[Fwd: Re: [tied] Re: IE lexical accent]

On Tue, 3 Aug 2004 enlil@... wrote:

> > Mono-syllabic nominal forms are always given added length.
>
> Jens:
> > Not true.
>
> Counterexample? I don't know of a clear one.

But there are many root nouns without length in monosyllabic forms: Lat.
nex, prex, Gk. phlóks, phléps, Gk. voc. Zeu^. And if the rule is
supposed to be a phonetic one that keeps you free of analogy which you
ridicule so much every time you see a dose of it in any argumentation of
mine, it ought to apply to other words than nouns also. Then verbal
forms such as Ved. han (2sg *gWhén-s and 3sg *gWhén-t), the numeral
*swék^s 'six', the adverb *g^hTyés 'yesterday', and pronominal forms like
*tóm, *tód, *tóy are counterexamples too.

>
> > You can get *wed-n- by shortening **we:d-n-,
>
> We see an alternation of *o and *e, so that's enough problem to deal
> with without inventing more, such as your unsubstantiated lengthening.
>
> The collective form *wedo:r doesn't need to be an "analogical
> creation" (and you say _I_ rely on analogy?!) if you get our head
> straight about *o/*e alternation. Based on the *e of the first
> syllable, we can safely see that the accent was on the second syllable
> when the word was first coined. The form being *wadá:rx in eLIE, it
> later raised pretonic unaccented *a to *e (*wedá:rx). Vowel Shift and
> other changes would produce the target form *wedo:r without any
> pathetic plea to analogy.

You are shouting and cursing at analogy as if it were some way of
cheating that I am introducing now. By any rules we imagine there is
less alternation in the attested facts than we predict, so analogy has
certainly taken its toll. It is more a matter of how little we can make
do with. Your own rules are widely of an analogical nature, but you
cover by a smokescreen of bewildering terminology.

> The origin of the form lies in late MIE where it was *wat:ár-ha. The
> accentuation obeys the penultimate rule.

Then what rule did the accent of Greek húdo:r obey?

>
> > I would suppose the lengthened-grade root vocalism expressed a note
> of durativity or habituality, perhaps also collectivity.
>
> Or nominality... but I admit I can't think of a strong arguement in
> its favour. It's a pattern that I see and I've pursued it.

Why am I unimpressed?

> > If the thematic nominative in *-o-s were a resegmented old genitive
> in *-os it should not be possible for it to be unaccented which it
> often is.
>
> Yes it is possible. You forgot a point I was making on that...
> Acrostatic Regularization
>
> The accent alternation between nominal and adjectival thematic stems
> was facilitated by Acrostatic Regularization. The rule only affected
> nouns while adjectives kept the original accent on the final syllable.
> The accent difference came to be a morphological rule when converting
> adjectives to nouns.

In what way is this a phonetic rule? Is it just the contrastive accent
under another name?

> It was at the very point when adjectives and nouns were differentiated
> by accent position that this adjectival category developped.
> Adjectives were then free to misanalyse their unchanging genitive
> endings as nominatives just like their accompanying nouns were marked
> in, at which point adjectives adapted the nominal paradigm to suit new
> requirements for _case agreement_. It's a small change from */newos
> ekwom/ "new horse" (ie: morphologically either */new-os ekwo-m/ or
> */newo-s ekwo-m/) to */newom ekwom/ (ie: *newo-m *ekwo-m). It's a
> trivial change that can easily have took place in this language over
> time and further explains the use of *-m in the inanimate, borrowed
> from thematic inanimate nouns in *-om just as the animate *-s was
> borrowed from thematic animates.
>
> There can't be a more fulfilling explanation of IE adjectives than
> that.

Oh sure: the -o- expressed adjectival function. If "[i]t's a small
change from */newos ekwom/ "new horse" [...] to */newom ekwom/ (ie:
*newo-m *ekwo-m)", it's an even smaller one from *newos *ek^wos to
*newos *ek^wos. That even goes without analogy.

