Re: [tied] Re: Mi- and hi-conjugation in Germanic

From: Miguel Carrasquer
Message: 36699
Date: 2005-03-11

On Thu, 10 Mar 2005 17:22:27 +0000, elmeras2000
<jer@...> wrote:

>--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
>
>> >> What I object
>> >> to, with Jasanoff, is the completely arbitrary and
>> >> rigour-less way in which reduplication or lack of it is
>> >> handled in "standard theory". The Hittite hi-conjugation is
>> >> seen as a dereduplicated perfect, in spite of the fact that
>> >> reduplication is not otherwise lost in Hittite, even in
>> >> forms clearly associated with the perfect, like wewakta.
>> >
>> >I do not think there is any particular lack of rigour in the
>> >conservative theory. wewak- is every bit as much of a loner as
>beben
>> >is in German and ded-/da:d- are in Balto-Slavic.
>>
>> Jasanoff makes a good case for it being a pluperfect. Now a
>> pluperfect presupposes a perfect. And it's striking that we
>> see the same thing in Hittite as in Gothic (perfect wak-,
>> pluperfect wewak-, just like Goth hait-, preterit haìhait-).
>
>I am quite convinced wewakta *is* a pluperfect. That also convinces
>me that wewakki is a perfect, but one of present meaning. That is
>not really pertinent evidence in a discussion of perfects that have
>developed a preterite meaning. Is your quote "perfect wak-" from
>Jasanoff's book?

No, it was a mistake. Forget it.

[...]
>> >For the Balto-Slavic
>> >facts I have even added rigor by advocating a conservative
>> >derivation from reduplicated intensive structures like *ml-mólH-
>> >/*mél-mlH-, which I have supported by the evidence of the accent:
>> >the action of Hirt's law demands an asyllabic laryngeal, which
>can
>> >only be supplied in *molH- if this was earlier reduplicated.
>>
>> Doesn't work for me.
>
>I don't understand you. Do you mean it does not work for you because
>you refuse to posit a reduplicated form, or do you mean it would not
>work even if you did reduplicate the protoform?

I mean that in my view of Slavic accentology, any acute root
in a non-mobile paradigm attracts the accent, even if
unaffected by Hirt's law. Let's call it the "jábloko-rule".

>> >I also see dereduplication in the Hitt. sk-verbs which are
>> >iterative; I can understand an iterative function on the basis of
>> >earlier reduplication, but not so well from the sk-form itself
>which
>> >was inchoative. I can then also understand the lack of syllabic
>> >reflex of the laryngeal in zikizzi from *dhH1-sk^é- if this is
>from
>> >earlier *dhi-dhH1-sk^e-.
>>
>> Melchert claims that laryngeals do not syllabize in
>> Anatolian (although I'm not sure I agree).
>
>I know, I'm still waiting for an argument.

From Melchert?

[...]
>> Indeed not. But I have an explanation for o/e ablaut which
>> works for nouns and verbs alike: *ó is the reflex of an
>> earlier lengthened vowel (**a:) under the stress, and *é
>> results from the same lengthened vowel in pretonic position.
>
>Then you don't have a place for /é:/ in the acrostatic paradigms.
>You just killed Narten and now act innocent.

Absolutely not. You are right of course that Narten-forms
are closely connected to the o/e-Ablaut I describe above.
They are different aspects of the same thing.

Actually, most of the phenomena falling under "Narten" are
equally well described as cases of o/e-Ablaut according to
my definition above. For instance, if we look in LIV for
verbal forms classified as "acrostatic root presents", that
includes:

(1) forms that show /e/ in otherwise weak paradigms (e.g.
the middle);
(2) forms that have /a:/ in Indo-Iranian in an open
syllable, not after a palatal.

These might as well be cases of o/e-Ablaut.

We are left with a relatively small but important group that
shows strong forms with /e:/ and weak forms with zero or /e/
(Ved. dá:s.t.i, má:rs.t.i, rá:s.ti, s'á:sti, tá:s.t.i;
ks.n.áuti, stáuti; perhaps some forms with BS /e:/, like
sêkoN, smêjoN, tré.s^kiu).

These forms reflect a lengthened vowel *i:, which regularly
develops into /é:/ under the stress. That there are much
fewer cases of /e:/ than there are of /o/ is as expected,
given that /o/ is the result of the lengthening of both */a/
=> /a:/ > /o/ and */u/ => /u:/ > /o/, while only */i/ =>
/i:/ gives /e:/. In the weak grade, we would expect *i: and
*u: to give zero, while *a: gives /e/. In practice, there
has been some levelling (zero grade in má:rs.t.i, mr.jánti;
stáuti, stuvánti; but /e/-grade in e.g. tá:s.t.i
[*té:tk^-ti], táks.ati [*tétk^-n.ti]).

