There are a couple of points in bibliopage’s Staff Report, Many English words have Latin roots, but what are the roots of Latin? that I would take mild exception to.
Thus, bibliophage writes:
This is correct in a very narrow sense, in that we don’t have written records or even a reconstructed proto-language. However, unless we postulate that PIE sprang into being full-blown, like Athene from the forehead of Zeus (or perhaps *Dyeu-p@ter, since we’re talking PIE), or somehow preserved intact the first human language, it must have descended from “something else”. The oldest strata of PIE that we can reliably reconstruct are thought to date from about 4500BCE; human existence (even defining “human” quite narrowly) is some tens of thousands of years older than that.
The question “Did PIE descend from something else?” should probably be interpreted as meaning, “Can we connect PIE with other known languages?” Here, bibliophage writes:
The relationship of PIE to Proto-Uralic is also generally (although not universally) accepted. That PIE speakers were in contact with Proto-Uralic speakers is almost certain (that the Pontic-Caspian region in which the Kurgan culture flourished was part of the PIE homeland is accepted by most, although not all, linguists). Many PIE loans and, even more tellingly, definite Proto-Indo-Iranian loans (showing the /k/->/s/ shift) are found in Uralic languages (e.g., Finnish porsas). However, there is enough unity in the noun and verb inflections that most linguists are willing to postulate a Proto-Indo-Uralic language, although it cannot be reconstructed. As an affinity between Uralic and Altaic is often also recognized (although some linguists question whether there is an “Altaic” family at all, as distinct from a bunch of essentially unrelated languages with contact relationships), a Proto-Indo-Uralo-Altaic language is implied.
bibliophage is correct in stating that the Nostratic hypothesis remains controversial, to say the least.