Fuzzy Language Boundaries
ECOLING at aol.com
ECOLING at aol.com
Fri Dec 10 17:36:48 UTC 1999
The following message was drafted long ago,
though a few minor revisions have been made since.
Two paragraphs at the end were substantially revised
at the request of the moderator.
Mr. Trask has criticized others for using the term "language"
in ways in which he himself has used the term.
When they did, he repeatedly put different words in their mouths.
Even when Steve Long quoted from Trask's book to show this,
he still did not acknowledge that others are using terms the
same ways he is. Examples from Trask today [many days ago] follow...
In a message dated 10/20/99 6:02:33 PM, larryt at cogs.susx.ac.uk writes:
>When a language splits into daughters, there comes a time when we must
>finally speak of distinct languages.
>But one of my points on this list has been that the time when this occurs
>is not, and cannot be, well defined.
>That is, there was no moment when people
>ceased to speak Latin and began to speak Portuguese.
>And all of my objections to various postings on this list
>have centered on this fact, and on related facts.
>All the points I have objected to, it seems to me,
>depend crucially upon the assumption that such a moment exists,
No, they do not depend on that.
People on this list never made that assumption.
Everyone else accepted that there are fuzzy borderlines.
The points Trask does not like
were stated in terms of being able to differentiate in some situation
between a daughter dialect which is the "same" language, by whatever
criterion adopted, and a daughter dialect which is a "different" language.
That can be done easily when the two dialects being compared are far
apart on the scale of relative similarity to the parent, one so close
that one has no problem calling it the "same" language,
one so far away that one has no problem calling it a "different" language.
There is no need whatsoever to deal with close-calls, borderline cases,
in order to discuss the paradox that a daughter distinct language can
co-exist with a parent. McLaughlin agreed that Trask was wrong here.
>In my view, the relation "is the same language as" cannot be coherently given
>the kind of precise and principled content which you appear to require for
As pointed out, the kind of "precise and principled"
version that Trask feels the lack of,
perhaps because he does not like definitions with fuzzy borderlines,
is NOT that required by any of us here.
So we agree that the expression cannot be given Trask's (!) kind of
precise content, but do not require it for the argument.
>Except in a very broad and rough sense -- which is insufficient for
>the kinds of conclusions you want to draw -- language varieties simply cannot
>be divided absolutely into "the same" and "not the same".
A very broad and rough sense *is* sufficient
(the word "absolutely" is simply not appropriate
to that broad and rough sense,
and it was not required for the statement of the paradox).
Again, all that was required was that one be able,
in *some* cases, to distinguish "clearly same language"
from "clearly different language".
We need not have any criterion
so precise as to be able to deal with all cases
and give an easy answer or even any answer in the hard cases.
If Trask had discussed what other people were discussing,
this issue would have been resolved long ago.
>and that the notion of "same
>language" can be given a precise and principled sense. This I deny.
That is not a difference between us. I agree with Trask, yet again,
Like him, I deny that his kind of precise sense is possible --
as my statement that there are fuzzy borderline cases implies,
there is no possible reasonable definition
without fuzzy borderlines.
>Nor does the passage in my book commit me to any view that mutual
>comprehensibility is the sole or principal criterion for drawing language
No one has said it was the sole criterion.
It does appear to be "a" principal
criterion, *one used also by Trask himself*.
The paradox was framed in terms of that simper technical criterion,
but as pointed out many times, a more complex criterion will do just as well.
And again, the criterion does *not* have to be able to deal with the hardest
borderline cases in order for the paradox to be statable.
A rough one is sufficient.
>> But please don't pretend that a language
>> "changing enough" to be become a different language - ancestor or otherwise
>> is something you don't understand.
>In the context in which this notion has featured recently on this list,
>I find this notion incoherent -- as I hope I have made clear by now.
No, that is not true.
Trask, just like the rest of us,
finds the notion incoherent that there is
a sudden point at which one language becomes a different language,
or finds the notion incoherent that one can make sharp,
non-fuzzy distinctions in all cases
(however any of us wish to phrase that), or etc.
(Trask can word it in his favorite way).
Steve Long's quotations from Trask's book quite literally quote Trask
as saying that a language can change enough to that we are forced to call it
a different language (and of course there are cases where it has only changed
imperceptibly, so we then normally call it the same language).
>But you and Lloyd Anderson apparently want to defend a notion along the
>lines of "A and B are the same language but are not identical". And this
>notion, I think, cannot be given any coherent content -- at least not
>sufficiently coherent for the purposes you appear to have in mind.
*Any* gradient definition has those properties, that one can change
less than X, and "same" will still be used, or one can change more than Y,
and "different" will be used. Perfectly coherent, but with fuzzy boundaries.
As *Trask himself* uses the terms
(never mind whether he was speaking to beginners or not)
the quoted material two paragraphs above makes perfect sense.
The English of Trask today and the English of Trask
two months ago are so nearly identical that no one (including Trask)
would have any valid reason to call them "a different language",
in the sense of a *normal* opposition of "same" vs. "different" language.
Yet none of us think that means they are identical.
This is *not* difficult thinking.
The composite phrase "same language"
can perfectly normally have a meaning, for most of its users,
not entirely predictable from its parts,
in an absolute sense of predictable, specifically not including
a sense of absolute sameness, because this is the ordinary language
use of the word "same", which does have fuzzy borderlines.
[ moderator edit, by agreement with Mr. Anderson ]
More information about the Indo-european