glen gordon wrote:
>>In those positions where we would expect *e:
>>itself to be lengthened (i.e. superlong), we find
>>*o: (or *o as a conditioned shortened reflex
>>thereof).
>
>
> [gLeNny crInGes and shIvErs] Ick...
What's so icky about it? Almost needless to say after so much
discussion, this is best illustrated by the long-vowelled nom.sg. of
animate root nouns. They fall into two classes: "e-nominatives" and
"o-nominatives".
The e-nominatives are typically simple action nouns or agent nouns with
stative values like *ne:k^-s 'death', *pre:k^-s 'prayer, request',
*spe:k^-s 'watcher', *gWHe:n 'killer' or *dje:u-s 'shining'. In the weak
cases the accent moves to the inflectional ending: (*pr.k^-ós, *diw-ós,
etc.).
The meaning of the o-nominatives always involves iterative or
resultative nuances, e.g. *bHo:r 'robber, thief' (= 'one who regularly
takes things away'), *wo:kW-s 'utterance, speech' (= 'the effect of
speaking') or *do:m 'house, building' (= 'the result of construction').
The o-nominatives have week cases with *e, such as *wekW-ós or *ped-ós,
and there are traces of an archaic-looking acrostatic pattern: gen.sg.
*dém-s, *nékWt-s, loc.pl. pé:d-su, etc., suggesting "Nartenoid" vocalism
in pre-PIE. Here and in his publications, Jens has presented a very
strong case for interpreting the *o: as a reflex of an originaly
overlong (trimoraic) root vowel, i.e. a fundamental *e: further
lengthened in the nom.sg. Given the meaning of these nouns, the
underlying length of the vowel was most likely iconic and had a fuction
akin to that of reduplication.
> Now, with *qep- /k.Ap-/, the lengthened grade is
> simply *qe:p- /k.e:p-/. Long vowels are not subject
> to lowering because their added length helps to
> reinforce their original quality. That takes care
> of this all.
Of course it does. I haven't said anything else.
> And as for *qo:pax, there's nothing
> about its ablaut that needs further explanation
> unless you're swayed by a very assumptive conviction
> that pre-IE must not have ablaut. The unnecessary
> O-fix rule is steering us away from the big picture.
How, then, do _you_ get a long *o: in such derivatives? If you merely
label it as a "simple" lengthened o-grade you don't _explain_ the length.
> In short, this supposed ablaut pattern of *a/*e:/*o:
> is only what you get when you mix the phonetics with
> the phonemics. Phonemically it's still *e/*e:/*o:
> like in a bunch of other non-uvular examples.
Depending on how abstract one's phonological analysis gets, or how deep
internal reconstruction one indulges in, it may be morphophonemically
just *//e:// (or diachronically ***e:) across the board, with various
rule-governed surface realisations (or historical reflexes). I didn't
mention that example in order to impress the list with unusual ablaut
patterns, but merely to demonstrate that "ordinary" Nartenian ablaut may
surface _phonetically_ in quite surprising ways. It's a safe bet that
most historical linguists are still puzzled by alternations seen in
roots like "*kap-".
Piotr