In some languages that have a uvular /r/, as Danish and Northern German,
the main allophone of /e/ before this /r/ is like the English vowel of
hat. Danish actually has exactly the same degree of aperture in words like
kerne, stjerne, hjerte, persisk, hverve and salme, tante, katte, vaske,
i.e. pretty much that of Eng. hat. Now, languages can be very different
and have very different rules. Old Engllish seems to offer evidence that a
velar spirant is enough to back a vowel like that of present-day "hat"
(i.e. the OE unaffected <{ae}> digraph [does æ come through at your end?])
into something written <ea>, as in eahta 'eight', neaht 'night', meaht
'Macht'. I know this is breaking where only the final part of the vowel is
assimilated to the backness of the spirant, but other languages could have
it lower the whole vowel, or contraction could finish the job. I do not
really believe you are on the right course if you think about the
opposition in terms of "hard" and "soft". For the prestage of PIE in which
laryngeal coloration took place I do not think we can really know whether
/H2/ was a velar or a post-velar fricative, but it was almost certainly
one of the two.
As for /H3/, we know it was voiced at one point (following the working
of ablaut, for *píbeti had to lose the root vowel to make the consonants
of older *pe-peH3-é-ti meet), and we know it changed (what would otherwise
show up as) e into o which indicates rounding; also in the selection of
/m/ or /n/ as the reduction of /mn/ it works just as well as /w, p, bh, m,
kW, gW, gWh/ in taking /n/ by dissimilation, which also means it was in
fact rounded; in Celtic it was dissimilated away before labiovelars, as in
*eni(H3)kWos, *ma:tri(H3)kWaH2 (Hamp); in the lexeme *gWiH3wós it
completely assimilated to the initial in pre-Germanic, so that *gwigwos
could give Gms. *kwikwaz with the sound shift; this all practically proves
it was a voiced labiovelar fricative, i.e. the fricative counterpart of
The "first laryngeal" was not very different from a simple [h]. The
list of examples of its aspirating affect on a following voiceless stop is
quite long by now, much longer in fact than any list of examples of the
aspirating effect of /H2/ on a preceding stop, a rule accepted by
practically everybody. It *may* have had a slight palatal shade to it
also, for how else could Greek develop sequences of syllabic sonants into
/ne:, me:, le:, re:/ which all have a-timbre without laryngeals? So it was
something like an /h/ with e-timbre, a voiceless /e/ if you will. If
languages playing with aspiration as a feature in their stops are demanded
to have a phoneme /h/, here it is.


On Sun, 29 Sep 2002, Sergejus Tarasovas wrote:

> --- In cybalist@..., Jens Elmegaard Rasmussen <jer@...> wrote:
> > Could it be of interest that the Arabic letter names ka:f and qa:f
> have
> > prescribed pronunciations with very different vowels, ka:f having
> the long
> > counterpart of Eng. man (which actually is long in most varieties of
> > English), while qa:f has the vowel of Eng. call or crawl? Great
> variation
> > is expected in systems with few phonemes.
> >
> Yes, that agrees with my (as a speaker of Russian and Lithuanian)
> perception of the Arabic uvulars or pharyngealized consonants
> as "hard" (velarised) and the rest as "soft" (palatalized) (the same
> applies mutatis mutandis to post-consonantal vowels). But "hardening"
> is not that causes [e] ->[a], at least my language intuition says
> that when "hardened" [e] is lowered (as in Russian) or shifted one
> step back to become mid-central vowel (resembling Russian [y] in
> _mylo_) as in some High Lithuanian dialects, but not lowered 2 steps
> and shifted back 1-2 steps to become some kind of [a].
> Sergei
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.