tgpedersen wrote:
> Duly noted. They start from different points. Why don't those two
> Slavic palatalisations crash?
>
>
> I had this thought: Suppose PIE /e/ and /i/ went palatal, /ye/
> and /yi/ in Slavic? That would explain the 'palatal element' of /c^/
> vs. no such thing in /c/.
That's impossible. Inherited *je and *ji contrasted with *e and *i after
consonants, which shows that *e and *i had no palatal onglide in
Proto-Slavic (except word-initially, but this is irrelevant). If
anything, the reflexes of the tense vowel *e^ (triggering the first
palatalisation if from *e: and the second if from *oi or *ai) are often
accompanied by palatal glides in Slavic languages. But the *t in *te^
does not behave like the combination *tj, so it would be a mistake to
assume the pronunciation *[j]e^ after consonants at the time of the
second palatalisation. There was no collision because:
(1) The first palatalisation was already non-productive (except in
derived environments, where it had become grammaticalised as a
morphophonological rule of *k/*c^ alternation, not a living phonetic
process).
(2) The results of the second palatalisation at the time of its
productivity were different from the *c^, *z^, *s^ series. They were,
respectively, *c', 3', s' (predorso-postalveolar, then alveolar). The
traditional transcription of the affricates as *c and *3 conceals the
fact that they were still palatalised in Common Slavic; actually their
dispalatalisation took place much later and independently in different
dialects (and sometimes not at all). The fricative s' (from *x) merged
with *s^ in West Slavic but not elsewhere.
Non-intersecting alternative paths from velar stops to alveolar or
dental affricates/fricatives are possible because the articulatory space
of oral obstruents is not linear: the configuration of the tongue tip,
blade and body has to be taken into account as well.
Piotr