Dating the final IE unity

[Richard M. Alderson III]

On Tue, 22 Feb 2000, Steve Long (X99Lynx at aol.com) wrote:

>>> Do the similarities between Latin and Hittite 'leap off the page" as you
>>> say?  (Please recall how long it took for relationship to even be
>>> detected.)

In a message dated 2/22/00 3:08:16 AM, Rich Alderson replied:

>> Knudtson published the Tell-el-Amarna letters in 1902, as I remember, and
>> put forth the claim that Hittite was Indo-European at that time.  Hrozny'
					^^^^^^^^^^^^^^^^^^^^^^^^^^
>> demonstrated the IE-ness of Hittite in his 1917 monograph to the satisfac-
>> tion of the general IEist populace.  How long did you think it took?

> Did it take 15 years for the similarities to "leap off the page?"  That is
> still rather slow leaping.

No, it apparently took Knudtson looking at it for him to state that this was an
Indo-European language.  I'd call that "leaping off the page".

Hrozny''s book was simply a setting forth of large bodies of evidence for the
claim.  If we date the founding of IE studies to Sir Wm. Jones' little dictum,
but the *scientific* study of the question to Rask, Hrozny' beats Rask by about
15 years.

> (And it appears that laryngeals are still more hesitant about any such
> acrobatics.)

No linguist who ever examined with an open mind the evidence set forth by
Saussure in 1868, and that of Hittite, ever doubted that Saussure was correct.
I can't speak for what *close-minded* folks might have done.

> What are the chances that the gap between Hittite and Classic IE languages is
> actually much older?

[ than 500 years or so, as stated by several posters, myself included ]

What are the chances?  Not good.  Not zero, but not good.

Yes, we have the example of the extreme conservatism of Old Lithuanian, but
that is so unusual that we *do* make note of it.  Historical linguistics is,
after all, a *historical* science, and so must make assumptions about such
things as rates of change based on *historical* obversation of similar occur-
rences--the development of the Romance languages from Latin, the development of
modern English from Anglo-Saxon, and so on.

In order for Hittite (and the other Anatolian languages) to have diverged from
the rest of the Indo-European languages very much more than 1000 years earlier
(my own outside estimate), we would have to reject the evidence for rates of
change provided by all the historical obversations we can make and instead say
that Old Lithuanian is the expected result, and *every* *single* *other* *IE*
*language* underwent accelerated development.

An archaelogist may be willing to do that; I, and I think most if not all of my
colleagues as well, will not.

> If Hittite were separated from Sanskrit-Latin-Mycenaean by an additional 2000
> years, how would the comparison be different than it is now?  What would one
> expect in the comparison to change if in fact Hittite separated 1500 or 1000
> years earlier?  If you wanted to see what Hittite would have been like if its
> ancestor were a distinct language in 6000-5500BC, how would it reconstruct
> differently?

I'm sorry, but the question is meaningless.  Linguistic change is not determin-
istic; we *cannot* say "these changes must, or should, or will, take place if
languages are separated by X centuries/millennia".  In a historical discipline,
we can only say "similar changes, or similar *kinds of change*, took place over
a period of X <unit of time> in these families or languages under our control,
so we expect that it will have taken a similar amount of time for a group of
languages *not* under our control to have undergone similar changes".

> Does the degree of variance in the reconstructed forms become greater in some
> way?  Do the numbers of retentions or innovations increase?  What changes
> would one expect to reflect the greater effects of a longer time period?

Lexical retention as a measure is of course the usual stalking horse, although
this means not only "words on a list" but occurrence both of free and of bound
morphemes when being done by a non-glottochronologist.  Innovation is of course
simply 1 - (measure of retention), so one only gets the one when one gets the
other.

I've already address the last question above, which answer also covers the next
query:

> If I have been successful in posing this question understandably, then one
> should see the value in considering what the reconstructed proto-Hittite of
> 6000-5500BC would look like.  What would it be missing?  What would it have
> lost?  What additional signs of age should we expect?  This would give us a
> way of saying 'Hittite texts would need to look like this if proto-Hittite
> indeed separated from PIE about 7500 years ago.'  And that would seem to me
> to be of great value.

> Go back to the original post and you'll see that agnis/ignis was being used
> to selectively support the 2000 year separation between those early IE
> languages.

No, it wasn't.  It was being used as an easily accessible example of the large
number of similarities which "leap off the page" to anyone familiar with both
languages.  Nothing selective about it, not in the secret-cabal sense I read
into what you've written.

> My point that this was very convenient for Latin and Sanskrit to be compared
> this way.  And equally inconvenient not to find anything like the same
> similarity in either Greek or Hittite.  If agnis/ignis prove something about
> the degree of differentiation over time, then what does the absence of
> agnis/ignis in other early IE languages prove about time and differentiation?

Absolutely nothing.  Similar sets from each pair of IE languages can be set up,
as well as sets from larger groups; several have been posted in the last couple
of days.

> (As a matter of fact, I'm surprised that the closeness between agnis/ignis in
> Sanskrit and Latin does not suggest a much more recent date of commonality
> for those words by themselves, without regard to the rest of those two
> languages.

But as linguists, we *don't* disregard the rest of the languages--we leave that
to the people who claim kinship between Basque and Xalxa Mongolian or the like.
If the form of the word in either Latin or Sanskrit were somehow anomalous, we
might suspect contact later than (near) the time of PIE unity, but since both
show all the expected developments for all sounds within the word, there is no
reason to look for zebras in the stable.

> If all the words in Latin and Sanskrit matched like this, you could argue 50
> years separated the two languages.

And if the differences between these two words were the only differences
between them, you might even be correct.  That's why we don't focus on any word
pair in particular to the exclusion of all other data, to see whether such a
claim *would* be justified.

> Lehmann tells me that Sjoberg and Sjoberg showed why words in south Asia like
> 'sun' should be eliminated from the "glottochronological core" precisely
> because they reflected very early and widely borrowed religious vocabulary.
> Forgive me for asking whether agnis/ignis might not fall into the same
> category.)

You're forgiven.  They don't, even if there were such thing as a "glottochrono-
logical core" for them to fall into.  (I agree with the sentiment that we have
to be wary of such possibilities, just deny that anyone can come up with an _a
priori_ list of things we *must* leave out, or in, or whatever.)

								Rich Alderson