Grzegorz Jagodzinski wrote:
> First, a technical note. I follow the old and GOOD tradition to spell H,
> never h, for "laryngeals" - but we do not know their place of articulation
> (my opinion is they were just velar spirants x^, x and xW). So, "gH" means
> for me "g" plus a laryngeal. And, I use "gh" the same way as in Sanskrit =
> for a voiced aspirated sound. Such a spelling is correct from the view of
> tradition and from the view of IPA (h = aspiration). And I cannot understand
> who and why changed this good old custom with the new one, inverse - perhaps
> only for making troubles.
Some (not all) of us use capital letters in lieu of superscripts or
subscripts, hence *gW, *gH, gWH, and *h1 etc. If you don't like this
convention, use anything that suits you, as long as it's clear what you
mean.
> I still see no reason for such an assumption. And, which is more, *k, *g and
> *gh seem to have been as marginal phonemes as *b in PIE. It is possible that
> they were present only in loanwords.
Do you mean items like *legH-, *(s)ker-, *kap-, steg-, *kreuh2-, etc.,
look like borrowings?
> Just take into consideration front [a] (like in German or French), central
> [a] (like in Polish) and back [a] (different character in IPA, like in
> Dutch) which are all low but which have different places of articulation. If
> you state that low vowels have no place of articulation, tell my what the
> difference between them is in.
Most of the difference depends on shape of the tongue, and in particular
the position of the highest part of the dorsum. But that's different
from the place of articulation, which is _not_ a tongue feature. If the
dorsum approximates the roof of the mouth, as it certainly does for high
vowels, it makes more sense to talk of "the place of articulation". But
for fully low vowels, even the front/back contrast rarely matters in
phonemic terms. Some British English accents have front [æ] in <cat>,
others have central [a], but in both cases the vowel in question is
simply the only unrounded short low vowel in the system.
> If I am right, *k, *g, *gh were not pharyngeal (and of course not uvular),
> and the a-colour is just the trace of the previous, pre-IE state.
It might be so. But then you'd have to agree that by PIE times the
"a-coloured" sound had been phonemicised and contrasted both with *e and
with *o.
>>The a-colouring effect of *k, *g and *gH was not as strong as in the
>>case of *h2, and the resulting vowel was not always phonemicised as
>>different from fundamental *e.
>
>
> Or, sometimes the former *ka would enter into the ablaut as *ke ~ *ko.
The former seems more likely, if in a root like *kap-, with a largish
family of derivatives, no restoration of short *e took place.
> I see these examples not-so-striking because I have doubts concerning the
> reconstructions you give.
Sorry, for lack of time I can't discuss the individual reconstructions
just now, but I'll begin with a general statement of my position and
will try to write more tomorrow:
I see no reason to deny the existence of _fundamental_ *a in PIE quite
independently of laryngeal and other colouring. Indeed, its existence
would seem to be required for the results of the colouring to be
phonemicised. Otherwise any *[a] would have remained an allophonic
variant of */e/ until such time as the conditioning *h2 was lost, in
which case at least some branches would be likely to have merged *[e]
and *[a] again.
The difference between old and new *a is as follows:
Old *a occurs in contexts where colouring by *h2 (or *k etc.) can be
ruled out, mostly (though not exclusively) in nouns and adjectives (this
is because the verb system was dominated by e/o/zero ablaut). Examples:
*k^aso- 'grey, hare', *g^Hans- 'goose', *h1albHo- 'white', *h(?)na:s-
'nose', *wa:stu- 'habitation'. It never alternates with vowels of other
quality, and undergoes only quantitative ablaut *a:/*a (perhaps also
/zero, but it's hard to be certain).
New *a occurs next to *h2 (or, sporadically, *k, etc.) and enters into
the same ablaut patterns as *e, i.e. can alternate with *o, *e:, *o: and
zero. Examples: *stah2-, *h2ag^-, *kap-, *bHag-.
Piotr