>>voiceless aspirates.
> > There are only 4 that are supported by both Greek and Skt, one of those
a
> > Lall-word.
> > Is this really "ample" evidence, which "demands" the reconstruction of
> > voiceless aspirates?
>
> Oh yes, that's more than the basis of a phonemic opposition between /gWh/
> and /g^hw/.
Alas Jens, when you make this claim, your facts are wrong; the distinction
between *g^hw and *gwh is very well founded.
The difference between *g^hw and *gwh is reflected in Sanskrit, Avestan,
Armenian, Albanian, Slavic, and Baltic.
But the difference between */p t k/ and */ph th kh/ can be reflected
securely only in Greek and I-I.
For *g^hw and *gwh, even though the difference only appears root-initially,
we have 3/4 minimal pairs, 3 other words in *g^hw and 8 in *gwh. Some of
these are very wide spread over the IE languages which can show the
difference.
For */p t k/ and */ph th kh/, there are only 3 words where *kh is supported
by both Greek and Sanskrit, and 1 for *ph. One of these is the "laughing
noise". In all other cases either a laryngeal is known to follow an
original */p t k/, or the evidence is contradictory, difficult to interpret
safely, or the aspirate is restricted to a single language.
You don't need me to spell out the reflexes for you, but in case anyone else
is following, we would expect to find:
gWh > Skt h/gh Av j^ Arm j^/g Alb gj- Lith g- OCS s^/g
g^hw > Skt hv/juhv- Av zb Arm j Alb z- Lith z^v OCS zv
If we consider only the minimal pairs, and only the languages that show the
difference, then we get:
gwhen "hit" Skt h / gh- Av j^ Arm j^/g- Alb gj- Lith g- OCS z^ / g-
gwhen "swell" Skt h-/ gh- Arm -g- Lith g OCS g
g^hwen "resound" Arm j- Alb z- Lith z^v OCS zv-
gwher "hot" Skt h-/gh- Arm j^- Lith g- OCS g-
g^hwer "wild animal" Lith z^v OCS zv-
g^hwel "become bent" Skt hvarate redupl juhur- & jahvar- Av zb- Lith z^v
OCS z-
gwhel "want" OCS z^
Since these roots are widely attested across Greek, Latin, Celtic and
Germanic, and since there are other roots showing the same distinctions, but
not as minimal pairs, the evidence from the satem languages can be taken as
establishing that *g^hw and *gwh were indeed different phonemes.
But no such certainty can prevail with */p t k/ and *ph th kh/. I accept
that the degree of certainty we give it is a matter of interpretation, but
the facts are that the claim for the phonemicity for voiceless aspirates is
much less securely based than that for *g^hw and *gWh.
Peter