Re: [tied] Re: Various loose thoughts

From: Miguel Carrasquer
Message: 36329
Date: 2005-02-16

On Tue, 15 Feb 2005 19:04:42 +0000, pielewe
<wrvermeer@...> wrote:

>--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
>
>[On the mechanism that gave rise to BSl. mobility.]
>
>
>> That still doesn't explain -ovi, unless the dative falls
>> under the "one or two other case forms", in which case one
>> may as well dispense with Ebeling's law altogether.
>
>
>I doubt one can build much on -ovi, for the following reasons:
>
>
>(1) Granted for the sake of the argument that it did not retract the
>stress in accordance with what Kortlandt calls Ebeling's law (it
>can't have), it risked receiving stem stress analogically because all
>other singular datives were stem-stressed.
>
>
>(2) By the time Slavic is attested, the u-stems as an inflexional
>pattern were no longer a distinct entity but were merging with the
>masc o-stems, increasing the chances of mutual influence of any
>conceivable kind.
>
>
>(3) What about modern Russian end-stressed "domój" and "dolój"?

domU (~ Grk. dómos, Lith. nãmas) and dolU (Grk. thólos) are
originally a.p. b, so domò(v)i, dolò(v)i are regular, by
Dybo's law.

>> A much
>> simpler solution is to see the mobile stress as a direct
>> continuation of Indo-European mobility, only extended
>> analogically to vowel stems. The cases with end-stress in
>> the paradigm of a PD stem like *dhug(h)&2té:r (Nsg., Gsg.,
>> Isg., Gpl., Dpl., Lpl., Ipl., GLdu., DIdu.) are plus, the
>> others are minus (although the Dsg. is a bit of a problem
>> here too).
>
>
>Perhaps also the o-stem Gsg (= Ablsg).

The ablative is also a "non-oxytonic" case (*-óod).

The vowel stems, when they became mobile in BS, did
introduce a few idiosyncracies of their own, like the
(originally) end-stressed o-stem Npl. *-ój, or the
end-stressed i- and u-stem L.sg's (*-é:i, *-é:u).

>Whatever may have been the case, I do think that we urgently need an
>update of this section of Kortlandt's theory, particularly in the
>light of everything that has happened since 1975, notably the arrival
>of the valency theory, which was not yet on the scene when "Slavic
>Accentuation" was published.
>
>> [...]
>
>I'd written:
>
>> >If I remember rightly, K puts the analogical introduction of the
>> >pronominal ending *-o < *-od into the (at the time) oxytone neuter
>o-
>> >stems at some relatively late Balto-Slavic stage preceding
>Ebeling's
>> >law. The attested stem stress in these examples is then due to the
>> >retraction he calls Ebeling's law. In that conception, it is not a
>> >matter of circumflexes attracting the ictus, but of final
>syllables
>> >of certain shapes losing it.
>>
>
>Then you wrote:
>
>> But that cannot be right. Only the oxytone o-stems with
>> circumflex become mobile (meN^so, ja^je), not those with
>> acute (vêdró) nor those with short vowel (peró).
>
>
>Here again we are faced with a section of Kortlandt's theory that has
>not been formulated explicit enough for me to understand it
>completely. But let me see what I can do. After all one lives only
>once. Or twice.
>
>
>Synchronically speaking, the type of which "pero" is a representative
>is squarely Stang's (b), which implies that it had fixed stem stress
>in the period immediately preceding Dybo's law. (I'm no admirer of
>the practice of referring to (b) as "oxytone" because it was barytone
>during most of the pre-history of Slavic and only became oxytone as a
>consequence of Dybo's law, only to turn into a new kind of mobile
>pattern as a consequence of Stang's law.) Now in view of the way
>neuter nouns are supposed to have arisen (more or less along the
>lines developed by Illic^-Svityc^), there is no immediately
>recognizeable source for (b)-stressed neuter o-stems. To the best of
>my knowledge the rise of (b)-stressed neuter o-stems is nowhere
>explicitly treated. But I've probably missed something.

I have written about this on several occasions. The word
peró (= Grk. pterón) is an oxytone neuter, and as such it
was a.p. b _before_ Dybo's law.

The whole point of Pedersen's law (i.e. the spread of
mobility to the vowel stems), at least in the nouns, must
originally have been to distinguish the nominative sg. from
the accusative sg. prosodically (a secondary, but more
lasting, effect would have been to establish a prosodic
distinction between NA pl/du and oblique pl./du.). That is
why "oxytone" (better: theme-stressed) vowel stems were
affected, because they could retract the accent of the
accusative to the initial syllable, e.g. N. gol&vá: ~ A.
gól&va:m, mimicking dug&té:: ~ dúg&terim. This also means
that neuters were not affected. The neuter C-stems were
barytonic anyway (*nébhos, *n.'mn.), there were no neuter u-
and a:-stems, and the very few neuter i-stems (*mori) became
neuter jo-stems. That only leaves the oxytone neuter
o-stems, which remained theme-stressed and were unaffected
by Pedersen's law.

