http://www.armazi.com/georgian/unicode/class_4_verbs.htm

It seems a sensible enough strategy and I have an idea that something similar exists in Basque. ‘To me a horse is visible,’ is perhaps more natural than ‘I see a horse’. The horse may just carry on grazing, blithely unaware that he is under observation. If he is left unaffected by my action, why designate him as an object?

If you found Deutscher’s book rewarding you might enjoy ‘Dying Words’ by Nick Evans, which strikes a nice balance between the concrete and the abstract, including the coming back into fashion of Sapir-Whorf. By way of a cross-reference you may find appealing, he compares the potential of endangered languages to provide counter-examples with Littlewood’s theorem about the eventual change of sign of Li(x) – pi(x), aka the strong law of small primes.

The isolation of Tabassaran would certainly be consonant with the ‘island’ principle.

I often wonder to what extent these grammatical differences change the view of a people. I have heard that in Russian colours are effectively verbs: “the sky blues” rather than “the sky is blue”. If this is right does it effect the way you perceive colour, as an active rather than passive property? Interestingly the other Detscher book “Through the Language Glass”, is about Sapir-Whorf vs. Chomsky and uses colour words as the running example.

I like the idea of the grammatical system of Arabic being so richly interwoven that it cannot be changed. Reminds me of so much code I know! And self-reinforcing structures do seem to be a core aspect of the essence of life and knowledge. However, as counter-example, I believe that modern Hebrew has lost all but 5 of its verb cases (I assume more complex expressions are now constructed with auxiliaries as in English).

A linguist called Joanna Nicholls observed that there is a striking morphological cleavage between the languages of the North West Caucasus and those of the North East. The first are so-called head-marking and the second are dependent-marking languages.

The former term denotes very heavy verbal morphology, where verbs are marked for subject, direct object, indirect object and much else besides, but negligible noun morphology, eg only two case markers, direct and indirect.

The latter term, typical of Chechen, Avar and about thirty others, is a sort of mirror image: negligible verb morphology (tense, aspect and mood are all done with auxiliary verbs) but a mesmerising system of noun classes and case markers.

As to why these languages might resist simplification … with Arabic etc, it is difficult to see how you could mess with the poles of the triliteral root without bringing the tent down The enduring complexity of the NEC languages might have less to do with any overarching structural principle which cannot be tampered with than the isolation and topography of the relevant terrain (42 locatives in Tabassaran, if memory serves, which it probably doesn’t).

(* Having no more than a pedestrian mind, I have never found out how to combine the two interests successfully.)