Re: Fun with prenasalized stops

> >According to Thurneysson, the OI neuter NA dual nasalizes, which
> >sets it apart from the NA singular, which doesn't, so that can't be
> >true. Either Thurneysson or you is right.
>
> I speak for Thurneysen.
>
> Thurneysen 285 [o-stem singualr] (p. 180):
> "Nom. voc. acc. neut., with neutral final, nasalizing"
> ibid. 287 [o-stem dual] (p. 182):
> "Nom. acc. Neutral final, leniting when masculine"
> "The neuter has the same form, but causes nasalization".
>
> The rest of the paragraph explains the course of events: the
> original o-stem n. NA dual *-oih1 should have given OIr.
> palatalization and no nasalization (as does indeed the
> a:-stem NA dual of identical shape).

Thurneysen bases that ending (*-oi) on Skt. and Greek, which is wrong,
Skt. has one dubious example of it (see Burrow quote). He must mean
Greek and Lithuanian. But other than that:
Ernout-Meillet:
qui:, quae, quod: qui, que. - Pronom relatif italique commun. Le thème
est en -o-, kWo-; il s'y est ajouté la particule épideictique -i, d'ou
le nom.masc. *kWo-i > quoi, quei ..., qui:; le féminin quae represente
*qua-i. Le neutre n'a pas cette particule; cf osq. puí, paí, púd "qui,
quae, quod", ombr. poi (poe, poie), "qui:", pur^e "quod"
In other words, again the pronoun turns out to be made up of two words
*kWo-i-, *kWa-i-
Subtract that -i from the supposed original dual NA *-oi and there is
only one ending, *-o, from *-oNw/*-o: <- *-omW. Therefore m.f.NA
-omW-i didn't nasalize, n.NA *-omW (if we assume the -*d is strictly a
sg. thing) did.

> But in Celtic, as in Latin and Greek, the neuter adopted the ending
> of the masculine (*-oh3),

*-ow < *-omW (*on,W?)

> which in OIr. should have resulted in -o > -0, thus merging with
> the NA singular, except for the nasalization, which the NA dual then
> took over, as a means to distinguish (again) the NA dual neuter from
> the masculine.

The nasalization was already there.

> >> The nasalization in the NA n. dual of vowel stems is
> >> unetymological (and so is the nasalization in the NAsg. n.
> >> of i- and u-stems).
> >
> >If you were right, that would be true. If received linguistic
> >wisdom says it is so, it is because they didn't know that *-w could
> >actually be *-nW, which you were the first to point out.
> >
> >
> >> >> The Dsg of "two" is Skt. dvabhyam.
> >> >>
> >> >
> >> > I forgot: In Old Irish the numerals 7, 8, 9 and 10 all nasalize
> >> > the following initial. PIE 7, 9 and 10 end in nasal, but 8 is a
> >> > dual.
> >>
> >> Cf. Greek combining okta- "8-", based on hepta- "7-", Slavic
> >> osmI "8" based on sedmI "7", Lith. septynì, as^tuonì, devynì
> >> "7, 8, 9",
> >
> >How do you explain -y- in 7 and 9, and -uo- in 8?
> >
> >> and the I-I ordinals saptamá-, as'tamá-, navamá-,
> >> das'amá-.
> >
> >*ok^tnW.- -> Gk okta-, Slavic osm(I)
>
> These are from analogical *ok^tm., after *septm., with loss
> of the original ending *-oh3(w).

Please snap out of textbook mode and start argueing against my proposal.

> >*ok^tonW -> *as^to:n-
>
> -on(W)- does not give a long vowel in Lithuanian. <as^tuonì>
> is from *ok^to: plus -n- as in 7 and 9.

Ok. Make it *ok^to:mW then.

> >*ok^tonW-ó- -> Skt. as´tamá-
>
> -on(W)o- _does_ give a long vowel in Sanskrit.

Assuming mW is a well-behaved consonant., so that the first syllable
in *-VmWV is open

> <as'tamá-> is from *ok^tm.(m)ó-, analogical after *septm.(m)ó-.

You never give up, do you?

> >and
> >*ok^tmW-ó- -> Latin octáv-us
> >How do you explain the -v-?
>
> *ok^toh3w-ós > *okto:wos, and then apparently o:w > a:w in
> Latin.

Apparently. I should have asked you about the -a:- instead of course.
*ok^tmW-ós -> *oktm.wós -> *oktaNwós (Reduktionsstufe in impossible
environment) -> octa:vus


Torsten


PS Nw = mW are two ways of writing the same phonmeme (*n,W is different).