Illich-Svitych correctly identified the fate of oxytone
o-stems in Slavic (his [sg.] examples are peró (Grk.
pterón), gnêzdó (Skt. ni:d.ám), jeNdró (Skt. a:n.d.ám),
oNtró (Skt. a:ntrám), sidló (Gmc. saila), gUrnó (Skt.
ghr.n.ás), dUnó (*dhubnóm), sUtó (*k^m.tóm)), but this did
not fit into the Lithuanian-inspired "two-paradigms
paradigm", and was apparently forgotten by Dybo (who writes
that there is "insufficient data" to clear up the fate of
the PIE oxytone neuters).

>The case of "ve^dro" is different, but yet again in a way that
>highlights a point where the theory has not been formulated very
>clearly, at least clearly enough to avoid misunderstandings. Read on.
>
>
>It has not often been realized that Ebeling's law splits the mobile o-
>stems into two subtypes, to wit one that undergoes retraction and one
>that doesn't (in other words: remains oxytone). The latter type is
>continued most clearly by such l-participles as modern
>Russian "neslo" (see SA 5-6). However, the further history of the
>subtype that did *not* retract the stress has not been reconstructed
>in very sufficient detail, at least in published texts. As I
>understand it (but I may be totally wrong here and I take all
>responsibility for the collapse of the bridge -- might it collapse) K
>assumes that masculine members of this subtype just merged with
>ordinary mobile nouns at some Balto-Slavic stage, in other words that
>they carried out Ebeling's law analogically. The case of neuter nouns
>is different. In their case the unretracted subtype lived on and Rick
>Derksen has argued in a way I tend to find attractive that it can be
>used to explain certain types of "metatony" in Lithuanian that have
>proved elusive and cannot be brought under some version of
>Hjelmslev's law. [Please read Derksen's book, I can't summarize it
>here or anywhere.]

I think I wrote about that too:
http://groups.yahoo.com/group/cybalist/message/35434

That was concerning the metatony in Lith. dz^iau~tas, dõtas,
garu~z^tas, -mau~tas; klõstas, mõstas, au~ks^tas; -bu~:klas,
-de~:klas etc.

>The noun ve^dro may well be a specimen of this subtype. Since its
>stem ends in -dr- it is formally of the non-retracting sub-type. At
>first sight it looks like a (b), but it obviously isn't a genuine (b)
>because in a genuine (b) the *e^ should be long and here it is short,
>showing that it was unstressed
>before Dybo's law caused genuine (b)'s to become end-stressed.

I'm confused by your terminology. "Genuine (b)" for me
means *pre-Dybo* (b)'s, in other words: (1) the neuter
oxytones like peró, vêdró, okUnó; (2) the compound words
made with dominant theme-stressed suffixes like *-ikós; (3)
the theme-stressed verbs in *-jé-, *-né-, *-í:-.

I think vêdró was end-stressed before and after Dybo's law.
So yes, the *ê was unstressed all along. This fits in
nicely with Shintani's amendment to Winter's law (which you
probably disagree with), which requires a vowel lengthened
by Winter's law to be pretonic (which automatically explains
the Latvian Brechton).

>Then you exclaim in the context of plural datives and locatives.
>
>
>> That's what seems totally backwards to me.
>
>
>Tut-tut.
>
>
>> The concept of
>> "weak" and "strong" yers in the sense of Havlík's rule
>> (pIpIrI'cI > pIprI'c) is clear enough, even given a few
>> exceptions (irregular strong yers in a.p. a initial
>> syllable, or sometimes before *j). But in Kortlandt's
>> account yers are first weak (unaccentable) in final
>> syllables only, although they are also unaccentable in
>> medial syllables, which get skipped, except before *j. Then
>> medial yers become accentable again, to account for the
>> behaviour of Dybo's law, and only after that Havlík's rule
>> sets in. I could accept that if it explained all the facts,
>> and if it were the only way to explain the facts, but it
>> doesn't, and it isn't. There is sufficient evidence that
>> shows that when a final yer lost the stress, a preceding yer
>> (strong, by definition) could and did receive the stress
>> (*dhworikós > dvorIcI' > dvorI'cI > dvoréc; *moldikós >
>> moldIcI' > moldI'cI > molodéc).
>
>
>You haven't understood K's theory, which [and *please* pay attention
>now, because I explained the same point on this very list only three
>days ago] at no point of the chronology generates accentuations like
>**dvorIcI'. Since the stressability of final jers is held to have
>been lost *before* Dybo's law operated there is no way of getting o-
>stem singular accusatives with stress on the final jer. This is a
>central point of the theory which for some reason keeps getting
>misunderstood.

Tut-tut. I *have* paid attention and I understand K's
theory.

But the PIE accent was still *dhworikós and *moldikós (and,
incidentally, *bhrah2trikós). The suffix *-ikós is
dominant, i.e stressed. The a.p. of the base noun doesn't
matter. All words in *-ikós (and similar suffixes) were
a.p. b in Proto-Slavic. The retraction law which I have
dubbed "minus Dybo" (or "Kurylowicz's law"), which is
simultaneous to Dybo's law, pulls the accent back to any
preceding acute (e.g. bra/tIcI' > bra"tIcI), but all others
(i.e. from a.p. b and a.p. c base words) remain a.p. b, the
only thing happening is retraction of the stress from a
final yer (dvorI`cI, dvorIca`; moldI`cI, moldIca`).