The levelling may have affected the expected Ablaut of roots
with original *u (o ~ zero) even more, at least I can't
think of a good example right now (the o ~ 0 Ablaut of the
perfect must explained otherwise). Presumably, there was a
tendency to unite the three expected Ablaut grades (o ~ e, o
~ 0 and e: ~ 0) into either o ~ e or e: ~ e.

There is one instance where perhaps Narten is endangered by
something you proposed, namely the explanation of Vedic 1sg.
past middle -i as from reduced *-h2. If we compare the
behaviour of the element *-e- added to the stative endings
*-h2, *-th2 etc. in the middle versus the
perfect/hi-conjugation, we see that this *-e- always carries
the stress in the middle forms (*-h2ái, *-th2ái, *-ói,
*-ntói, leaving the root in zero grade), while in the
perfect it is unstressed in the singular (o-grade + *-h2e,
*-th2e, *-e), but stressed in the 1/2 plural (zero-grade +
*-mé, *-té or something similar), and appears to be absent
in the 3pl. (perhaps expected *-né (*-ré) was replaced by
*-én > *-ér(s) here). But if -i indeed comes from *-&2 (I
would actually prefer *-h2-i, but that's not the point),
then perhaps the element originally behaved similarly in
both middle and perfect, and the middle singular once
regularly had e-grade and stress on the root, before it was
polarized as an end-stressed form. In that case, "Narten
middles" are in fact archaisms unrelated to the rest of the
Narten system. If my analysis of the middle vs. the
perfect/hi-conjuation is correct ("I have to X" vs. "I have
X'ed"), we wouldn't expect "lengthened grade" (e: or o) in
the middle anyway.

[...]
>> But the ugly truth is that the relict reduplicated forms in
>> Anglian do not support the "dereduplication with contraction
>> to e:2"-theory all that well. We have:
>>
>> la:can: leolc (vs. le:c)
>> le:tan (læ:tan): leort (vs. le:t)
>> re:dan (ræ:dan): reord (vs. re:d)
>>
>> What these forms actually suggest is that the e:2/eo/e does
>> *not* originate in reduplication-with-contraction.
>
>Do they not show the liquids twice? Is that not reduplication?

Exactly. That means that where the reduplicated forms were
maintained, we didn't get *e:2, so *e:2 cannot come from the
reduplicated forms.

>> [BTW, why is there no fracture in <heht>? There should be,
>> if the form is old].
>
>So <heht> is permitted to be not "old". Then why must the e-forms of
>the hi-conjugation be?

Why is there no fracture in heht?

>> The van Coetsem theory, on the other hand, explains the NW
>> Germanic facts quite elegantly.
>
>
>Van Coetsem and you show a complete disregard for the complementary
>distribution between the a-verbs of class VI and the a-verbs of
>class VII. The a-verbs in class VII are those that could not form
>the class-VI o:-type slo:h, fo:r, mo:l, ho:f, because they have
>closed syllables: falthan, staldan, faNhan, ha:han, aikan, fraisan,
>haitan, laitan, maitan, aukan. If you make the prestage of *fo:lth,
>that also becomes *falth. Therefore nothing is gained by replacing
>the reduplicated structure by the o:-form in this subgroup, so
>faifalth lived on. I consider it extremely unelegant to have a
>putative *far-/*fer- become far-/fo:r- if hait-/*heit- is retained.
>Why would that be regulated by the root structure?

Most likely for the same reason that /e:/ is regulated by
root structure in Classes IV and V. The parallel is
complete in the preterite plural, where verbs in -eC/-eR
have -e:C, verbs in -oC/-Or have -o:C. In the singular, we
would expect -oC in both cases (perhaps with reduplication
in the o-verbs), but we find -oC and -o:C. I'm not entirely
convinced that the /o:/ in the singular of class VI verbs is
due to analogy after vowel-initial verbs, as you suggested,
because there seems to be no contraction in Gothic vowel
initial verbs of Class VII (aìauk, etc. [the really
interesting cases, like alþan [*aì-alþ?] are not attested]).

If we had in the preterite:

e-verb: o-verb:
sg. bar- (fe)far-
pl. br- ~ bor- fer-

which was replaced by:

e-verb: o-verb:
sg. bar- (fe)far-
pl. be:r- fo:r-,

then the innovation was to extend o: to the singular of the
o-verb preterite (Class VI).

Verbs with other root structures did not introduce e: or o:,
so we have:

e-verbs o-verbs:
sg. staig- he-hait-
baud- he-hlaup-
band- he-hald-
pl. stig- heit- (> heet-)
bud- hleup- (> hleop-)
bund- held-,

which explains all the attested forms.


=======================
Miguel Carrasquer Vidal
mcv@...