[aatharuv]

It makes sense but it's not trivial to discover, since languages separated by about 10,000 years have changed enough that it's almost impossible to find similarities. This paper talks about separations from 12-24 thousand years.

Proto-Afro Asiatic (edit: ancestor of Coptic, Hebrew, Amhara, Arabic, Tigre, the Berber languages, Aramaic, Hausa (IIRC), and a number of less commonly spoken languages) likely dates back to more than 10K years ago, but we have the benefit of more than 5,000 years of writing in both Egyptian and the Akkadian. We don't have that benefit with any of the indigenous languages of Siberia or North America.

Also, linguistic typology is not the best way to show relatedness, since languages change. For example, proto-Indo European had Nominative Accusative alignment, and was highly inflected, but modern English is almost completely uninflected, and Hindi has Ergative-Absolutive alignment for some tenses.

I looked at the structural features they mentioned, and at least some of them are well known to change quite dramatically. (Some Indo European languages picked up a difference between exclusive and inclusive we, places of articulation change, languages develop and lose gendered nouns, others have lost some distinctions of number (dual has been lost in most IE languages).

But I'm not a linguist, so any linguist can comment on whether these are fair criticisms.

Also, as others mentioned Na-Dene (Navajo is the most spoken language in this group), is already believed to have a connection to the Ket language of Siberia.

[thaumasiotes]

> but modern English is almost completely uninflected

That really depends what you're counting; if you ask which words in a sentence must be inflected, English still looks highly inflected. If you compare it to other inflected languages, the number of distinct forms for any given word is low, but this doesn't do much to help learners coming from languages that don't observe the same distinctions that require inflection in English. They don't do a lot better choosing between the fivish forms of an English verb than between the dozens of forms of a Latin verb.

> dual has been lost in most IE languages

This is a good tangential example; the dual persists, in reduced form, in modern English! We don't inflect verbs for it. But we do inflect determiners; we scrupulously distinguish both from all in a manner that most people in the world find confusing.