IE pers.pron. (dual forms)

[Jens Elmegaard Rasmussen]

On Tue, 25 May 1999, Patrick C. Ryan wrote:

> The point was that Beekes, a trained IEist, does not feel that
> "sva'sa:rau" indicates no laryngeal.

I respect those feelings: It is a highly subjective matter whether one
takes this apparently solid counterevidence to disprove a laryngeal or
chooses the alternative of analogical levelling. Still, there is no
material demanding a laryngeal. The analysis with the laryngeal is
widespread in IE studies and is not Beekes' invention. I suspect it just
looks more "technical" than a vowel without. - It appears to matter a
great deal to you what Beekes says in the book you have, so you may look
at p. 138 and find it correctly reported that a laryngeal blocks the
working of Brugmann's Law in Indo-Iranian (Kurylowicz' discovery).

[...(Jens:)]

>> That makes the vowel of the genitive "ending" a part of the preceding
>> stem.

> Pat answers:

> Yes, that was my point.

> Jens continued:

>> If that is so, there will also appear a vowel before other endings
>> that cause the accent to move from one vowel to the next. Then why does
>> this not happen when the endings *-bhyos, *-bhis, *-su are added?

> Pat responds:

> If Beekes is to be believed, that are a number of inflectional patterns in
> IE. This makes your question a bit tricky to answer untill we know the
> circumstances. How about an example?

> Jens continued:

>> And in
>> root nouns where the stem is identical with that of a radical verb (root
>> present or root aorist), why does the vowel appear in the genitive *-os
>> (*-es), but not when verbal endings are added? These problems evaporate if
>> the stem is posited as consonant-final, and the gen. morpheme is *-os
>> (*-es).

> Pat responds:

> I am dependent on Beekes for the most current viewson IE morphology. Why do
> you not give us an example of this phenomenon since Beekes does not.

Beekes treats root nouns (189f) and root verbs (234) in a very clear
fashion. Look anywhere in the book, and you'll _never_ find the vowel you
ascribe to the root on the basis of the gen. in *-os reappearing before
_any_ other inflectional ending. Am I to give examples of something that
does not exist? You are the one making claims. You may also look at p. 162
of Beekes' book for the information that "each PIE root begins and ends in
one or two consonants".

[...]

[Pat:]

>>> if I understand Beekes correctly, OCS would not have oc{^}i but
>>> rather oc{^}e{^}. Is that incorrect?

> Jens responded:

>> Yes and no. Slavic neuter consonant stems are few in kind, and their
>> nom.-acc. dual regularly ends in -e^ which must be pure analogy with the
>> o-stems. The dual of 'eye ' _is_ oc^i (also so given by Beekes 173), along
>> with us^i 'two ears' a relic of the inherited form.

> Pat responds:

> Analogy to the rescue. What would linguists do without analogy and
> laryngeals to explain anomalies?

Tell lies.

> Perhaps you can explain to me why oc{^}i could not derive from *ok{w}-ye?

It could not fit Sanskrit /aks.i:'/ 'two eyes' if it did. But they can
_all_ come from a form in *-iH1. Since yo-stems have replaced their
vocative with -u from u-stems and the imperative has been replaced by the
reflex of the optative, I do not know of a case to show unambiguously what
does come out of PIE *-ye in OCS.

[... (Jens):]

>>>>>> u-stem *-u-y cannot give Skt. -u:.

>>> Pat asked:

>>>>> Why cannot su:nu{'}- + -y yield su:nu:{'} with compensatory lengthening?

>>> Jens responded:

>>>> There is no such rule. In one instance, a stem amu- got segmented off
>>>> by a funny analogy in the inflection of the pronoun asau 'that one'
>>>> (acc.sg.M amu-m) and was used in the formation of a pl. with /-y/, this
>>>> giving ami:, not **amu:.

>>> Pat responded:

>>> I think it is dangerous to assume that combinatory rules have acted
>>> identically at different periods, do you not?

> Jens objected:

>> Sure, but you were using completely unknown rules.

> Pat rejoinds:

> Since when is compensatory lengthening "unknown"?

I meant "completely unknown for the language concerned", which of course
is what matters. I don't believe such a compensatory lengthening rule has
ever been known for Sanskrit. If you assume -uy > -u: in Sanskrit, it is
your task to demonstrate that there is such a "rule", meaning that the
same change occurs in other cases where -u- and final -y meet. It would be
an interesting discovery if you have examples to show that (for Sanskrit,
mind you).

[...]

>>> Pat responded:

>>> Sorry, I cannot accept the idea that laryngeals still functioning in
>>> Sanskrit made yuge{'} sandhi-resistant.

>> It is a descriptive fact,

> Pat rejoinds:

> Your idea of a "fact" and mine are obviously totally different. That yuge{'}
> may be sandhi-resistant could be a fact. That the cause is your convenient
> laryngeal, is not!

But facts ought to be given explanations, and in this case it lies right
at hand. What is simpler than assuming that a neuter dual contains the
neuter dual ending? Now, in consonant stems the neuter dual in Sanskrit
ends in /-i:/. The most common (in Beekes' phonology, if I understand him
correctly, the only) source of that is a PIE sequence of i + laryngeal.
Then, if /yuge'/ is regular, and the stem is *yugo-, we are made to posit
*yugo-iH. That fully explains its sandhi-resistence, for before a vowel,
the H goes to the following syllable, leaving -oi to form a diphthong in a
syllable of their own, whence Skt. -e, even before vowel in the following
word.

[...]

Jens