Revised Re: Campbell's def. of "language"

[ECOLING at aol.com]

The paradox that, with the usual meaning of the terms,
a language and its own descendant can coexist,

arises precisely because ANY definition which involves
matters of degree, and which uses a cutoff point to make the
answer to its question a discrete answer, will yield that result.

(One may of course call it a paradox or not, that is immaterial.
It's not a paradox when one understands it,
just like many other paradoxes,
but the result still remains, however inconvenient.)

"Mutual intelligibilty" is a minimally complex criterion
which has this property of turning a gradient measurement
into a yes-or-no answer.  So it is ideal for stating the
paradox clearly, and I used it precisely for that purpose.

Non-transitivity of "is the same language as" is an automatic
consequence of converting a gradient measure into such a
categorical answer, just as non-transitivity will be a property
of any other predicate whose meaning is of the type
"is in the same category C as", when it derives
from a gradient measure (sufficient similarity for some
practical purpose).

***

Even with Larry Trask's or anyone else's preferred definition,
adding other parts to a complex criterion,
the paradox originally mentioned will normally STILL obtain
as long as that definition contains a component which thus
derives a discrete answer from a gradient.
(I add "normally" only because a clever mathematical logician
might possibly be able to devise some odd combination of conditions
which indirectly make the paradox impossible to set up,
not because I believe any definition anyone has ever actually used
would be able to escape the paradox.)

Unless one defines the terms circularly to evade it,
or unless one denies (implausibly) that two dialects of the same
language can change at radically different rates so that one of them
would be considered the "same language" as the parent,
the other would not (under one's preferred definition,
WHATEVER that preferred definition is) .

WHATEVER definition one uses,
of the gradient measure type,
this is simply the paradox of different rates of gradient change,
combined with a categorial concept based on a gradient measure.

***

Lloyd Anderson
Ecological Linguistics