>>The irregularity of *pxte:r and *maxte:r was not created out of
>>nothing. It was created by analogy.
>
>No. Analogy *eliminates* irregularities.
Yes, but at the same time, it can introduce new irregularities.
In this case, with a new word *pxte:r, the word for "mother"
and the words of other kinship terms no longer matched. One
might healthfully presume that **apa- and **ama- were the
original mommy and daddy words. So, a pair like *pxter- & **ama-
looks lopsided. This is the reason for the change, producing
a more balanced *pxte:r/*maxte:r.
>An old story, which I don't believe.
Unfortunately for you, *pxte:r < *pax- + -te:r remains the most
convincing explanation for the root.
>You forgot to explain the accent, which is all I wanted know.
>It's ph2t�r-, m�h2ter-, bhr�h2ter- and dhughh2t�r-. Why?
You need to look at these changes in a specific order. First,
*pxte:r was adopted as a new term for "father". The word
*m�xte:r, which pairs so nicely with *pxt�:r, resulted via
analogical change but retained accent on **ma- because **ma-
was not a verb like *pax- was (and therefore wasn't a typical
agentive noun like *pxte:r). Next, **bhr�xte:r, also retaining
accent on Early Late IE **bhrox, formed by analogy with *m�xte:r
and *pxt�:r. Subsequently, **bhroxte:r was further affected by
*maxte:r, producing *bhraxte:r.
Finally, a little later, the "daughter" word **dheuk-s (**dheug^-)
followed suit and adopted the emerging *-xte:r suffix, but it did
it differently because, by now, much time had elapsed since
"mother" and "brother" had been affected. So, it acted much
like *pxt�:r, with zero grade on *dhug^-, producing *dhug^xt�:r.
>>I clearly gave the explanation: The careful replacement of final
>>vowels efficiently regularizes the otherwise irregular accent
>>system of IndoEuropean.
>
>That doesn't explain anything. Which vowels dropped and which didn't?
Aarrrgh. I told you.
Stems with accent on FINAL SYLLABLE have lost their final vowels.
Look for indivisable, originally polysyllabic roots with accent
on final syllable, Mig! It's not hard. Examples include
*k^won- (lateMIE *kewane), *g^lou- (lateMIE *k:elaxWe) and
wed�r- (lateMIE wet:ar-xe).
In the case of *k^won-, it's straightforward because the initial
consonant cluster indicates an earlier polysyllabic stem
(*k^Vw�n-) with accent on what would be the final syllable *-w�n-.
However, by adding a final vowel (*k^Vw�nV), we REGULARIZE
and EXPLAIN the "mobile" accent in this and every other
paradigm with accentual shifts:
*k^Vwon & *k^VwVn-�s
becomes...
*k^Vw�nV & *k^VwVn�-sV
(Ultimately this is to be reconstructed as *kewane/*kewena-se
showing vocalic constraint and penultimate accent, as
well as being linked to the rule of vocalism-dependant
*k/*k^ allophony)
So we KNOW that this stem had a final vowel at one time. We know
that *wodr did not have a final vowel because the accent doesn't
fall on the final syllable. Rather, it falls on the penultimate
as it originally did.
>I think *h1s�nti is more correct. Yes, we all know that the accent shifted
>in the sg. vs. pl. of the verb, as it did in the
>rect./obl. of the noun, and that it looks like it's basically a
>syllable thing (C�Cr ~ CCR�s, h1�s-m ~ h1s-m�). But it's much
>more complicated than your "penultimate accent rule".
Of course it is! That's because the PA Rule isn't a panacaea. Your
problem is that you're looking for holy water. I don't sell holy
water. The PA Rule must be understood in combination with the
heteroclitic rule (*-n > *-r), the PreIE vocalic shift
(*� > *e, *a > *o), the PreIE vocalic constraint theory
(*a may only be stressed) and vocalism-dependant velar allophony,
combined with common sense to sift out the irregular developments
caused by analogy (cf. *-mn/*-m�n).
>If you've ever wondered why the standard handbooks don't carry it,
>it's because it's just not good enough. For one thing, by your
>"final vowel" rule, all IE suffixes should be split in two (e.g.
>the suffix *-men, as in d�imo:n and the suffix *-mene, as
>in poim�:n), without any difference in meaning between them.
Yes, when used blindly. If we used Grimm's Law blindly, we might
fail to see connections that are established by Verner's Law
or by irregular, analogical changes. Luckily some of us aren't
that daft.
To answer your question about *-mn/*-m�n, they are two different
MidIE suffixes, inanimate *-n and animate *-ene (< IndoTyr *-ena,
seen in Etruscan). The unaccented suffix *-(e)n is obviously the
source of the heteroclitic stems (*wodr, *yekWr...) while *-�ne
irregularly became *-�r- due to association with the inanimate
*-n>*-r.
Before *-n > *-r took effect, *-(e)n and *-�ne were already
perceived to be related to each other (even though they had
seperate origins). It was made into an inanimate/animate opposition.
When *-n became *-r, this aesthetic opposition was threatened
since it would produce nominoaccusative *-r in the inanimate
and *-�ne in the animate. So, *-�ne compensated, becoming *-�re,
and then *-�r after loss of unstressed vowels.
Of course, in the case of *-m-n and *-m-�n, the original *n
was preserved in both suffixes, but they are NOT the same
suffix and they did NOT serve the same function originally.
Sufficed to say, *-n/-r and *-er-/-en- are used in combination
with other artifact appendages like *-m- and *-t- (an old stative
marker), so the accent difference in this case is caused by
two different suffixes entwined together as one. The end.
- love gLeN
_________________________________________________________________
Get your FREE download of MSN Explorer at
http://explorer.msn.com/intl.asp