> > And its vocalic part should not alternate the way the thematic vowel
> does, seeing that the vowel of the genitive ending alternates in a
> totally different way.
>
> The alternation of the thematic vowel is not based on accent.

Exactly, we agree on that point.

> Everything
> indicates that the genitive was originally accented *-ás except in a
> few cases of *-s which are predictable. The thematic of adjectives was
> _unaccented_ and therefore free to lengthen before voiced consonants
> and not before voiceless ones.

The e/o alternation of the thematic vowel has nothing to do with accent.
Its oldest form, however, must have been accented, since it is neither
-u- nor -i-.

> >> Alright. Let me see if I understand correctly. You're saying that
> the durative and the aorist were both athematic at one time.
> >
> > Yes.
>
> So then, why is the thematic added again? Subjunctive transferring? My
> god, not more analogy?! My solution doesn't require it.

Certainly, the subjunctive morpheme continued to exist and was now used
also on stems which already had one. It's like plural morphemes added to
collectives in some cases. We can afford some analogy if we save all yours.

> > Yes. It was apparently an ongoing process before and after the
> disintegration of the protolanguage. Some original subjunctives had
> already become indicatives in PIE, others followed in the individual
> branches.
>
> This seems untenable considering that thematic verbs are far more
> common than athematic ones. Something's fishy with your account. I'm
> siding with the _majority_ and explaining the _minority_ of cases that
> don't fit.

There is no guarantee that the oldest forms are in the majority. It's a
strange argument really, more often one hears the opposite fallacy that
something is necessarily old and original because it is not productive.
Both conclusions are of course baseless.

> > Durative ("present")
> > inj. *gWhén-t
> >[...]
> > Punctual ("aorist")
> > inj. *kWér-t
>
> Hmm. I think the overwhelming advantage to my position is that I only
> have to explain a handful of verbs, whereas you have to explain how a
> whole onslaught of aorists became duratives. You don't even attempt to
> attain logical simplicity.

Well, explaining "a whole onslaught" is superior to explaining less.

> Of course, I'm not saying that subjunctives can't be made into
> duratives. I guess I can see how they can but I hardly think that this
> can be a widespread, large-scale process. There's no motivation for
> such a deluge of subjunctives into the indicative category. None.

It is something we can see happening, so it does not matter how any of
us feels about it.

The subjunctive is frequently used in the Rigveda in combination with
the adverbs <nu:nám> and <nú>/<nú>: 'now', to express that the action
is to begin (says Grassmann). The present indicative is even more
frequent with the same adverbs, and a difference of pragmatic meaning is
not really palpable. In 8.40.10 nú is combined with the aor.sbj.
bhédati, the exact match of Gothic beitiT 'bites'. A special usage is
the aor.sbj. of verba dicendi in the first person, meaning 'I hereby, we
hereby ...', a content also expressed by the aorist injunctive and by
the present indicative.

> > Perhaps they were all _aorist_ subjunctives.
>
> This just seems completely lopsided grammatically. Do you have any
> respect for markedness? I'm sure that markedness has been violated by
> your account.

So they violated markedness when they did it. Who cares? The facts are
there. The Rigvedic combinations of subjunctive and nú/nú:/nu:nám
actually are more frequently in the present aspect, but the lexicalized
continuations of subjunctives as presents in the individual branches are
almost exclusively based on aorist stems. Perhaps the 'hereby' usage is
related to this?

> > No, read again, it's the other way around. I do not assume syncope
> for Finnish, and I do not need it. You assume an underlying vowel in
> the ending *-it, which in some forms is not present, so I ask you by
> what rule you made it disappear. That sounds fair to me.
>
> Alright. With IE, the rule is clear because that is the language group
> that I've been zeroing in on for so long. This is the rule of Syncope
> and indeed we have *-s in the accusative plural *-m-s.

