Degrees of similarity and early IE.

[JoatSimeon at aol.com]

As the saying goes, once may be coincidence, twice may be happenstance, but
the third time, something is going on.

When considering broad questions of IE origins, it's helpful to step back and
take a look at the situation -- the relative similarity -- of the first
attested examples.

Eg., Mycenaean Greek, RV Sanskrit, early Avestan, Latin (particularly the
older pre-Classic) examples, what's attested of early Celtic, the early runic
Germanic, Tocharian, etc.

Then throw in the more secure of the first-generation reconstructed
protolanguages; Proto-Germanic, Proto-Slavic, and so forth.  Both the latter
seem to be products of the 1st millenium BCE; the basic characteristic PG
changes are fairly securely datable to the Iron Age and not earlier, and
Slavic somewhat later still.

No two single examples would be conclusive in estimating the time-depth;
after all, if you look at, say, modern Lithuanian in comparison to, say RV
Sanskrit, it makes you blink.  If Lithuanian didn't exist, and we discovered
it via a corpus of written texts buried in caves the way we did Tocharian,
you'd swear it was 2000 years old.  Nobody would dispute that there are
potentially very wide disparities in the speed with which individual
languages _can_ change, given special circumstances, or that the rate of
change can be "jerky" rather than smooth.

But overall, the earliest attested and the most securely reconstructable
early IE languages are of a really startling degree of similarity, and there
are so many of them across such a broad geographical expanse, three-quarters
of the whole width of Eurasia.

Anatolian slightly apart, they seem to be no more diversified than the
Romance languages today; and if you throw in French, Anatolian doesn't look
at that much more so.  In fact, Anatolian is less aberrant than English is
vs. a vs. the other Germanic languages, and we know that they were all
mutually comprehensible only a little over a millenium ago.

Given the extremely broad _geographical_ spread of the IE languages at their
earliest attestation -- from the Atlantic to Chinese Turkistan, which means
pretty well complete linguistic separation -- the high degree of uniformity
between so _many_ widely separated languages virtually forces, I should
think, the hypothesis of relatively recent origin and (by historical
standards) extremely rapid spread over an area previously occupied by many
different languages.

That's certainly the mechanism by which we see similar situations occurring
in historic times -- the spread of Latin and its diversification, the spread
of the (quite unified) Slavic in the early medieval period, the growth of the
Arabic-speaking area, the spread of Chinese, and the spread of English.  If,
every time we have records, Phenomenon A has Cause B, then we're on fairly
sure ground in attributing Cause B when we meet Phenomenon A.

This fits in very neatly with the other evidence -- eg., the technology
expressed in the PIE lexicon -- to bracket the late neolithic as the period
of PIE unity. (Or to be more precise, as the _end_ of the period of PIE
linguistic unity.)  Much earlier, and the degree of geographic spread should
logically and by comparison have resulted in much greater linguistic
diversity.  The hypothesis of recent (within two millenia) spread from a
relatively small area at the time when our first records emerge most
parsimoniously explains the data.

None of this, of course, delivers the degree of conclusivity provided by a
mathematical theorem -- or even the degree of certainty we can use to show
that Latin spread out from a small nuclear area in central Italy.  But it's
about as much certainty as we can expect, given the sparse evidence and huge
stretches of time involved.