IE "Urheimat" and evidence from Uralic linguistics

[Robert Whiting]

On Sat, 5 Feb 2000 petegray <petegray at btinternet.com> wrote:

RW>No.  Relationship is an absolute.  ....
RW>Genetically related languages were once the same language.

>Sorry, Bob, I can't agree, and I suspect you're in a minority
>these days (though I may be wrong!).

No problem -- differences of opinion are what make for horse
races and book reviews.  But I'm not entirely sure what you
are disagreeing with here:  the restriction of related to
"genetically related" or the definition of genetic relatedness.

I find it difficult to believe (although that doesn't mean that
it isn't true) that the definition of genetic relationship that
was the cornerstone of historical linguistics for over 200 years
-- from Jones' 1786 "sprung from some common source" to Anttila's
1989 "'Related' is a technical term ... meaning that the items
were once identical" -- has been dropped in the last 10 years and
replaced with something like 'related languages are those that
have some features in common or are somehow connected', or 'there
is no such thing as genetically related languages'.

I note that David Crystal in his _Cambridge Encyclopedia of
Language_ (1987) has the following entry in the glossary (p. 429:
"related (hist)  Said of languages or forms that share a common
origin."  It will be interesting to check the second edition to
see if he has switched over to what you suggest is now the
majority position:  that related languages do not have to have a
common origin.

>(a) Unrelated languages do produce offspring - for example,
>creoles.  It has even been suggested that the entire Germanic
>branch of IE is in fact a creole.

Yes, I've even heard it suggested that IE was a creole, but I
don't think that such theories have many adherents (doesn't mean
that they are wrong, just unlikely).  But creoles are a different
animal and one that is not yet well understood despite intensive
study.  I think that the mainstream view is something like the
following:

     Creoles develop from pidgins; pidgins are not natural
     languages (have no native speakers), but auxillary languages
     used for communication between speakers of different
     (usually typologically widely divergent) languages, intended
     for limited purposes such as trade; pidgins have limited
     lexicons and minimal morphology and syntax (essentially they
     are mini isolating, bare-bones, no-frills languages);
     pidgins are often (but not necessarily) based on one
     language (usually the socially dominant one) but with some
     elements taken from other language(s); pidgins often die out
     when the need for communication between the groups ceases or
     with the development of bilingualism or the assimilation of
     one of the language groups into the other.

     Creoles arise when a pidgin becomes the mother tongue of a
     group of speakers (presumably the children of a group that
     communicates primarily in pidgin hear only the pidgin and
     begin expanding it to provide some of the syntactic features
     that have been stripped from the original language(s) to
     create the pidgin); this expanded pidgin becomes the native
     language of the next generation and continues to expand to
     provide all the syntactic features that are necessary to
     normal communication; the creole is once again a natural
     language.

Pidgins and creoles are thus two stages of a single process.
Many pidgins never become creoles, either dying out when no
longer needed or simply continuing in use as pidgins.  But I
don't know that creoles arise other than out of pidgins (it
wouldn't bother me to learn otherwise, however).

So the steps involved in the creation of creoles actually form a
cycle of contraction and expansion (natural language --> pidgin
--> creole [natural language]).  The thing is that the creole can
arise very quickly from the pidgin (within a generation or two)
and the creole is usually not easily intelligble to the native
speakers of the language(s) on which the pidgin is based.  This
is in contrast to the normal development of natural languages
where mutual comprehensibility is usually preserved over at least
3 generations (children may think that gramps uses a lot of
'quaint' expression and gramps may think that the youngsters
aren't being taught the language properly, but they can still
communicate easily).

This, I think, is one of the fascinations that creoles hold for
current research.  The accelerated pace of change in creoles is
is distincltly different from non-creolized natural languages and
thus forms sort of a laboratory for studying language change.

>I think it is unhelpful to restrict our understanding of
>relationship to a yes-no either/or. You might have trouble
>describing a creole without distorting facts to fit your
>definition.

I don't think so.  I have just described what I think is
considered the mainstream view of creoles without having to use
the term 'related'.  One doesn't have to distort facts to fit a
definition.  If the definition doesn't apply to the facts, then
one just doesn't use it.  It is of course, easier to distort the
definition to fit the facts.  Facts exists in nature; definitions
are arbitrary conventions agreeded upon by a speech community.
And definitions are subject to change either by the discovery of
new facts or by an agreement to amend the convention.  But if you
change definitions unilaterally, you run the risk of not being
able to communicate with your audience.  You can redefine "roast
beef" to mean "spinach quiche" and "file cabinet" to mean
"kitchen sink", if you want to.  But if you invite people over
for roast beef they may be surprised (and perhaps even dismayed
if they are real carnivores) when they are served spinach quiche.
And the evening may take a disastrous turn if you tell them to
just put the dirty dishes in the file cabinet.