You are just repeating the error that was pointed out to you, as if it
did not matter. There need to be a motivation for the assumption of a
vowel in the ending at an earlier stage. Not only syllabic sequences
turn into pure consonants when syncope reduces the words, also
pre-existing sequences of pure consonants yield that result. The modality
again is "can": It is possible that a vowel has been lost between any two
consonants in IE, but it does not have to be so in every example. It
seems to me you are putting one vowel too many in the underlying form of
every single ending in the whole of your IE grammar. No small error in
my eyes, and a strange choice based on an ever-repeated gospel of
simplicity.

> The motivation
> for preservation, as I've long said, is to differentiate it from the
> thematic nominative. Tyrrhenian shows *-er (Etruscan /-ar/) and is
> even present in Minoan. The *e in both IE and Tyrrhenian reflects
> earlier *i in Proto-Steppe.

I see no reason at all to accept that.

> Uralic itself shows *-t but the "loss" of vowel here is a mirage given
> that all nominal stems have been made to end in vowels! So naturally,
> *-it will give way to a vowel-ending stem to produce what appears to
> be a plural *-t in favour of the stem vowel.

That's why I used the example I gave: Finn. <lapsi>, gen.sg. <lapsen>,
nom.pl. <lapset>, gen.pl. <lasten>. The stem is clearly /laps-/, and a
supporting vowel /e/ is added where needed, in the uninflected form in
final position, where it develops further to -i, and in heavier clusters
before the final consonant, so that -ps-n, -ps-t, -ps-t-n gave -psen,
-pset, -psten > -sten. Surely there is no underlying vowel in the plural
marker of <lasten>. And if there is no rule of vowel loss, there never
was a vowel in it, as far back as analysis goes.

> Furthermore, the strict CV(C)
> shape of the syllables in Uralic (the same syllable rule found in
> ProtoSteppe) would forbid the cooccurence of two vowels, so one of
> these vowels was destined to go.

This is not necessarily relevant: The syllabic structure you observe for
Uralic (assuming it is correct) is the result of a development, not its
input. Old Church Slavic has only open syllables (in the sense that it
tolerates such clusters as appear also word-initially); by your
principle that would have to be original, but we know it isn't.
Phonotactics is one of the things that change the most. Even
synchronically there are competing registers depending on articulatory
accuracy, typically involving a reduced variant that shows the way to the
next stage, as indeed the case was in Slavic which got all its many
clusters that way; and by your principle that structure should also be
original, and again we know it isn't. So this is not a principle worth our
time.

> Now, you don't expect the vowel of the suffix
> to be given priority over the vowel of the stem, do you? The lack of
> vowel here is straightforward if we take into account how Uralic
> works. If we don't and reinvent it in your image, we lose sight of
> simple things like this. We don't need any more rules other than this.

If there are no rules needed, no change can be assumed, and there was
then no vowel to salvalge your analysis.

> > What *is* the point? The gen.sg. morpheme *is* *-os. In pronominal
> forms like *tóysoom we do seem to have a collocation of a plural
> stem and the genitive morpheme (*-s, zero-grade of *-os).
>
> Ugh. The point, yet again, is that the genitive originally did not
> mark plurality because it served to mark the indirect object, an
> entity in the sentence without prime focus in the topic. The form
> *toiso:m is a synthetic hodgepodge of suffixes lying around since
> early Late IE and does not represent anything terribly ancient except
> for the pronominal plural in *i which is connectable to the same
> suffix found in Uralic and EA to mark plurals.

If EA is Eskimo-Aleut, I am very surprised. Where does an *i mark plural
in that branch? Surely the ending -i meaning 'his/her - (pl)' (West
Greenlandic aqq-i, Chaplino atX-i, from PE *atR-i 'his names') does not
allow the isolation of a plural morpheme *i, for the -i is also a 3sg
possessive. It must represent the product of the morphemes otherwise seen
as plural in -t + the 3sg poss. in -a (in that order). Are you talking
about something else?

>
> > It is anybody's guess what *-oom is.
>
> The genitive plural of the Late IE period.

You don't say? Well again, where did it come from? Oh yes, it was lying
around.

Jens