>Shënim: *dvorIcI is a derivation from (pre-Dybo) *dvòrU, which is (b)
>and as such is assumed to retain the stress in suffixal derivations.
>The MAS scheme (which K adheres to) unambiguously generates *dvòrIcI
>here, which is shifted to *dvorÌcI by Dybo's law.

That is wrong. The word is dvoréc, dvorcá in Russian.

What happened is that after Dybo's law and its inverse, the
distinction between the two classes of suffixes (stressed
and unstressed) became blurred. Previously, we had had:

stressed suffix:
(a) *bratr-IcI' = a.p. b
(a) *dvor-IcI' = a.p. b
(c) *mold-IcI' = a.p. b

unstressed suffix:
(a) *bra'tr-IskU = a.p. a
(a) *dvo'r-IskU = a.p. a
(c) *'mold-IskU, *mold-Iska` = a.p. c.

After +Dybo/-Dybo, we got:

"dominant" suffix:
(a) *bra'tr-IcI = a.p. a
(b) *dvor-IcI' = a.p. b
(c) *mold-IcI' = a.p. b

"recessive" suffix:
(a) *bra'tr-IskU = a.p. a
(b) *dvor-I'skU = a.p. b
(c) *'mold-IskU, *mold-Iska` = a.p. c

The words with dominant suffix made from a.p. b (by Dybo's
law) words tended to shift to the model of the a.p. b words
extended with a recessive suffix. In other words, the
valency model, which works from left to right, tended to
replace the old scheme inherited from PIE which, as it were,
works from right to left. Instead of original dvorI`cI,
dvorIca`, we also got dvorI`cI, dvo`rIca, as if *-IcI were a
recessive suffix here.

The stress pattern of a.p. c words extended with a dominant
suffix remained a.p. b, however, so the accent retraction
from *moldIcI' to *moldI'cI remains valid in any case.

>> K.'s soundlaw would predict
>> that the ins.sg. of mobile i- and u-stems (*-ImI, *-UmI)
>> would be barytone/enclinomenic, which is incompatible with
>> the evidence (adverbial R. verxóm cannot be secondary, and
>> the word is an old oxytone, Lith. virs^ùs, so the u-stem
>> a.p. c Isg. was *-U'mI without a doubt).
>
>
>Nice.
>
>
>> I also think the consequences of such a soundlaw have not
>> been thought through completely. I notice that in his reply
>> to Olander, Kortlandt says: "The obvious objection to
>> Olander's proposal is that the accent would have been
>> retracted in accordance with Hirt's law throughout a
>> Balto-Slavic paradigm *suHnù- with fixed stress on the
>> second syllable, so that accentual mobility could not have
>> been preserved." But the same can be said for Kortlandt's
>> model with regard to the "retraction from final yers
>> skipping middle yers"-law (for want of a better name). K.'s
>> model predicts *sy'nU, *sy'nU, *sy'nu, *sy'novi(?), *sy'nu,
>> *sy'nUmI; *sy'nove, *sy'ny, *syno'vU, *sy'nUmU, *sy'nUxU,
>> *synUmi', where it is hard to see how accentual mobility may
>> have been preserved based solely on the G and I pl. (Stang
>> p. 81 gives OR pl. sy'nove, sy'ny, syno'vU, syn(ov)o'mU,
>> syno'xU, synmí).
>
>Interesting point, but the Lsg should have been added to the list of
>forms with a stressed ending. (Whatever its precise origin it is
>obvious that the Lsg was end-stressed in (c)-stressed u-stems in
>PSl.)

I know, but how does Kortlandt explain that? Are *-e:i and
*-o:u closed syllables?

>And don't forget that an important component of a moble
>paradigm was also the fact that the stem-stressed forms were clitic.
>A learner of the language needed to hear, say *zà synU or *synU bò
>only once to know that the word was mobile. And how many (a)-stressed
>and (b)-stressed u-stems were available as a model for analogical
>introductino of fixed stem stress? Preciously few if any. The
>situation cannot be compared with the Olander reconstruction , which
>renders the BSl predecessor of synU consistently stem-stressed, just
>like, say, the word for 'brother'.

In the case of synU, I would say that the BS form *was*
stem-stressed (OLith sú:nus, 1>3). There are no a.p. a
u-stems in Slavic, so something must've happened to them
(they have become mobile).

Hirt's law did affect the Nsg. (*suHnús > *súHnus), which
was enough to make the whole paradigm barytone, cf. C-stems
like z^I"rny < *gWr.h2núh2, where _only_ the N.sg. was
affected by Hirt's law [_all_ other case forms have two or
more syllables, and Hirt's law could not affect them]. When
Hirt's law killed mobility's raison d'être, it was given up
(which makes it more remarkable that mobility wasn't killed
off later when nom. and acc. merged in most root types).


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