Urheimat in Lithuania? (was Re: the Wheel and Dating PIE or NW-IE)

[Robert Whiting]

On Thu, 16 Mar, JoatSimeon at aol.com wrote:

>In a message dated 3/16/00 12:29:13 AM Mountain Standard Time,
>whiting at cc.helsinki.fi writes:

<snip>

[quoting JoatSimeon at aol.com on 13 Mar]
>>>What seems to have happened is that at one point they _were_ in
>>>the "innovative core", but that subsequent to the final breakup
>>>of PIE they became extremely conservative; Baltic more so than
>>>Slavic.

>> The evidence fairly clearly shows that both Baltic and Slavic
>> were transitional areas in both the satemization (palatal
>> assibilation) and RUKI palatization changes that had their core
>> in I-Ir (although RUKI is more generalized in Slavic than in
>> Baltic).  Essentially all this shows is that Baltic and Slavic
>> were still in fairly close contact with I-Ir. in contrast to
>> Greek, Italic, Germanic and points west which were not affected
>> at all by these changes (relic areas).  In short, all the other
>> IE stocks have already broken off or are simply out of range
>> (which is not quite the same thing as they could still be at the
>> other end of a dialect continuum) before this change.  And
>> conservatism after this point indicates that the language was
>> consistently a relic area in further changes originating in the
>> dialect continuum

>-- this seems to be a terminological problem here.  That's
>pretty much what I was trying to say.  Pardon the infelicities of
>expression!

Yes, I would say that this is pretty much a difference of
terminology as I simply tried to restate the proposition in terms
of a wave model.  But there is one difference, which may still be
terminological, but is nonetheless a difference.  You said that
Baltic and Slavic "_were_ in the 'innovative core'", while I said
that they "were still in fairly close contact" with it and were
transitional areas for the particular changes we were talking
about.  Whether the difference between being 'in the core' or
being 'on the edge' of it is significant or not does not seem to
be worth arguing about (sort of like "once you are in the forest,
can you go further in or can you only go out?"), but, as you say,
it's a matter of degree.

>>The example of Lithuanian would seem to argue against this.

>-- it's a matter of degree.  Lithuanian certainly changed far
>less; but it still changed.

Of course it changed.  Change is not just a linguistic universal,
it is a universal universal.  That is to say that it has to do
with the way the universe works -- everything changes.   This is a
by-product of the Second Law of Thermodynamics.  What would be
truly remarkable would be if Lithuanian had not changed at all.

And as for it's being a matter of degree, again, of course it's a
matter of degree.  Anything that isn't a binary opposition
(yes/no, on/off, 0/1) is a matter of degree.  But if we are
talking about a binary branch at a node defined by an innovation
on one branch and a lack of innovation on the other, then it is
not a matter of degree.  It is a binary opposition.  Now what
happens on the individual branches after the bifurcation is beyond
our control.  The less innovative branch may suddenly take it
into its head to innovate like mad without branching again while
the innovating branch may suddenly become very conservative.

Again, we are not saying different things here, we are saying the
same thing differently.  Lithuanian was obviously on the
innovative branch of the node that is defined by palatal
assibilation and RUKI but after this change this branch seems to
have become very conservative (Sanskrit, Slavic, Baltic).  By
contrast, the non-innovative branch of this node subsequently
underwent such extensive innovations as to take any of its
members out of the running for the most archaic surviving IE
language.  This has made the branch that actually innovated the
more conservative by default.  But this status only lasts until
the next bifurcation and then it starts afresh from the new node.
It must be remembered that not only are Sanskrit, Slavic, and
Baltic on the innovating branch of the palatal assibilation node,
but so are Armenian and Albanian, which are not particularly noted
for their conservatism.

Finally, it must be remembered that the changes that define nodes
may be in different areas (phonology, morphology, syntax, or
semantics [lexicon]) and so not every node will affect a given
area.  Lithuanian's claim to great archaism lies mostly in noun
morphology (especially in the number of cases and the forms of
the endings and in a number of particularly archaic looking
lexical items).  There may be other languages that are more
archaic in other areas.

>>But the point is that *something* has to be; if not Lithuanian,
>>then something else.  It is equally against the law of averages
>>for a team or an individual to win a single elimination
>>tournament; but someone always does.  In the IE superbowl,
>>Lithuanian is apparently the winner.  But that still doesn't say
>>anything specific about where it started out geographically.

>-- however, it's not just Lithuanian.  Apart from the other
>Baltic languages, there's the example of Slavic -- which, while
>not quite as conservative, is still notably so.

Again, we are not saying different things.  The node that
produced Lithuanian from the other Baltic languages would have
been in the branch that was less innovative from the previous
node (actually, one of the other Baltic languages, like Old
Prussian or Curonian, may have been even more conservative, but
since little or nothing of them survive it is not possible to
say).  Similarly, the node that produced Baltic on one branch and
presumably Slavic on the other (assuming a Balto-Slavic unity,
which is not universally accepted) would have again been on the
least innovative branch of the next higher node producing the
observed effect of two conservative branches with one slightly
less conservative than the other.  But again (or still) this
doesn't say anything specific about where they started out
geographically except that Baltic was always in close proximity
to Slavic.

>>But is it clear that there is no substratum influence?  There
>>are a number of Balto-Finnic loanwords in Baltic and a
>>considerably larger number of Baltic loanwords in Balto-Finnic.
>>Looks like about what one would expect if Baltic were a
>>superstratum language over Western Finnic.  Just because there
>>wasn't an unknown substratum, as there apparently was in
>>Germanic, that provided a lot of words of unknown origin to the
>>language doesn't mean that there wasn't some substratum.

>-- its a question of degree.  There are certainly substrata in
>Baltic, but not to nearly the same degree as in Germanic.  The
>number of lexical items which can be traced to PIE is
>proportionately much larger and Baltic that in, say, Germanic or
>Greek.

Here, I am less inclined to see it as a matter of degree,
especially as the claim was originally stated "That's where the
absense of substratal influence comes in." If the absence of
substratum influence is important, then surely the presence
of a non-trivial amount of it changes the situation.  But the
presence of more archaic lexical items and fewer obvious
substrate borrowings could equally well be a result of the
refusal of the language to accept loans as of its still being
in its original home.  German, for example, for some considerable
time simply refused to accept loan words or neologisms unless
they blended seamlessly with German phonology, preferring to
calque them instead when absolutely necessary (e.g., Eng.
exposition, Ger. Ausstellung; Eng. rhinoceros, Ger. Nashorn).
English, on the other hand, just gobbles up loanwords and
neologisms regardless of whether they violate English
phonotactics or not (e.g., aardvark, gnu, syzygy).  This is
simply not a function of how close these respective languages are
to their original homelands.

If we didn't have a historical record of the situation, the
argument of conservatism plus lack of substratum influence could,
other things being equal, be used to claim that Iceland is the
original Scandinavian homeland.

>>>There's also the lack of non-Baltic river names and other
>>>features in the area of Baltic speech (and the area where it was
>>>historically attested).

>>Again, useful for indicating that (a) the language was in its
>>original home, (b) the language moved into a previously
>>uninhabited area, (c) the speakers of the language
>>systematically renamed or calqued all river names in their own
>>language, or (d) the speakers of the language drove off or killed
>>all the original inhabitants before they could learn the original
>>names of the topographical features (i.e., no period of
>>linguistic contact).

>-- one has to make a balance of probabilities in these cases.
>Even the case of the Anglo-Saxon conquest of England, which is
>about as complete a case of linguistic replacement as exists in
>the historical record, not all the river and place names were
>changed.  River names in particular seem to persist.

Yes, river names seem to be particularly persistent.  But it must
be kept in mind that, in non-literate societies, linguistic
contact is essential for the transfer of any linguistic
information.  Unless a language has been recorded somehow, that
language exists only in the minds of its speakers.  When the last
speaker of a language dies, unless the language is recorded
somehow, that language is gone forever.  There is no hope of
recovering it through DNA analysis of the remains of its speakers
or archaeological excavation of their material culture.  The only
hope of recovering it is time travel with a camcorder, not yet a
serious option.

Now one way that a language can be recorded (albeit imperfectly)
is through loanwords in a surviving (at least surviving until it
can be recorded) language.  If there are such loanwords that do
not fit the normal patterns of a language and that cannot be
identified as coming from any known language, we can postulate
contact with an otherwise unknown language.  If there are no such
loanwords, however, we can only say that there are no such
loanwords.  Negative evidence cannot be used to construct
specific scenarios.  Negative evidence only means that there is
no evidence.  Negative evidence could mean that there was no such
language, that there was no linguistic contact (bilingual
speakers), or that the second language simply refused to accept
loanwords (or possibly even consciously purged them at a later
date).  Each of these scenarios will be identically represented
in the evidence:  no loanwords.

