PIE vs. Proto-World (Proto-Language)

[Larry Trask]

Pat Ryan writes:

> I simply do not understand why some find it difficult to understand
> that reconstructing the Proto-Language is only primarily different
> from reconstructing Indo-European in the wider selection of source
> languages for data.

Others have already made substantial points in reply, but I'd like to
add one more.

Before reconstruction can be attempted, there must be a persuasive
*prima facie* case that the languages under consideration are genuinely
related.  Our predecessors did not begin reconstructing PIE until they
had persuaded themselves that the evidence for a genetic link among the
IE languages was too massive to be ignored.

But you can't just pick some languages that catch your eye and then
"reconstruct" a common ancestor for them.  That is, you can't pick, say,
Zulu, Sumerian and Korean and "reconstruct" Proto-ZSK.  You must first
make a good case that a common ancestor for these languages is
preferable to any other scenario.

And the same goes for "Proto-World", or whatever.  Until a persuasive
case has been made that *all* of the world's 6000 or so known languages
are genuinely related, there is no point in attempting a
"reconstruction" of "Proto-World".  The result of such a rash attempt
can be no more than legerdemain.  Of course you can show that this, that
and the other *might* have a common ancestor, but you can do this in
countless entirely different ways, none of them superior to any other,
and you cannot show that these things really *are* related.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk