[front] = ‘palatovelar’, i-coloured;
[low] = ‘uvular’,
a-coloured;
[round] = ‘labiovelar’, u-coloured.
Frontness and lowness may occur together in vowels, but since front
dorsal consonants are redundantly [high], uvularity is incompatible with
palatality. Roundness is a labial feature and as such may accompany any
tongue-body feature, but both rounded palatovelars and rounded uvulars are
doubly marked and predictably rarer than less complex types.
If we accept that Brugmann’s ‘plain velars’ were in fact uvulars, the
three rows should be analysed as follows:
*K^ [(front)] -- velars with positional palatovelar variants;
*K
[low] -- uvulars (including h2);
*KW [round] -- labiovelars.
In the *K^ series frontness was not distinctive and remained an
allophonic feature in Kentum languages. This is why both in PIE and in the
Kentum branches *K^ is the unmarked series. Frontness became a distinctive
feature in the Satem group, and the Satem shift introduced [front] as the
primary contrastive property among the dorsal articulations. As a result,
dorsals that were not [low] or [round] tended to undergo fronting, while the *K
and *KW series merged as non-front, i.e. unmarked, velars. The *K/*KW lexical
set also absorbed a number of *K^ words in which, for various reasons, fronting
was blocked. Eventually, [front] dorsals evolved into a variety of coronals in
most Satem branches.
The feature [low] was lost everywhere, so *K merged with whatever was
the unmarked velar in a given branch. In the branches that retained *KW, *K was
automatically absorbed by *K^ as soon as the former’s lowness became
nondistinctive. One can speculate that the phonological reinterpretation of PIE
[low] stops and the lenition and loss of *h2 (probably something like *X > *h
> zero) are two aspects of a single process.
Finally, if *h2 was a uvular fricative, it should be transcribed as [X
~ R] ([R] here stands for a uvular fricative/approximant, not a trill). I
presume that the alophone [R] occurred in typical voicing contexts (like [z]),
or when *h2 was syllabic. Fans of symmetrical systems will perhaps like the
following arrangement (allophones bracketed):
[(front)] [low]
[round]
k/(c)
q
kW
STOPS
g/(J)
G
gW
gH/(JH)
GH gWH
FRICAT.
X
APPROX.
j
(R) w
VOWELS:
HIGH
i
u
NON-HIGH
e
a o
Piotr