Thanks for the input on laryngeals. My ideas have been
revised. Basically I now consider the laryngeals to
have the following values:

*H1 = *?-/*-h-
*H2 = *x
*H3 = *hW

Whether *H1 codes for two phonemes *? and *h in initial
position is another problem to be solved. The laryngeals
are completely parallel to the velar series:

*h *k [-low -round]
*x *q [+low -round]
*hW *kW [-low +round]

Phonemes marked for "low" are uvular and there are no
double-marked phonemes like **qW or **xW in Late IE.
Another way of putting it is that there are no phonemes
marked for both [+low] and [+round] at the same time.

The whole reason why IE *x and *hW both merge to Anatolian
*h while IE *h disappears is simple once we understand
these laryngeals according to their phonetic weakness. Of
all the laryngeals, *H1 is the weakest (a plain aspirate)
while *H2 is a nasty bugger that refused to go away the
most. Thus, visually we may represent this as follows:

*H2 > *H3 > *H1
*x > *hW > *h

In Anatolian, *h (*H1) would have disappeared first.
Without a plain counterpart, *hW filled this gap through
delabialization (and residual vowel colouring). This left
only *x (< *H2) and *h (< *H3) but the two later merged
together as *h, making it look like *H2 and *H3 must have
both had a same feature (like uvularisation) - Not true.
Mea culpa. You live, you learn. Go figure.

Origins of the velar/laryngeal system
I also came to this conclusion thanks to other yummy
considerations. It appears to me now that in Old IE
(c.7000-6000 BCE), there was only *?, *h and *hW. These
are roughly the ancestral values of later *h (*H1), *x
(*H2) and *hW (*H3).

Only labialisation was a special feature of Old IE phonemes
and there were no snazzy phonemes marked for palatal,
uvular or anything else. Also, labialisation was more
widespread occuring even with dentals, producing phonemes
like *tW and *sW that later were dephonemicized to *tw
and *sw in Late IE.

It's important to understand that uvularisation was once
due to allophonic variation beside neighbouring *a in Old
IE. So, with a two-vowel system, *[k] occured next to *e
(-low central) and *[k.] occured next to *a (+low central).
This allophony occured for all velars and aspirates both
voiced and unvoiced and was originally completely automatic.

However, around the Early Mid IE stage (6000-5500 BCE),
it appears that these allophones became seperate phonemes,
causing a special uvular series amongst both velar stops
and aspirates:

*h *x *hW *xW
*k *q *kW *qW

This system quickly collapsed into a three-way contrast
once roundness became an automatic feature of uvularized
phonemes. This caused *x and *q to merge with their
rounded counterparts *xW and *qW:

*h *x(W) *hW
*k *q(W) *kW

This is when *x (with automatic rounding) decided to
change into *w (as I suspect is the case for *kwon- "dog"
< MIE *kewane < *kexane). This caused a temporary gap
in the uvular series:

*h -- *hW
*k *q *kW

However, mediofinal *-?- softened to *h, pushing the
original *h down to uvular *x, thereby satiating the
previous system:

*h *x *hW
*k *q *kW

From here, we enter the Late IE stage and the automatic
rounding previously present in uvulars must have surely
ended since we know that *x (our *H2) did not cause
rounding of *e, only lowering.

Does everyone like that story?

- gLeN

Join the world�s largest e-mail service with MSN Hotmail.