A shift like g>gh>h>0 is very frequent, in a lot of languages. This is very
clear in Dutch & its dialects. We have no plosive /g/ in standard Dutch
(Hollandic, the dialect of Amsterdam etc.), but pronounce a fricative /gh/
when we say, eg, "goud" (Engl.-Germ."gold"). In Holland this "g" is hard
(guttural) and voiceless. In the South (Belgium) "g" is mostly soft (velar)
and voiced. In the West-Flemish dialect (Bruges etc.) they only say /h/ when
they pronounce "g"; and they don't pronounce "h", eg, they pronounce "hout"
(Engl. "wood", Germ."Holz") without /h/ (=0). My wife's dialect belong to
the same dialect group as mine (Brabantic: Brussels, Antwerp etc.), but she
doesn't pronounce "h" (which rarely causes misunderstandings...). So we have
a gradation in the pronunciation of continental Germanic "h" and "g":
W-Flemish<W-Brabantic<E-Brab.<Dutch<German. Obviously the W-Flemish
pronunciation of "g" as /h/ could only be possible if the "h" had already
lost its pronunciation (not only in W-Flemish, but also in the western
dialects of Brabantic), IOW, there must be an intermediate dialect where "g"
still has its normal pronunctiation /gh/ and "h" is not pronounced (ie, my
wife's dialect).
IMO there's nothing puzzling in the fact that different dialects (eg, Czech
& Ukrainian) independently can undergo the transition g>gh>h. For the same
reason the distinction centum/satem IMO doesn't say much about affiliations
of languages (k>ch>sh>s>0 is a "normal" decay process, seen in a lot of
languages, just as g>gh>h>0). An interesting question is whether these
"decay" processes are more frequent in peripheral dialects: W-Flemish &
Czech lie at the border of the Dutch-German & the Slavic group. IMO such
"shifts" are less likely to occur in more central dialects because there's
constant correction by neighbouring dialects.
I think g>gh>h is +-comparable to the Germanic consonant shift (parallel
softening of p>f, t>th, k>h) --perhaps at a time when proto-Germanic was a
rather isolated dialect at the borders of PIE. The second consonantal shift
in German (water>Wasser etc.) probably originated far south, at the borders
of the German dialects and then spread north (and still are spreading north,
after more than thousand years!, as people from Low Germany are adopting
standard German). Possibly such shifts are due to substrate influences of
neighbouring unintelligible languages (long-standing bilinguism, eg, the
non-pronunciation of "h" in French & W-Flemish).
Marc
+++++++++
>It's more of a South-North split in this case. And it's not really a "h" in
Ukrainian, more of a "gh". They do have a separate "h" or, rather, [x], that
sounds quite different. Oh, and I wouldn't really call it a "consonant
shift", cause it's nothing like the Germanic consonant shift, but Petr would
be able to find a better term for it. GK PS: And no wonder you
could understand Ukrainian after studying Polish - Slavic languages are
still very close to each other, kind of like Spanish and Portuguese. ;)
>The example that puzzled me arose with Slavonic languages. I notice a
consonant shift from g to h happening (or perhaps the reverse - I don't know
which came first). Example (one of many): Russian/Polish grad / gorod and
Czech / Ukrainian hrad / horod. What puzzles me is the way this seems
to cut across the standard linguistic groupings, with Russian / Ukrainian in
an Eastern group and Polish/Czech classified as Western. It would seem more
logical to me to have a "G group" and an "H" group (although I have a vague
worry this would split Sorbian down the middle).