This map of Kura Araxes sites is from Wilkinson 2014.

 This map of Kura Araxes sites is from Wilkinson 2014. Its possible that he didnt mention all known sites but overall it shows well the pattern of their geographic distribution.

As You can see there are very little known Kura-Araxes sites in middle and lower Kur river region. This is exactly the region where Early Kurganic culture of steppe origin mixed with locals (2550-2300 BC) according the archaeologic studies that i posted earlier.

In Late Chalcolithic period this region was occupied by Leila tepe culture people genetically related to Chaff faced ware horizon. Those people had much more Anatolian/Levantine neolithic related ancestry and less CHG than Kura-Araxes. This is the reason why the succeeding Trialeti Vanadzor culture and Lchashen-Metsamor show excess of Anatolian related ancestry. This shift is visible also in Y DNA. By the presence of haplotypes like E-M84 and T1a1a in Armenia MLBA period. Those haplotypes are very unlikely to represent typical Kura-Araxes haplotypes.

The second picture is a qpadm model made by Nareg Asatrian where it is visible that Yamnaya Moldova works quite well as a source for ancestry for Armenia MLBA. Its a reproduction of the model done in Yediay et al. 2024.

The real reason for samples from Moldova working well is that they have the extra Anatolian farmer related ancestry necessary for Armenia MLBA. For reasons that i explained above and absolutely unrelated to Moldova. Another reason is the extra WHG but it will be a subject for a different post.

In conclusion the linguistic substrate for Proto Armenians was not the Kura-Araxes culture or lets be more cautious was not exclusively the Kura-Araxes. Chaff faced ware and Leila tepe's hypothetic language(s) also must have important influence on Armenian.