On Thu, 01 Apr 2004 19:36:25 +0400, Âàäèì Ïîíàðÿäîâ
<
ponaryad@...> wrote:
>I've understood your main idea on the origin of the pronominal *e/*i alternation. Further, this point attracted my attention:
>
>The same rule (deletion of *i before *á-) also explains the
>feminine forms:
>
>> nom. *-á-ih2 > *-á-h2
>> acc. *-á-ih2-m > *-á-h2-m
>> voc. *'-a-ih2
>> gen. *-a-íh2-as
>> dat. *-a-ih2-ái
>> dat. *-a-íh2-ai
>> ins. *-a-ih1-át
>> pl.
>> nom. *-á-ih2 + átW > *-á-h2-àtW
>> obl. *-á-ih2- > *-á-h2- + oblique pl. endings,
>
>So, I see, you reconstruct the original marker of feminine gender as *-ih2. I like it very much, because it seems just excellent for external comparison. But why not *-h2 ?
Because the athematic feminine marker is *-ih2, as we see it
in the de:vi: and vr.ki:s stems.
These behave like ordinary consonant stems, with
hysterodynamic, proterodynamic (as well as amphidynamic and
static) paradigms.
PD AD HD
nom. *'-îh2-z *'-ih2-z *-íh2-z
acc. *'-îh2-m *'-ih2-m *-íh2-m
voc. *'-îh2 *'-ih2 *'-ih2
gen. *-íh2-âs *-íh2-âs *-ih2-ás
DL. *-íh2-a(i) *-íh2-a(i) *-ih2-á(i)
IAb. *-íh2-ât *-íh2-ât *-ih2-át
The lengthened vowel in the PD paradigm (*-î- > *-ye:-) is
resolved in different ways. *ye: is retained in unstressed
position in e.g. the Latin stems in -ie:s (materie:s besides
materia:). Otherwise, unstressed *ye: is reduced to *ye
(and *-yeh2 > -ya:, as in Latin materia:). But most often,
the reduction to *ye apparently took place when unstressed
*e was still capable of being reduced to zero, and the
result is *-ih2, as it is in the AD and HD types:
nom. *'-ye:h2(s) > *-ye:(s) *'-yh2 > *-i: *-íh2s > *-i:s
*'-yah2 > *-ya:
*'-yh2 > *-i:
acc. *'-ye:h2-m > *-ye:m *'-yh2-m > *-i:m *-íh2-m > *íym.
*'-yah2-m > *-ya:m
*'-yh2-m > *-í:m
voc. *'-ye:(h2) > *ye: *'-i(h2) > *-i *'-i(h2) > *-i
*'-ya(h2) > *-ya
*'-i(h2) > *-i
The oblique forms develop into:
gen. *-yáh2-os > *-yá:s *-yh2-ós *-yh2-ás > *-iyás
DL. *-yáh2-i > *-yá:i *-yáh2-i *-yh2-ái > *-iyái
IAb. *-yáh2-ot *-yh2-ót *-yh2-áh1 > *-iyá:
Cf. Sanskrit:
de:vi: vr.ki:s
G. -ya:s -iyás
D. -ya:i -iyái
L. (-ya:m) --
I. (-ya:) -iyá:
The plural forms:
nom. *'-îh2-ès(W) *'-ih2-ès(W) *-íh2-ès(W)
acc. *'-îh2-ms *'-ih2-ms *-íh2-ms
gen. *-íh2-âm *-íh2-âm *-ih2-á:m
DAb *-ih2-bhi-á:sW << <<
loc. *-ih2-sW-í << <<
ins. *-ih2-bhí-sW << <<
becoming:
nom. *'-ye:h2-es *'-yh2-es *-íh2-es
*'-yah2es
*'-yh2es
(> *-ye:s, *-ya:s, *-i:s) (> *-íyes)
acc. *'-ye:h2-ns *'-yh2-ns *-íh2-ms
*'-yah2-ns
*'-yh2-ns
(> *-yé:ns, *-yá:ns, *-i:ns) (> *-íyn.s)
gen. *-yáh2-om *-yh2-óm *-yh2-óm
(> *-yá:(o)m) (> *íyom)
DAb *-yh2-bhi-ós (> *-i:bhiós)
loc. *-yh2-s-ú (> *-i:sú)
ins. *-yh2-bhí-s (> *-i:bhís)
The feminine marker *-ih2 is originally a diminutive suffix,
related to the thematized diminutive ending *-ik-os (> Slav.
-IcI, etc.).
>> Yes. And Proto-Altaic *m corresponds to Proto-Nostratic *m,
>> Proto-Altaic *b corresponds to Proto-Nostratic *b.
>
>> This is what led Sasha Vovin ("Nostratic and Altaic", in:
>> Salmons/Joseph "Nostratic: Sifting the evidence") to state
>> that "... the inescapable conclusion is that PA personal
>> pronouns are unrelated to Indo-European and Uralic personal
>> pronouns".
