ways, it is incredibly similar to Middle Atlas Berber, and it's easy to switch between the two. However, unlike Middle Atlas Berber, Tashelhiyt has (almost) no adjcetives whatsoever, and almost all 'adjective-like' expressions are expressed through relative sentences with stative verbs, making it significantly different from what I'm used to.
The only two adjectives I've run into so far are 'new' and 'old'. Considering their clear semantic connection, you'd maybe expect their form and syntax to be similar too. But this is not the case at all.
new is lždid and connects to the noun it modifies with the possessive preposition n 'of': lqbr nns n ljdid 'her new grave'.
While the adjective 'old' behaves like a noun that agrees in gender and number with the noun it modifies: lqubbt taqdimt 'the old tomb'
In terms of adjective-like expression, I ran into another interesting construction in a text I was Reading:
ɣayda ittwiyššan ula ntta zund lluz lmksur ma irẓan ...
(Listing things that can be bought at a fetival) Likewise, things that can be eaten like broken broken almonds ...
The construction lluz lmksur ma irẓan is especially interesting.
lluz 'almonds'lmksur 'broken' an Arabc passive participle of the verb ksər 'to break'
ma 'what, which' used as a relative pronoun
i-rẓa-n 'to break, be broken' in the subject relative form of the perfective stem.
So literally: Broken almonds which are broken.
lmksur here, is an Arabic style participle, a form for which Berber does not readily have an equivalent. While Berber languages do occasionally borrow participles, these are always borrowed in a nominalized form, and cannot be used attributively (Except in Ghomara Berber, where this is perfectly possible).
The attributively used participle here, therefore presents a problem. It seems that, to the speaker, this was also considered a problem, which is why a relative clause with a native Berber construction was placed behind it.
The construction lluz lmksur should probably be considered a quick 'code-switch', because of the common occurence of this participle with lluz in Arabic. Realizing the 'code-switch' the speaker seems to have corrected it, by introducing a Tashelhiyt-internal construction.
Sadly without recordings or insight into the original context in which this text was recorded, it's difficult to determine exactly how this construction appeared in the texts, nevertheless I thought it was striking and worth sharing.