IE pers.pron. (dual forms)

[Jens Elmegaard Rasmussen]

[ moderator re-formatted ]

On Sat, 24 Apr 1999, Patrick C. Ryan wrote:

> Dear Jebs and IEists:

[ moderator snip ]

> > The Gk. /-e/ cannot be a syllabic *-H1 if it is to match OLith. augus-e
> > 'the two grown ones' or OIr. di: pherid 'two heels', only IE *-e will do
> > here.

> The main problem, however, is one that I think we run into far more often
> than we generally recognize, and that is that some linguists *contrive* very
> complicated rules to be able to ascribe a common origin to forms that are
> simply not commensurable.

It only makes things worse if you reconstruct _against_ the rules: The
dual forms mentioned can all derived from *-e, the Greek one alone also
from *-H1, but then the forms indeed are not commensurable. BTW, I fail to
see the serious objection (if it is meant to be one): What is complicated
by deriving /-e/ of one language and /-e/ of another from a common
protoform *-e ?

> Beekes, I feel, does just this when he attempts to link the Sanskrit
> masculine and feminine dual forms (-a:[u], -u:, -a:[u], -e) with those of
> other IE languages like Greek: -e, -ei, -o:, [-a:]).

> It is as if Beekes had never heard the word "Nostratic"! Egyptian, for
> example, has a simple mechanism for forming masculine duals: -wj, with a
> plural in -w. Also, nearly every cardinal number has a -w suffix; and AA
> plurals (collectives in origin) in -u{:} are well-known.

It even looks as if Beekes considers the IE languages more closely
related to each other than to Egyptian.

> I interpret these facts (and others) to indicate that Nostratic had a
> collective suffix -w(V), and that this suffix was one of those employed
> to form a dual in IE.

I fail to see that such a morpheme has left any palpable imprint on IE.
But show us where!

> I would analyze Sanskrit -a:(u) as (C)wa in opposition to Beekes' -H{1}e.

> [ Moderator's comment:
>   The final -u in the Sanskrit dual is not Indo-European, but an Indic
>   development that is not present even in Iranian.
>   --rma ]

It is present in Goth. ahtau, Skt. aSta:/-au, Av. ashta. It appears to be
a special Indic _choice_ out of an Indo-Iranian pair of sandhi variants
which proceed from an IE pair of variants.

> But, another method of indicating the dual was almost certainly the
> suffix -y, here, not an adjective formant but just a suufix of
> differentiation. This will be the source of those dual endings like
> Greek -e, and Beekes recognizes the phonological process when he suggests on
> p. 195, that "Gr. o{'}sse, 'eyes' comes from *ok{w}-ye" but then goes on to
> derive *ok{w}-ye, IMHO incorrectly, from earlier *ok{w}-iH{1}, in a
> misguided attempt to unify -e and -a:(u).

I think Beekes is basically right here. The neuter dual did end in *-iH in
IE, and judging by Greek the laryngeal was H1, so *H3(o)k{w}-iH1 is
correct. Note that the laryngeal is even proved by the Skt. thematic
ntr.du. in -e which is sandhi resistent ("pragrhya"). As for the
connection of -e and -a:(u), the Skt. ending is of course that of the
thematic stems, i.e. identical with Gk. -o: ; -e is the original ending of
non-neuter dual of consonant stems, so the thematic *-o:(w) quite
obviously represents the combination of thematic stem *-o- + ending *-e.
Note that you get exactly the same result in the perfect of "long-vowel"
verbs like dha:-, viz. dadhau from *dhe-dhoH1-e (Avest. dada here too
without the diphthongization), i.e. IE *-o- + *-e with or without
intervening laryngeal gives Indo-Ir. *-a:(u).

> Of course, there are sporadic forms like Greek no{'}: from *no:wi, combining
> *ne/o + *wi:, 'two', and a 'laryngeal' is not necessary to explain the <o:>
> in a stress-accented open syllable; a mechanism as simple as transference of
> length back to the stress-accented syllable from *-wi: could explain it.

Are you talking about a metathesis of quantity, *nowi: > *no:wi? If so,
what makes you think that was a rule? Or is it an invention - mother
of comedy, huh - like some of my early stages?

[... On the 1st person dual pronoun:]

>I [...] would go the further step of suggesting that a
> 'laryngeal' is not required to be reconstructed at all.

Then what would be the enclitic form meaning "us two" in IE? */no:/ ending
in a long vowel?

[... On my notation:]

> As an aside, am I correct in presuming that <c> indicates /ts/ and <D>
> /dz/?

No, I meant <D> to be a spirant d ("edh"), but the ts is right.

Jens