There's some evidence for it. Ruhlen's book The Origin of Language makes the case decently, but also admits that the evidence is pretty much all in--the world's modern and historic languages have been poked into the computer, the computer has crunched--and the picture is still ambiguous. Thus, we are unlikely to ever know for sure. Some information simply does not exist anymore.
Exodus as an echo of something tens of thousands of years before? Pass the bong! I like the idea that the pillar of fire and the darkness are from the explosion of Thera, which also figures to have caused the Atlantis legend.
For sure, the data he quotes is from Ruhlen's book.
Greenberg was, IMO, one of the greatest scientists of the last century.
but also admits that the evidence is pretty much all in--the world's modern and historic languages have been poked into the computer, the computer has crunched--and the picture is still ambiguous.
Ruhlen also says that the 27 etymologies he presents are the tip of the iceberg. I am personally optimistic that enough will be done to demonstrate monogenesis, but I don't expect to reconstruct the pronouns and declensions and so forth of "Proto-World". Maybe the phonology.
Actually, it's kinda like tracing multicellular life back to unicellulars - because lateral gene transfer is common among single-cells, there's really no tree when you go back that far. It's quite possible that there never was a single ur-speak, just a lot languages and dialects and multilingualism. But like life, there would be a single tree after that.
Part of the reason I'm optimistic is the bottleneck in human population something like 70K years ago. If they've already found cognates between Khoisan and Australian, and the Aboriginals have been there at least 40K years, how long ago was a language that contained the ancestors of the cognates spoken? Seems like we're more than 1/2 way there.