Tea, that good old-fashioned drink made of dried up leaves hung in cooking water. Quite a few countries can be considered to have a true Tea Culture. England would be the first to come to mind for people native to Europe, but also Japan and China can be considered to have true tea cultures.
But also in Morocco, tea plays a big role. Thus the Berber speaking people will be delighted to hear that the word for tea can be reconstructed perfectly in Proto-Berber!
Wait what? Yes.
Tea is a word I did not end up using in my thesis, but was available in my corpus the lexical data that I have collected for this word is printed below.
Tamazight attay (wa-) pl. attayn `tea'; Tashelhiyt atay `id.'; Ouargla atay (a-)/latay `id.'; Ghadamès atay `id.'; Zénaga äʔḏäyi `id.'; Mali Touareg: ătayy pl. ătaytăn `id.'
So, from this evidence we could reconstruct aa Proto-Berer *aʔtay(ǝʔ?) quite perfectly, with really Proto-Berber looking glottal stops too!
The only irregularity there seems to be is that in Ghadamès the sequence *aʔ is expected to yield initial o. But this should hardly concern us.
What should concern us though, is that Tea was only introduced in North Africa through trade with the Dutch VOC in the 17th century! Surely Proto-Berber was not spoken a mere 400 years ago but much earlier, yet this word manages to look very Proto-Berber.
In fact this word probably is a very early loanword from Dutch thee 'tea'. This point is argued by Harry Stroomer (2007) Quatre emprunts anciens au néerlandais en berbère tachelhiyt and seems quite convincing to me.
This strange anachronism of something being both reconstructable for Proto-Berber and can be proven to be a loan no more than 400 years old should teach us to be humble about our reconstructions of Proto-Berber.
The Berber languages have managed to stay so close even after the Proto-Berber period that dialectal diffusion of a loanword can happen extremely quickly, and become nativised in little to no time.
If even the glottal stop in Zénaga cannot be completely trusted to be original, it is difficult to determine when it is a true native Proto-Berber word with a glottal stop, and when it isn't. Luckily for us, we find that in this word actually has an irregular reflex of the glottal stop in Ghadamès, so we have some evidence that this glottal stop is not original. It does make me wonder, how this glottal stop ended up in Zénaga. The final glottal stop in Zénaga (represented by the absence of a h) is thought to be from *ǝʔ usually, but I feel that this and several other words show evidence that also a vocalic y can yield this result.
More strange anachronisms can be pointed out in Proto-Berber reconstruction.
The Proto-Berber *γ is found reflected as such in nearly every Berber language, except for Zénaga where it is a ʔ and Ghadamès where it often is a ε (That is a voiced pharyngeal approximant ʕ in IPA).
The long counerpart to *γ is not **γγ but *qq which makes us suspect that the *γ may have originally been *q. Internal reconstruction shows this, but the comparative method leaves it without a doubt that *γ is the Proto-Berber phoneme. Yet, we find names of Berber people written down by Roman people where the *γ written with a <C> and we find loanwords from latin where the C before back vowels is loaned as *γ such as taγawsa 'thing' < Lat. causa.This clearly points to a pronounciation q.
Proto-Berber was almost certainly already diversified and spread before Latin contact, and Latin clearly points to Post-Proto-Berber having q, so from historical evidence q should be reconstructed while in all Berber languages q has shifted to *γ apparently after the languages had split up, the sound law was still able to spread along all dialects.
I hope this has somewhat illustrated the strange anachronisms you run into when reconstructing Proto-berber, and what kind of mindset you should have about reconstruction of this Proto-language due to its amazing ability to spread words and sound laws long after the period that Berber could still be considered one language.