Making a balance of probabilities is fine so long as you know
what all the probabilities are.  But a balance of probabilities
doesn't tell you what happened, it only tells you what is likely
to have happened.  And if history is "was eigentlich geschah"
that is not quite good enough (although it is done regularly).
Using a balance of probabilities on historical events that have
already happened and claiming that it has to represent what
actually happened is simply a misuse of probability.  Probability
is used to predict the outcome of multiple events in the future
based on the number of possible outcomes and the relative
frequency with which the possible outcomes can occur.  But even
when used correctly for this purpose, probability still doesn't
tell you what is going to happen in a single event, only that the
distribution of the outcomes of multiple events will accord with
the probabilities.  The laws of probability have no memory of
previous events.  Otherwise, the chance of heads with a fair coin
being 1:2, two tosses of the coin would guarantee one head -- and
it ain't so.

If you look at the paramutual odds and see that the favorite is
listed at 1:3 and Beetlebomb is listed at 200:1, that means that
the people who make a living giving odds on horse races think
that the favorite is 600 times more likely to win the race than
Beetlebomb, or expressed another way, if the race were to be run
600 times, the favorite would win 599 times and Beetlebomb would
win once.  But the race isn't going to be run 600 times, it is
only going to be run once.  And if Beetlebomb has a legitimate
chance of winning the race once in 600 times, there is nothing
in the laws of probability that says that that once can't be the
first (and only) time that the race is run.  If it is, then the
rules of probability just say that the next 599 times the race
is run, the favorite should win every time.  So long shots do
come in (sometimes) and the favorite doesn't always win.

So it is with historical events (including historical
linguistics).  There is only one event and it is possible,
depending on the circumstances, for a long shot to come in (cf.
the destruction of the Spanish Armada in 1588).  So in looking at
events for which there is no historical record, saying that it
is likely for a certain event to have had a certain outcome does
not prove that it did have that outcome unless it is impossible
to have had any other outcome.  In eliminative induction only the
impossible can be eliminated.  The merely improbable can still be
the truth.

Now there is nothing wrong with saying that the most likely
outcome is the most likely to have happened (in fact, it is sort
of a truism, or at least a definition of 'most likely').  But it
must be kept in mind that doing this is what is known as a
"purely hypothetical" solution.  The solution relies on the
hypothesis that the most likely outcome is what actually
happened.  But in much of history, and particularly in
prehistory, purely hypothetical solutions are the only ones that
are available.  And in such cases, purely hypothetical solutions
are frequently not marked as such.

So, once again, I don't say that it is an unreasonable
explanation that the Balto-Slavic homeland is very near the
IE Urheimat.  I just say that it is a purely hypothetical one
and therefore not necessarily the correct one.  I don't say it
can't be right, I just say it doesn't have to be right, and as
long as we are agreed on this, then there is no dispute about
the hypothesis.

But I do dispute the contention that linguistic conservatism or
archaism per se is an indication of proximity to the Urheimat.
There are too many counterexamples of conservative and archaic
languages that have moved from their original location and are
now surrounded by heterochthonous languages where the archaism or
conservatism is a result of a phenomenon known as "language
loyalty" that resists the influence of foreign languages (or any
kind of change) in order to retain the speakers' linguistic (and
hence ethno-cultural) identity.  By contrast, the languages that
have "stayed at home" are not subject to similar pressures and
are more free to innovate, to follow linguistic fads, and to
experiment with more expressive modes of speech.

These different pressures are one of the things that contributed
to the "archaizing periphery -- innovative core" model, but it
can be seen here that, unlike wave model of linguistic change,
the core and the periphery are subject to different but
counterbalancing pressures.  The periphery is subjected to
influences from outside languages but tends to resist them
because of language loyalty, while the core is not subjected to
outside linguistic influences but has no need to adopt a
conservative stance in self-defense of its linguistic identity.
The response to these various pressures tends to be governed by
sociological factors (prestige, intragroup bonding, desire for
novelty or for the familiar, etc.) rather than by purely
linguistic ones, so geographical location with respect to the
homeland doesn't have anything specific to say about whether a
language is going to be archaizing or innovative.

Once again, I don't say that the archaic nature of Lithuanian
speaks *against* its being at or close to the Urheimat.  I just
don't think that it necessarily has anything to say *for* it
either.  As Brian Scott pointed out (23 Mar), if we have a tree
of binary nodes (admittedly, an oversimplification), if we follow
the least innovative branch from each node (however we choose to
measure "least innovative", assuming that we can, and even
assuming that we can tell the difference between the archaism and
the innovation) then we will eventually reach a branch that
contains the language that is the most archaic.  We have to.
Similarly, if we always follow the most innovative path we will
reach a branch that contains the most innovative language.  We
have to.  The most innovative and the least innovative languages
*have* to be there, but I don't think that the fact that they are
there means that they *have* to tell us where these languages
started out with relation to the Urheimat.

Bob Whiting
whiting at cc.helsinki.fi