Reply to Ghiselin (long)

[Larry Trask]

----------------------------Original message----------------------------
 
The question of whether Old English and modern English, or classical
Greek and modern Greek, are the `same' or `different' languages has, I
think, not commonly been seen as an issue of principle in linguistics.
Indeed, this strikes me as reminiscent of that Greek philosophical
query about whether one can step in the same river twice.
 
For certain purposes, it is convenient to regard the various
historical stages of English as "the same" -- for example, in writing
a single continuous history of English.  For other purposes, it is
convenient to regard them as different -- for example in contrasting
earlier and later English.
 
If pressed for a response, however, I think most linguists would
prefer the response `different'.  Spoken Latin has broken up into
several quite distinct languages, such as Italian, French and
Romanian.  Since presumably nobody would want to say that these are
the same languages as one another, it is impossible to claim that any
one of them is the same language as Latin.  But the relation of modern
English to Old English is the same as that of Italian to Latin, save
only for the point that modern English is the *sole* descendant of Old
English.  So, if we tried to maintain that modern English was the same
language as Old English on this ground, we would have to abandon this
claim if we discovered that Old English had given rise to a second,
distinct, descendant previously overlooked -- which seems a very odd
outcome.
 
Anyway, our recognition of a single modern language called `English'
is largely an artefact, deriving from political and social factors,
and above all the widespread recognition of a single standard form.
Several centuries ago, the speech of the Scottish Lowlands, which was
largely incomprehensible to Englishmen, was well on the way to
acquiring its own quite distinct standard literary form.  But the Act
of Union, a political event, put paid to this, as Englishmen and
English influence poured into Scotland, and the Scots eventually
abandoned their own standards and accepted the standard English of
England as their standard.  So, without the Act of Union, we might
have had two languages, not one.  Even today, a few Scottish scholars
prefer to regard Scots as a language distinct from English, but this a
minority view.
 
We have our own counterpart to ring species, in the form of dialect
continua.  In a typical dialect continuum, the local speech just
changes slowly and gradually as you travel across the country, but the
differences accumulate to the point of mutual incomprehensibility and
beyond.  Hence *anybody* in the continuum can talk easily to his near
neighbors, with more difficulty to people farther away, and not at all
to people still farther away.  Consequently, there is no principled
basis for deciding how many distinct languages we are looking at, or
where we should put the boundaries.  In practice, the "solutions",
where there are any, usually come from non-linguistic factors, most
often political factors.
 
For example, much of the Low Countries, Alsace, Germany, Austria, much
of Switzerland and part of Italy are all covered by a single Germanic
dialect continuum.  Speakers who live close together can understand
each other regardless of any political boundaries, but speakers far
apart cannot understand each other at all.  It is not just the
citizens of Zurich and Amsterdam who cannot understand each other's
mother tongue: the citizens of Bonn and Berlin cannot understand each
other's mother tongue either.  So how many different languages are we
looking at?  There is no linguistically principled answer.
 
But there is a political one.  Citizens of Germany believe they are
speaking German, while citizens of the Netherlands believe they are
speaking Dutch.  This is so even though people living on both sides of
the Dutch-German border can understand each other easily, while
neither can understand speech varieties which are "dialects" of his
own language spoken farther away.  And, of course, this perception is
now reinforced by mass education and the mass media: everybody in the
Netherlands learns in school a single variety called standard Dutch,
while everybody in Germany learns in school a quite different variety
called standard German.
 
Indeed, the very existence of these two standards is a political
accident, resulting from the early political separation of the Low
Countries from the rest of the Germanic-speaking world.  If the
Netherlands had been absorbed politically into Germany, then standard
Dutch would probably not exist, and the speech of Amsterdam would be
regarded as merely another regional variety of German.
 
What about the other countries?  Well, the Swiss, the Austrians and
the South Tyrolean Italians decided some time ago that they too speak
German, and they have accepted standard German as their own standard.
The Luxembourgers, after much vacillation, are now seemingly
concluding that they do not speak German, and they are making efforts
to construct a standard form of their own Letzebuergesch speech and to
recognize this as a national language.  The Alsatians in France, with
long-standing political grievances against Germany, have apparently
decided that they do not speak German, but a different language called
Alsatian.  However, if Germany had succeeded in her repeated efforts
to annex Alsace, the outcome would be different.
 
The Germanic-speaking Belgians have changed their minds during my
lifetime.  Formerly, they maintained that they did not speak Dutch,
but a different language called Flemish, and they made efforts to
establish a standard Flemish language distinct from standard Dutch.
But, some years ago, they gave up on this, and agreed that they too
speak Dutch, so they now accept standard Dutch as their own standard.
(But note that the speech of West Flanders is just as incomprehensible
to people in Antwerp as it is to people in Amsterdam.)
 
This sort of thing is pretty much the norm with languages.  Individual
languages just do not exist "out there" as a general rule; instead,
they are imposed *ex post facto* by non-linguistic means, especially
political ones, today commonly reinforced by education and mass media
(largely absent in the past).
 