>
>
>PA *b- ~ PU / PFU *w-:
>
>*bo:lo- ~ *wole- "to be"
>*biujlu ~ *wire "blood"
>*ba3a ~ *wa(n)c'V "early"
>*bior'u ~ *wa"De "a young domestic animal (calf, colt)"
>*bojl'o ~ *welV- "to learn"
>
>(Note that the correspondences PU *-l- ~ PA *-r-/*-l- and
>PU *-D- ~ PA *-r- are regular. So the connection of these
>words is quite clear, and they are not incidental similarities
>only in the initial sound.)
>
>PA *b- ~ PIE *w-:
>
>*bi:re "wolf" ~ wlkWo- "wolf"
>*beta "sea; ford" ~ *wet- "water"
>*beje "man, self" ~ *wiHr- "man"
>*bedu "thick, large" ~ *weto- "old"
I'd have to take a closer look at these etymologies (*bi:re
~ *wl.kWo- seems prima facie unlikely to me).
The PA ~ PIE correspondences, if confirmed, would support my
point: the regular outcome of *mW in PIE is *w (cf. 1pl.
*wey- < *mWéy "we", versus dissimilated *mWésW > *mésW >
Lith. *mes, Arm. mek` < *mesW. If in Uralic *mW > *w, taht
would mean that Uralic 1sg. *mi and 2sg. *ti derive directly
from PN *mi and *ti, and not through *mwi, *twi. The shape
of the 1/2 pronouns *mu/*mu-i, *tu/*tu-i would then be an
exclusively Altaic-Indo-European isogloss (perhaps also
Chukchi-Kamchatkan [pl. muri, turi], and --speculatively--
Sumerian g~a-, za- < *wa, *tsa < mu-a, *tu-a).
>> Vovin is of course right: if the Nostratic pronouns were *mi
>> and *ti, Altaic *bi and *si must be unrelated.
>
>PA *bi = PIE *we; PA *si = PIE *se > Hitt. zi-k, verbal 2 pp. -s.
Hittite zi:k comes from *ti:-g (Palaic ti), and the acc. tuk
(Palaic tu), as well as all the other oblique forms (tuel,
tuedaz etc.) show that the pronominal root was *tu(:), as it
is everwywhere in IE. Hittite/Palaic -i: in the nominative
is a special development of *-u: > *-y: > -i: in the Auslaut
(there is an exact duplicate of that soundlaw in Albanian).
Hitt. zi- comes from *ti- not from *si-.
The verbal 2sg. ending *-sW comes from the suffixed 2nd.
person pronoun *tu, which regularly becomes *-sW in the
Auslaut (cf. nom.pl. *-es(W) [Arm. -k`] < *-atu, Uralic *-d
> *-t, Altaic *-r, vs. oblique *-ati > PIE *-ey, Uralic *-j,
Altaic *-r2).
I find no evidence for *s- as a 2nd. person morpheme in
Nostratic. *s- is a demonstrative and a third person (or,
as Jens would have it, reflexive) pronominal stem.
The second person morphemes are *k- (Basque, Afro-Asiatic,
Kartvelian) or *t- (PIE, Uralic, Altaic).
>> However, the pronouns as we must reconstruct them for PIE
>> are not *mi and *ti, but *mu and *tu. If we derive the
>> Altaic pronouns from the same base, the problem of the
>> Altaic 1/2 personal pronouns disappears. The etyma now
>> become *mu- and *tu- (extended in the sg. with a vowel,
>> probably *i: *mu-i, *tu-i). We can now state a soundlaw to
>> the effect that the clusters *mw, *tw (or the labialized
>> consonants *mW, *tW) give *b and *s in Altaic (*b and *t in
>> Mongol), except when a nasal (at least *n) follows, when the
>> result of *mW is *m.
>
>All these soundlaws can be acceptable not earlier then additional examples of their application are found.
>
>> The same development *mW > *b can be seen in the
>> demonstrative (Turk. bu) and the accusative *-ba [both from
>> (oblique) *mu-a]. It would of course be nice to also have
>> some additional lexical items with the correspondence
>> PIE/Ural *m ~ Alt. *b[W], or with Turkic/Tunguz *s ~ Mongol
>> *t (PA *tW), but it's not essential.
>
>Really, this is essential. There seems to exist no correspondence PIE/PU *m ~ PA *b, and only one possible example of Tur. *s ~ Mong. *t (so"Nu"k "bone" ~ c^imu"gen < *timo"ge "marrow"), and even for the latter Starostin finds alternative etymological connections.
The place to look for such correspondences is not in
Starostin's database. Starostin doesn't know the soundlaw
*tW > Turk/Tung. *s ~ Mong *t, so no examples of it are
included (unless he has found, as you say, alternative
etymological connections). The place to look is in Turkic,
Tungusic and Mongolian (etymological) disctionaries, which I
unfortunately have no access to.
=======================
Miguel Carrasquer Vidal
mcv@...