Is most of the world's languages being in a small number of families a recent development? Or has there always been cycles of linguistic expansion?

Question

I've always found it curious that the languages spoken by an overwhelming proportion of the human population can be traced to a small number of proto-languages that were each spoken by only a small population of people, at some point circa a few thousands years ago.

For example, the Indo-European languages all descend from the language of the proto-Indo-European tribes in central Asia/eastern Europe, Chinese and all the Sino-Tibetan languages descend from the language of a relatively small group of Sino-Tibetans, and the direct ancestor of all the Austronesian languages was at one point confined to Taiwan.

But it is not the case that modern day Indo-European/Sino-Tibetan/Austronesian speaking people are all descended from the proto Indo-Europeans/proto-Sino-Tibetans/ancient Taiwanese etc. It seems that what happened was that throughout the world, over the past few thousand years, a select group of "linguistically successful" peoples (e.g. the proto-Indo Europeans, proto Sino-Tibetans) expanded to large areas via migration and conquest, and absorbed and intermarried with the local peoples, eventually replacing the local languages with their own. (In Europe this seemed to have happened by the time of the Roman Empire. In China this was happening as recently as the first few centuries AD, as the Chinese dynasties conquered and assimilate the various non-Sino-Tibetan speaking peoples in southern China.)

So the question is: is this a recent phenomenon that only emerged in the past few millennia, or has this probably happened many times before? For example, if we had a time machine and traveled 10,000 years to the past, would we see the same pattern as we see today, with several large language families (each of them descended from a proto-language from ~15,000 years ago)? Or will instead we just see lots of small families and language isolates? (I.e. in terms of the linguistic map, would the entire inhabited world, tens of thousands of years ago, just look like Papua New Guinea today?)

And for example, could many of the so called Pre-Indo-European_languages actually be related to each other, being descended from an even earlier wave of migration/conquest out of somewhere?

PS: I understand this question is highly speculative. Perhaps academic linguistics has little to say on the matter. If that's the case 1) I'd still like to hear from an expert why that's the case, and 2) please share your speculations, whatever they may be.

Answer 1

Probably yes, with a proper definition of "recent".

Daniel Nettle argues in his paper Is the rate of linguistic change constant that small speech communities have a higher rate of linguistic change than large ones (and for the purpose of the paper, 50 000 speakers is already large). Thus, there are two different modes in the evolution of language families: When the speech communities are small, we see a lot of unrelated language families, but when the speech communities are large enough to slow down language evolution we see a few large language families.

So starting with the neolithic era we see larger civilisations emerge in different places of the earth that have larger speech communities. Those larger speech communities start to absorb a lot of smaller ones, leaving fewer and larger language families in the areas where they expand.

So, the proper definition of recent is something like in the past 10k years which is a short time compared to at least 300k years marking the homo sapiens/Neanderthaler split where language was available to humans.

Related question: When were there the most languages