Re: THEORY: phonemics (was: RE: [CONLANG] Optimum number of symbols

Tom Wier:
> Quoting And Rosta <a-rosta@...>:
>
> > Anyway, I understand where you're coming from, but there really
> > is no a priori reason to think that unconditioned allophony
> > within a given position must be within a continuous or
> > intensionally definable range; in principle, [s] and [m] could
> > be allophones of the same phoneme, and if that is vanishingly
> > rare, it is merely because it is a historical improbability,
> > not a phonological impossibility.
>
> But "historical improbabilities" are a function of a set of
> distinct synchronic circumstances. IMHO it is very difficult
> to maintain that any phone can pattern along with any other
> phone, given that empirically this simply does not occur. In
> case after case, phones that are acoustically similar not only
> tend to surface in complimentary distribution with one another,
> but to undergo the same kinds of phonological rules.  So, even
> though [h] and [N] in English are in complimentary distribution,
> there is no reason to believe -- whether from dialectology, or
> from synchronic phonological alternations -- that they are
> underlying allophones of the same phoneme (whatever we might
> want to call that). The fact that [s] and [m] are not usually
> considered allophones of the same phoneme is therefore not an
> accident.
My contention is not that the rarity of [s] and [m] as coallophones
is an accident. Rather my contention is that it is not *precluded*
by anything intrinsic to a phonological system. Hence I agree
with the points you make, but I don't think they argue that
[s]-[m] coallophony is a phonological impossibility.

As for the [h]-[N] chestnut, it won't be a cause for contention,
for I argued elsewhere in this old thread against the very notion
of allophones being in positionally complementary distribution.

--And.