Miguel:
> That is of course the big difference with IE, where initial
> clusters are quite common. Glen's proposal of "automatic"
> o-prosthesis does not explain why we don't have o- in the
> vast majority of cases (paradigmatic Ausgleich won't explain
> *ph2tér- instead of +oph2tér-).
Proto-IE itself allowed initial clusters, obviously. However
the ablaut alternations, particularly the zerograding of
full vowels in unaccented position or the parallel reduction
of long vowels to short, shows that IE was more syllabic in
the past, having lost many of the vowels during an event of
syncope presumably caused by stress accent. This idea has been
proposed or adopted by many others before me so I feel
unashamed to pursue this established train of thought in
order to develop better ideas on Pre-IE.
The penultimate accent of Mid IE follows directly from
syncope since it can immediately regularize overall the
unpredictable mobile accent of IE by assuming lost final
vowels where accent occurred in final position. None of my
rules should be taken as a panacea anymore than Grimm's Law
is, nor should the fact that it doesn't explain ALL forms
make it a failure. Afterall, Grimm's Law required Verner's to
account for the more minor exceptions. These rules can only
be explanations of _general_ phenomenon, not all phenomenon.
In the case of a-prothesis in early Late IE, it is not meant
to be applied to every word for the simple reason that
not every word dates to that stage, nor do the phonetics
of that word necessarily invoke that law.
The word *pxter- does not date to then and is strictly
_Late IE_. It is clearly derived from *pax- "to feed, provide"
+ *-ter- [agent]. Other words like *maxter-, *bHraxter- and
*dHugxter- were artificially modelled on *pxter- and given
what seems like a "family" suffix *-xter-. This is because
*pxter- was seen to be *p(a)-xter- (*pa/*apa/*papa, nursery
term), rather than *px-ter. So naturally *ma/*ama/*mama >
*ma-xter. Note that it retained *a because it really _does_
derive from *ma, while *pxter does not derive from *pa.
This is not to say that Mid IE didn't have family terms! It
just means that the ones reconstructed in Late IE are recent.
As I said, the other terms are based on a misanalysis of
*pxter- which led to a pseudosuffix *-xter- attached to
preexisting forms. Such a family suffix or conception of
one such as *-xter- does not follow in the very syllabic,
anti-cluster phonotactics of Mid IE, but it does makes
sense in post-syncope times, which shows that *pxter- and
these other family terms are strictly Late IE.
Therefore, they also postdate a-prothesis which happened
as a concurrent backlash to syncope.
Speaking of which, I will be examining what has been
said so far by Jens and Piotr. I can't help but think
there is something else going on than what is being said.
Perhaps there is another opportunity for a hybrid theory
like what went on with the nominative *z-conundrum which
forced me to finally accept Jens theory of *z but only
as a non-phonemic allophone of *s in final position
during early Late IE.
= gLeN