I haven't studied this in detail, but the isolates all seem to be in parts of northern Eurasia and the Middle East which have been pretty thoroughly mapped. In the Americas, Africa, and Australia, and possibly in SE Asia and India, I doubt that we know enough to identify isolates.

Most of which will turn out to be, of course, of Dravidian origin.

Erm, how could a language family be an isolate? If it's a family, then it consists of a number of relatives, which are thereby not isolates.

I understand what he means, of course, but I've never heard the term "isolate" applies to families rather than individual languages. And I find the notion absurd: if you go high enough up the tree, every language family is unrelated to every other. (Unless you believe in Proto-World.)

I've said it before and I've said it again, the only solution is to nuke language isolates from orbit. Then, and only then, will the eldritch influence of the dread Cthulhu be cleansed from the Earth.

"the isolates all seem to be in parts of northern Eurasia and the Middle East which have been pretty thoroughly mapped"

I agree that in much of the world we probably don't know enough to identify isolates, but that certainly hasn't stopped anyone in the Americas, which are absolutely full of "isolates". In Africa, it depends who you ask, but there are certainly people who argue that Hadza (a click language of Tanzania), in particular, and maybe Shabo (in Ethiopia) are isolates.

"if you go high enough up the tree, every language family is unrelated to every other. (Unless you believe in Proto-World.)"

There isn't a single attested case of a spoken language emerging out of nothing in human history (even the one artificial language that has gained a few native speakers, Esperanto, still has a largely Indo-European vocabulary). On the plausible assumption that there haven't been notably more artificial language fans in the past than there are today, and that all human beings descend from a fairly small common ancestral population that spoke a language, it seems quite likely that every language family is related to every other one. Discernibly related - now that's another story.

"it seems quite likely that every language family is related to every other one."

That'll probably never get past the hypothesis stage, since all evidence seems to show otherwise. :) I have no problem with the idea that the world's proto-languages emerged independently 6,000 years ago among their respective tribes and ethnic groups.

Isn't Meroitic generally regarded as having been an isolate? You don't get much more African than that.

"I have no problem with the idea that the world's proto-languages emerged independently 6,000 years ago among their respective tribes and ethnic groups."

Do you really think people were painting, sculpting, building boats, making spear-throwers, living in villages, conducting complex funerary rites, and hunting mammoths to extinction without even having a language to speak to each other in? And that we didn't start using a language until hundreds of thousands of years after the common genetic endowment that allows all of us to learn a language emerged? Or was there a sudden craze for making up new languages 6000 years ago ("OK, everyone, we need to come up with a new way to form the accusative... how about we add an -m to the stem? The ayes have itm - let's use that suffixm from now on.")?

The least implausible way I can think of for the world's spoken languages not to have a single origin would be if the original Homo sapiens expansion only used sign languages, and different populations independently came up with spoken languages to replace their existing sign language in much the way that different villages with large deaf populations have independently come up with sign languages to replace their parents' spoken language. But if that had happened anywhere near as recently as 6000 years ago, you'd expect at least some areas (Australia or the Andaman Islands should easily be isolated enough) to have kept a sign language as their first language despite having hearing populations, and that too is unattested as far as I know.

Nothing new under the sun? It's probably true that all languages are linked. But that doesn't mean they haven't evolved so distantly from the parent as to make them independent. Some of them may choose to leave their families all together with a clean break. The term 'isolates' may be more about severed ties than about the language having been immaculately conceived. I've been inventing my own language for 30 years, incorporating a grammar based on cycles. Though Tapissary has many links to Indo-European languages, I believe its cyclic grammar and pictographic script make it an isolate.

I have no problem with the idea that the world's proto-languages emerged independently [...] among their respective tribes and ethnic groups.

Ockham Billy? Thy razor, please.

6,000 years ago

If you mean 60,000, that is AFAIK barely imaginable.

Are there any non-IE words in Esperanto, apart from proper names?

Last I looked, Wikipedia cited something saying recent that both Hadza and Sandawe are indeed Khoisan.

Then, and only then, will the eldritch influence of the dread Cthulhu be cleansed from the Earth.

Fool! There is no hope -- but to be eaten first.

"Are there any non-IE words in Esperanto, apart from proper names?"

Define "IE words". Would arabric borrowing through Spanish count as IE or non IE?

A less obstructionist answer would be 'some, but not that many'. For historical reasons relating to the time and place the language was worked out and started to be used, IE roots predominate.
Also, due to the mechanics of the language and the linguistic ethos of most users, there's a tendency for new concepts to be expressed by compounds made up of existing roots. New roots are not borrowed that easily.

I've read that at one time that the Japanese borrowing hashio(j)? was generally accepted for chopsticks by IE-speaking esperanto users but that Asian users preferred the compound (mangho)bastonetoj ((eat)-little-sticks). IME bastonetoj is the more commonly used term.