Northwest IE attributes

[petegray]

Rick asked:

> Why the exception kravih?

It is far from being the only such word.   PIEists reconstruct three series
of velar consonants, based on centum kw ~ satem k;   centum k ~ satem k;
and centum k ~ satem k'  (and simliarly for g, and gh, except that the
evidence is slighter.)

There are 166 PIE roots listed in Bird which begin with /k/ rather than /k'/
or /kw/, while 51 begin with plain /g/, and 57 with plain /gh/.

Baldi, Foundations of Latin, says (summarised):  The palatal series: satem
k>s' and kw>k except (i) before /r/ and /a/; (ii) after /u/.   This accounts
for nearly all cases of the plain series /k/ (as opposed to the palatal k'
or the labiovelar kw), so the plain series may not be needed for the
reconstruction of PIE.

See Beekes, Comparative IE linguistics 1995 p 109-113, and Lehmann,
Theoretical Bases, 1993 pp 100-101.

Baldi seems sadly out of line with the evidence.   Many of the roots in Bird
do not fit his "rules", and he also seems to ignore the evidence of a
three-fold series suggested for both Anatolian and Albanian.

Rick also said:

> Sometimes, it almost looks as if IE has recessive genes, e.g. Romance
> Satem,

No - palatalisation in Romance is always and only due to the following
vowel, while the whole point of the three-fold velar series in PIE is that
the palatalisation in the satem languages has no explanation in any
surrounding phonological features - at least none that we can yet discover,
pace Baldi.

Peter