Re: The palatal sham :) (Re: [tied] Re: Albanian (1))

From: Piotr Gasiorowski
Message: 30111
Date: 2004-01-27

27-01-04 10:36, Mate Kapović wrote:

> I believe that *h1, *h2 and *h3 were *x', *x and *xW and many others do.

Symmetry isn't everything, even if it looks nice on paper. There is
evidence that *h1 was phonetically weaker than *h2 and *h3 and was lost
before them at least in some branches (certainly in Anatolian). This
suggests a glottal articulation of *h1 (for example, */h/) as opposed to
dorsal *h2 and *h3 -- a reconstruction consistent with the fact that *h1
is the only "colourless" laryngeal (it simply lengthens any preceding
vowel but doesn't affect its quality, and its syllabic reflex easily
gets coloured itself by assimilation in Greek). The presence of */h/ is
also consistent with typological expectations (a system with aspirated
stops ought to have an /h/ phoneme).

The /a/-colour of *h2 is easier to explain if we assume that its typical
phonetic realisation was a back fricative in the uvular range (*/x_/) at
least in early PIE, though it may have undergone fronting to */x/ in
parallel with *k. I'd say that in phonological terms *h2 was the
fricative counterpart of _both_ *k^ and *k, being simply dorsal and
non-labialised, and that any finer shades of its articulation were
merely allophonic. Jens has suggested here that "hardened" *h2 yields
Indic <j> in a voicing environment.

As for */x'/, it may have existed as a short-lived phoneme derived from
*hj (h = any laryngeal) in some positions (always when word-initial and
also syllable-initially as proposed by Pinnault [1982]). This explains
the absence of prothetic vowels before *j in Greek.

Piotr