But I don't think that the linguistic awareness of creoles alters
the facts of genetic relationship.  The various parts that form
a pidgin are no more related than the bits and pieces of Dr.
Frankenstein's monster.  The pidgin is not the same as the
language on which it is primarily based; it is a severely
truncated form -- a mere stump.  Since this does not fit the
definition of 'related' as used as a technical term in historical
linguistics, I see no point in using the term.  For example, I
would describe Tok Pisin as an English-based pidgin influenced by
native Papuan languages that has been creolized in some areas.
There is no need to use the term 'related' and I don't think the
facts have been distorted.

The creole then develops out of the pidgin, but it is not, as a
creole, identical to the pidgin nor is it ever identical to the
language(s) on which the pidgin is based.  So if one continues
to use 'related', 'pidgin', and 'creole' as techinical terms,
then they each have their specific meanings and it is not
necessary to explain the "relationship" between natural
languages, pidgins, and creoles.  One only needs to start talking
about 'related' with references to possible daughters of the
creole.

There are, of course, some gray and muddled areas (as there
always are).  For instance, if a pidgin is in use over a wide
area, what happens if it is creolized more than once?  Are the
different creoles genetically related, having once been the same
language?  Depending on the circumstatnces, it is quite possible
that the two (or more) creoles from a given pidgin are not
mutually comprehensible.  Thus one would have cognate languages
that are not mutually intelligible within a generation, not a
usual occurrence.

>It is ultimately only a matter of which method of description we
>prefer, but I do believe it is unhelpful to restrict the term
>"related" to mean "genetically related".

I can't see why.  If you use 'related' in its dictionary sense of
"connected, linked, affiliated" rather than its historical
linguistics technical sense of "genetically related", then its
meaning becomes so diffuse that you always have to explain how you
mean 'related' so you might as well cut out the middleman and go
straight to a description of the 'relationship' and there is
really no point in using the term at all.  Then it is quite true
as Hans J. Holm says: "'Relationship' is _always and only_ a
question of degrees and ways."  The term has no specific meaning
beyond implying some kind of connection, however vague, so each
time you use 'related' or 'relationship' you have to define what
you mean.  I don't see what makes this so much more helpful.

If you allow 'related' to mean "has some connection", then
English is 'related' to Arabic, Hebrew, Chinese, Japanese,
Korean, Turkish, Etruscan, Finnish, Hungarian, Malay, Mayan,
Nahuatl, Sumerian, Swahili, Kaffir, Algonkian, Hawaiian, and
hundreds if not thousands of other languages from which English
with its capacity for swallowing foreign words whole has
appropriated words for its own use.  Similarly, it could be said
that most of the world's 6000 (give or take a few thousand) or so
languages are 'related' to English if they have at least one
English loanword.  'Related' has become more or less meaningless
in any linguistically interesting way.

Using 'related' in its non-technical sense would allow you to say
that English and Chinese are related (because there are a number
of Chinese loanwords in English; ketchup, gung ho, yen ['desire'],
and chow come to mind off the top of my head).  But almost any
historical linguist would take exception to a statement that
English is related to Chinese.  A historical linguist would say
that English and Chinese are not related.  And another historical
linguist would know, through the technical vocabulary of
historical linguistics, that this statement means that English
and Chinese were never the same language at any level that has so
far been uncovered by the methodology currently available.

>Genetically (in your terms), English is equally related to both
>French and Italian.

They're not my terms; I didn't invent them -- they are standard
in historical linguistics textbooks.  But I agree with the
premise -- except that I wouldn't say "equally related"; I would
say "related at the same level."

>I find it more helpful to accept a wider use of "related" in such
>a way that it allows me to indicate that plural forms and a range
>of other stuff in English actually are "related" to French but
>not "related" to Italian, and that therefore English has a
>different relationship to French and Italian, not an identical
>one

I can see what you are talking about (although I must confess
that I don't know what you mean by "related" plural forms).  But
it seems to me that this is more a problem of lack of terminology
to explain the situation rather than a limitation in the meaning
of 'related'.  I would simply say that English is related to both
French and Italian at the same level but that English and French
have (re)converged extensively due to lengthy periods of contact
with French as both a superstrate and an adstrate language
resulting in more similarites between English and French than
between English and Italian.

So I don't see the problem as one of limitation imposed by the
technical meaning of 'related' in historical linguistics, but
rather as a reluctance to use the vocabularly of historical
linguistics to describe the situation.  English and French have
converged as a result of extensive borrowing of Frech lexicon by
English due to contact; English and Italian have not.  This
does not mean that English is more closely "related" to French
than to Italian.  Languages that are in contact will tend to
converge.  But convergence does not make the languages related.
Languages do not become related (cognate).  Languages either
start out related (cognate) or they will never be.  Languages are
either related (cognate) or they are not.

The fact that we may not be able to demonstrate relatedness does
not change this.  Our perceptions of reality do not change
reality.  If everyone believes that the world is flat that still
doesn't make it possible to walk to the edge and jump off.

