Exceptions to sound changes

[ECOLING at aol.com]

In a message dated 9/2/99 12:10:25 AM, kurisuto at unagi.cis.upenn.edu writes:

>Yes, it's quite true that the Neogrammarian Hypothesis (exceptionlessness
>of sound changes) is a working assumption.  It's an assumption which seems
>warranted, and further, it is the crucial assumption which allows us to
>conduct the Comparative Method at all.  If it's false, all bets are off.
>It would leave us with no currently known and reliable methodology for
>engaging in language reconstruction at all.

I believe it is NOT TRUE IN PRACTICE
that exceptionlessness of sound change is the crucial assumption,
though it is a central working technique (working hypothesis),
unless one means the words circularly in some way by definition.

1.  We know perfectly well that sound changes are not without exceptions,
all over the world.  That does not discourage us in general...

2.  There are often more detailed conditions,
which look like exceptions until we discover more fine-grained
conditioning, yet we do not throw out the putative cognate sets
just because of that, rather we work to find the more
fine-grained conditioning by contexts.

3.  Grammatical morphemes are often exempted from following
ALL of the "laws" of sound change, yet are still allowed to be
regarded as cognate by practicioners of the Comparative Method.

4.  Even worse, sociolinguists and dialectologists know of many
cases in which sound changes spread slowly across the vocabulary,
when observed in process.  When observed from a great distance
in time, this means that some part of the vocabulary undergoes a
particular sound change, another part does not.  When combined
with fine-grained context conditioning as in (2), this becomes very
complex.

5.  Yet the bottom line is that, while we look for regular sound
changes, we only need to find an approximation to them, in practice,
to be satisfied.

6.  The above is NOT to be confused with the situation where we
are confident that we are WITHIN the limits where a change is
regular, and therefore can conclude that a word was borrowed
from one language into another, or even in the most difficult cases
if we have ideal information, one word was
borrowed from one dialect into another closely related dialect.

7.  So, if exceptionlessness is not an absolute requirement,
ONE of the ways of struggling to deal with this fact, while
being able to penetrate greater depths where we do not have
so many recurrences of a particular sound correspondence,
so less ability to judge whether it is exceptionless or not,
is to measure not "identities" but "near identities", with a metric
of "nearness" which must be empirically based.  This can maintain
rigorous method without absolutes.

Best wishes,
Lloyd Anderson
Ecological Linguistics