One Step at a Time

[Larry Trask]

--On Thursday, July 5, 2001 3:47 pm +0000 X99Lynx at aol.com wrote:

[LT]

> << OK, then -- fire away.  But you are defending a position which is
> wildly, ludicrously false -- not an easy task. ;-) >>

> Typically, your conclusions come first.  You don't have to wait for the
> evidence.

No.  The (snipped) position which Steve wants to defend is that the
comparative method can produce proto-languages which never existed.
This position is false.  Everybody in linguistics knows it's false and
understands why it's false.  It's only Steve who fails to realize this, and
who continues to maintain that a blatant falsehood can be true.

[LT]

> <<No.  This is fantasy.  The comparative method does not "assume" any
> number of parents at all.  Steve, when we attempt to apply the
> comparative method to some linguistic data, we *do not* "assume* in
> advance that we must be looking at a single parent....  If there *was* a
> single parent, then the method will tell us about it.  Otherwise, the
> method gives us only a nil return...

[LT on Steve's scenario]

> Now suppose we tried to apply the comparative method to the resulting
> collection of languages -- "collection", because this assembly would
> *not* be a family, as we use that term in linguistics.  [SL - The
> conclusion comes first, of course.]

No, Steve.  Will you please stop it?

In linguistics, a family of languages is, *by definition*, a group of
languages which are descended by divergence from a single common ancestor.
Languages which are so related constitute a family.  Languages which are
not so related do not constitute a family.  English and French do not
constitute a family merely because almost the entire abstract vocabulary of
English is borrowed from French.  Hebrew and Yiddish do not constitute a
family merely because they are both spoken by Jews.  Japanese and Chinese
do not constitute a family merely because they are both written in Chinese
characters.  There are all sorts of interesting ways in which languages can
be related, but descent by divergence from a single common ancestor is the
only one that makes the languages in question a family.

[LT]

> What would happen?
> Suppose,... that Basque and Spanish were to interact in just such a way,
> and give rise to half-a-dozen languages, each consisting of a different
> mixture of Basque elements and Spanish elements, a couple of thousand
> years later...., the method would once again give a nil return.  We could
> note the presence of many common elements in the languages under
> investigation, but we could not find the required systematic
> correspondences, and so we could reconstruct nothing. >>

> Let's just start this with a basic question, to be sure you are saying
> what you appear to be saying.

> If you assume Basque and Spanish are entirely lost and unrecorded in your
> example above, and all that is being compared are the "half-a-dozen
> languages" you mention...

Yes; exactly.

> Are you saying there will be no systematic correspondences to be found
> among those "half-a-dozen languages"?

Yes; that's what I'm saying.  Or, to be more precise, there will be no
systematic correspondences pointing to a common ancestor.  There will be
all sorts of seemingly common elements, seemingly connected in strange and
puzzling ways, but those common elements will not pattern in the way that
we call 'systematic correspondences'.  If we try to apply the comparative
method to the data, we will get some exceedingly strange results, but we
will not get any proto-language at all.  We cannot possibly get a
proto-language, because there wasn't one.

In any case, I regard this scenario as academic in the extreme.  There is
no known case of a whole collection of mixed languages arising by contact
between speakers of two languages, with different mixtures being settled on
by different groups of bilingual speakers.  And I know of no reason to
suppose that such a scenario is even possible, let alone plausible.

Steve, why are you so exercised about a scenario which never happens?

Anyway, let me try once more.

Descent by divergence works as follows.  A single language P is spoken in
some community.  Over time, P breaks up into what are at first regional
dialects but which eventually diverge into quite distinct languages, the
daughter languages of P.  When (as usual) P is not recorded, but some of
the daughter languages are, we can apply the comparative method to these
daughters, find systematic correspondences, work backwards, and reconstruct
the unrecorded ancestor P in some reasonable degree of detail (not in every
detail, of course).

In other words, the comparative method is a tool for rewinding divergence
from a single common ancestor.  It does this extremely well, but note: this
is the *only* thing the comparative method does, or can do.  It cannot do
anything else.  If the languages being examined are not in fact descended
by divergence from a single common ancestor, then the comparative method is
helpless: it can do nothing, and it can produce no result.

In particular, the method cannot conjure up an ancestral language which
never existed.  Steve, I wish you would get this undoubted fact into your
head.

Note in particular that the comparative method deals only with the
consequences of divergence.  It has nothing to say about convergence, and
it cannot be applied to instances of convergence -- which is precisely what
Steve is trying to do.  The comparative method can rewind divergence, but
it cannot rewind convergence.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk

Tel: (01273)-678693 (from UK); +44-1273-678693 (from abroad)
Fax: (01273)-671320 (from UK); +44-1273-671320 (from abroad)