Of course one can say that we should just dump the technical
meaning of 'related' and let 'cognate' carry the load.  'Cognate'
is perhaps better suited to this because it is etymologically
more transparent.  But the problem is that this needs to be done
across the board.  Mixing terminology just cofuses those who are
not trained in historical linguistics.  When different people use
the same word in different ways it leads to confusion if not
chaos.  Thus we have these interminable discussions about
terminology ("ungoing discussion" as posted [unintentionally I
presume] by Eduard Selleslagh on Feb 8 gets my vote as typo of
the century [so far] :>) that simply lead us around in circles
while the non-linguists who monitor the list conclude that
historical linguists don't know what they're talking about.

>(b) The idea that there must be a single language progenitor of
>daughter languages is widely disputed. Some people accept the
>idea that a collection of interrelated languages might never have
>had a single ancestor, but as far back as you care to go were
>simply a collection of inter-related languages.  The
>language/dialect issue comes up here.	We talk of IE "dialects"
>within PIE, but this is simply terminology.  The point is that
>there is no need whatever for there to have been a single unified
>PIE language.

There is quite a bit of material for thought in this paragraph.
Enough for several dissertations at least.  But fortunately, most
of these dissertations have already been written.

Yes, there are many people who dispute the existence of
protolanguages.  I have seen some of their web sites.  The
problem is that no one has yet come up with an explanation that
fits the data better.

Besides, if there are no protolanguages, then the comparative
method, the basic tool of historical linguistics, is useless.
Since you seem to be trying to tell me that the basic hypothesis
underlying historical linguistics (languages that have more
similarities than can be accounted for by chance or borrowing
were once the same language) is not true, and that the basic tool
used in historical linguistics (by comparing similar forms in
related languages one can reconstruct the probable form of the
common ancestor -- aka the comparative method) does not work
because there is no common ancestor, I get the impression that
you are trying to tell me that historical linguistics is a hoax
perpetrated on the academic community by a bunch of Germans in
the early 19th century (just joking, of course, but it does
start to sound that way).

The idea of infinitely converging languages instead of a
proto-langauge sounds like Trubetskoy to me.  This simply does
not take into account the overwhelming amount of detail with
which PIE can be reconstructed.  Much too detailed to be
accounted for with a loose federation of languages.  Now
individual IEists may not agree on the details, but that is due
to the wealth of data that is available for interpretation, not
on its scarcity.  The IE hypothesis (an explanation of observed
data) is inductive and thus cannot be proved directly.	Rather
proof of an inductive hypothesis comes from falsifying the
alternative hypotheses.  The detail with which the nominal,
pronominal, and verbal systems, as well of the syntax, of PIE
can be reconstructed do not prove that there was a PIE language.
Rather, they make it inconceivable that there was not a PIE
language.

On the other hand, a group of languages with similar features
does not have to be a language family.	Discussion continues over
whether Altaic is actually a family or a group of languages
connected by areal features and convergence through longstandin
contact.  The eventual outcome of this discussion, however, does
not affect the case for PIE.

Yes we talk of dialects within PIE.  But this is not simply a
matter of terminology.	PIE was a modern language much like any
modern language known today.  Any language that exists over a
sufficient period of time will develop dialects.  Since modern
languages have dialects we assume that PIE had dialects.  All
modern languages display variants in some forms.  We assume that
PIE had variant forms.	All languages that exist over a
sufficient period of time will change both from internal causes
and from contact with other languages.	We assume that PIE
changed with time.  PIE is a reconstruction of the state of this
language just before its first split, based the forms found in
its daughter languages.

The stage of the language before this we call pre-PIE (since we
don't know enough to be able to divide pre-PIE into old, middle,
and new phases, we call everything back to the point where
pre-PIE split off from whatever its ancestor may have been
pre-PIE).  Pre-PIE forms must be found by internal reconstruction
since the comparative method only works back to PIE.

Despite the different names, the comparative method and internal
reconstruction are essentially the same thing.	They have
different inputs and give different results, but the underlying
principle is the same:	similar forms that are in complementary
distribution are likely to be different aspects or outcomes of
the same thing.

Perhaps the most spectacular use of internal reconstruction was
de Saussure's reconstruction, in 1879, of what are now called
laryngeals for pre-PIE based on the reconstructed forms of PIE.
This reconstruction was not widely accepted, among other reasons
because it reconstructed a feature of the pre-protolanguage that
was not preserved in any of the known daughter languages.  With
the discovery that Hittite was IE and that at least one of the
recontstructed "laryngeals" were actually present in the language
in the words and places where predicted, the reconstruction was
considered vindicated and with it the methodology that produced
it.

This then is what any theory that would replace the concept of
PIE has to overcome.  It not only has to explain away the wealth
of detail with which PIE can be reconstructed, but it also has to
explain away the fact that it is possible to make accurate
reconstructions of pre-PIE on the basis of the reconstructed PIE
forms.

Oh, and it is very difficult to wish away the concept of
protolanguages while there is a clear example of the breakup of a
protolanguage into daughters that is entirely recorded in
historic times.

Bob Whiting
whiting at cc.helsinki.fi