As my Germanic example shows, mutual intelligibility has little to do
with it.  German is not a single language because all German-speakers
can understand one another -- they can't, if they use their mother
tongues.  Rather, German is a single language because its speakers
have decided that it is.
 
Anyway, mutual comprehensibility is not an either/or matter, but a
matter of degree, and moreover it can easily change with exposure.
Speakers of Basque from different parts of the country may have great
difficulty understanding one another at first exposure, but, with a
little practice, they quickly get used to one another's speech.
 
Indeed, I've had this experience myself.  I'm an American in England,
and, the first time I met a Geordie -- a speaker from
Newcastle-upon-Tyne -- I could not understand a single word the man
was saying.  I did not even believe he was speaking English.  But,
after a few days, my ear got attuned, and after that I was able to
understand him without much difficulty.
 
But there are limits to this.  Speakers from Zurich, Bonn and Berlin
cannot get used to one another's speech; the varieties are just too
different.  Either some people simply have to learn the other guy's
speech, or (as happens in practice) they just shift to the standard
German they have learned in school, which is the mother tongue of none
of them but which is, of course, bloody convenient.  This is perhaps
not really so different from the case of Basques and Catalans in
Spain, who cannot understand each other's (unrelated) languages and
have to switch to Spanish to communicate.
 
The `cladistic' model of linguistic descent, which we usually call the
`family-tree' or `genetic' model, has indeed been the mainstream view
in linguistics at least since the middle of the 19th century, and it
still is today.  Of course, linguists have been aware of the
complications for just about as long: both the problem of dialect
continua and the problem of contact -- the diffusion of linguistic
features across language boundaries, which appears to be vastly more
frequent than the diffusion of genetic material across species
boundaries in biology.  But the general view is that these
complications, while real, are tractable: that we can deal with them
by merely imposing some complications upon our basic family-tree
model.
 
Of course, there have always been linguists who took issue with this.
The 19th-century dialectologists often disliked the family-tree model,
which they saw as excessively idealized and not a good picture of
reality.  And such figures as Hugo Schuchardt (in the 19th century)
and Bob Le Page and Charles-James Bailey (in our own day) have often
protested against the family-tree model and drawn attention to the
importance of what we call `convergence phenomena'.  All these have
perhaps been widely perceived as intellectual gadflies, what the
Russians call `hooligans', stirring up trouble by over-emphasizing
interesting but peripheral curiosities.
 
But times may be changing.  For a long time now, we have been forced
to accept the reality of one type of what we call `non-genetic
languages': creoles, which descend from pidgins, which themselves are
not natural languages at all.  Fine.  But, in the last ten years or
so, the essential validity of the `genetic' or `family-tree' model has
been coming under sttack.
 
Thomason and Kaufman's 1988 book Language Change, Creolization, and
Genetic Linguistics argued that non-genetic languages were a real
possibility, at least in certain circumstances.  (A non-genetic
language is one that does not descend from a single ancestor in the
familiar way: either it has two or more direct ancestors, or it has no
direct ancestor which is a natural language.)  T&K were in fact very
cautious in their claims, but not everyone since then has been quite
so cautious.
 
In the last few years, a number of very striking cases of non-genetic
languages and possible non-genetic languages have been reported.
Mixed languages (languages descended from a mixture of two or more
existing languages), mooted about since the 19th century, have finally
been securely identified, the best example being Michif in North
America.  New and dramatic contact phenomena like metatypy (extreme
structural borrowing) have been identified and named.  Jeff Leer has
proposed that Tlingit, which has some odd characteristics, might be a
`portmanteau language', derived from a meld of several related but
quite distinct speech varieties.  Bob Dixon, in his recent book The
Rise and Fall of Languages (which has had a mixed reception), argues
that massive diffusion of linguistic features across language
boundaries is in fact the norm, and hence that family trees cannot
even be constructed, except in certain special circumstances when
divergence becomes temporarily more important than divergence.  Even
Thomason and Kaufman raised the possibility of `abrupt creoles',
creoles arising directly from contact with no intervening pidgins.
 
Many historical linguists remain unimpressed by this flurry of
activity, but others are enthusiastic.  Undoubtedly these new ideas
are still largely at the stage of waving our arms at some surprising
data and of tossing new ideas around excitedly.  Not much has as yet
coalesced out of this activity.  But it *may* be that the next
generation of linguists will come to accept convergence phenomena as
being at least as important as the more familiar divergence, and that
our models of linguistic descent may have to be revised accordingly.
Then again, maybe not.  Anyway, these are lively times in historical
linguistics, and I predict a lot of bitter arguments in my field in
the next twenty years or so.  This is perhaps not the ideal time to
ask us linguists what we think about models of linguistic descent.
Twenty years ago, virtually all of us would have given the same
answer.  Today, not so.
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK
 
larryt at cogs.susx.ac.uk