--- In
cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...> wrote:
>
> W dniu 2012-02-08 22:00, stlatos pisze:
>
> > If words looking slightly dif. w slightly dif. meanings were never
> > connected, we wouldn't have modern linguistics.
>
> Well, we wouldn't have it if we worked with "opt." sound changes and
> arbitrary metatheses all the time. Etymology would still be "une science
> où les voyelles ne font rien et les consonnes fort peu de chose".
>
The purpose of science is to describe reality. If many changes were opt., a description of them includes this. I can not choose for languages to be regular or irregular, I can only observe what exists.
Consider the set of words that supposedly someone demanding total regularity would consider unrelated:
*kes+ > c^esati = comb OCS;
*kas+ > kast = rake Lat;
*kasdH+ > haddr (m) = woman's long hair ON;
*kesk+ > késkion Hes \ késkeon = tow G;
*ke:r+ > cír (f) = comb MIr;
*ksan+ > xaíno: = comb / scratch, xánte:s = wool-carder G;
*kais+ > késara- = hair / mane S;
*kais.+ > kés.ara- = hair / mane S;
*kaikY+ > kéça-s = hair on head S;
*kYrixp+ > crib (m\f) = crest/comb/peak/apex W; crích (f) = end/boundary/furrow OIr;
*kYripHx+ > srifa:- = plume Av;
*kYrips+ > cri:nis = hair on head L;
*kYrip+ > krip = (short) hair on head / facial hair Al;
*kYipr+ > çípra: (du) = bearded lips S;
Some of these are obv. related by affixes (you already said you thought *kesr+ > *ke:r+ >), others just as obv. by met. (srifa:-, çípra: ). These are nec. to any good analysis. Why should I not extend those opt. changes already nec. to those not currently seen as nec., but still too similar to be in doubt by any but a Neogrammarian.
As such, since crib = crest/comb W; and srifa:- = plume Av; have two alt. (i:/i and p/pH), the variation can be caused by one met. (ixp vs ipHx) from an older common etymon. Mathematically, there is no simpler solution. A sim. argument can be easily made for most of the above.
Other affixes are not nec. from roots, but from nouns formed from roots, w noun der> verb by analogy in many cases.