INDO-EUROPEAN

A short summary of my views on PIE. I have given more elaborate accounts
elsewhere.


PIE PERSONAL PRONOUN

The 1st person sg. pronoun has a suppletive form in the nominative. The
attested forms can be derived from *h1eg^ (Gmc *ik, Slav. azU; with secondary
devoicing Lith às^, ès^, Arm es, Av. as). Sanksrit ahám (Av. azam) is from
*h1eg^hom, with *g^h instead of *g^, and added *-om (which appears in the other
personal pronouns too). Latin and Greek have final *-o: (ego, ego:(n)).
Hittite has <uk>. Whatever the ultimate origin of this form, it was created
within IE, and is not directly comparable to other Nostratic forms.

The accusative goes back to *mé(:) (Skt. má:-m, Grk. e-mé, Lat. me:, Alb. më,
Gmc *mi-k Arm. i-s), with enclitic is *me(:) (Skt./Av. ma:, Grk me, TochB. -ñ)
or perhaps *mu (Hitt. -mu).

The genitive is *méne(:) (Av. m&:na:, OCS mene) or *méme (Skt. máma).

The dative is *meg^hyo (Skt. máhya(m), Lat. mihi:, Arm. inz), with enclitic
*moi (Skt. me:, Av. mai, OCS mi, Grk. moi).

The corresponding forms for the 2nd. p. sg. are:
nominative *tú(:) (Skt. tv-am, Av. tv-&:m, tu:, OCS ty, Lith. tù, Hitt. zi-k,
Lat. tu:, OIr. tú, Alb. ti, TochB t(u)w-e, Gmc. *Tu(:), Arm. du)
accusative *twé(:), (Skt. tvá:-m, Av. Tva:-m, Hitt. tu-k, Greek sé, Lat. te:,
Alb të, Arm k`e-z) enclitic *te (Hitt. -ta, -tu, TochB -c)
genitive *téwe (Skt. táva, Av. tava, Lith. taveN~s)
dative *tubhyo ~ *tebhyo (Skt. túbhyam, Av. tabya(h), OCS tebê, Lat. tibi:),
enclitic *toi (Skt. te:, Av. tai, OCS ti, Grk. soi)

If we put the reconstructed forms side by side (as well as those of the 3rd p.
reflexive), we get:

N *h1ég^- *tú --
A *mé *twé *swé
enc *me *te *se
G *méne, *méme *téwe *séwe
D *még^hyo *t{ú/é}bhyo *s{ú/é}bhyo
enc *moi *toi *soi

We can derive this from earlier:

N **mú **tú **sú
A **m-mé **t-wé **s-wé
enc **me **te **se
G **mé-me **té-we **sé-we
D **m-mé-g^hyo **t-wé-bhyo **s-wé-bhyo
enc **mo-i **to-i **so-i

We know from other consonant alternations in PIE (e.g. *t ~ *s) that
a stressed **ú labialized a following consonant (**mát-nût-z > *méhno:ts, but
**mat-nút-âs > *m&1nésos), so the accusative and genitive can be seen as
variants of the same pre-PIE forms **mú-ma > *mémWe, **tú-ma > *témWe, **sú-ma

> *sémWe (and further with stress variation *méme ~ *mmé, *téwe ~ *twé and

*séwe ~ *swé). The dative was built on the accusative, using *-g^hyo in the
first person, *-bhyo in the 2nd. and 3rd. persons (why this is so is unclear,
as is the orgin of these forms: one is reminded of the thematic genitive in
*-syo).

The enclitic forms *me, *te, *se and *moi, *toi, *soi are not based on the
nominative **mu, **tu, **su, but on "absolutive" **ma, **ta, **sa.

Internal reconstruction thus leads to the following pre-PIE forms:

1. 2. 3.
nom. **mú **tú **sú
acc/gen **mú-ma **tú-ma **sú-ma
abs. **má **tá **sá

The plural forms are:

1pl.
N. *mé(:)s (Lith. me~s, Arm. mek`, ?Slav my < *mo:s)
*wéy-es (Skt. vay-ám, Hitt. we:s, TochA wes, Gmc *wi:z)
A. *n.smé (Skt. asmá:n, Av. ahma, Hitt. anza:s, Gmc *uns)
enc *no(:)s (Skt. nas, OCS nasU, Hitt. -nas, Lat. no:s)

2pl.
N. *jú(:)s (Av. yu:s^, Lith. ju~s, Alb ju, TochB yes, Gmc *ju(:)z)
*swéy-es (Hitt. sume:s, OIr. sí)
A. *usmé (Skt. yus.mán, Hitt. sumas, Grk. humâs, Gmc *izw-)
enc *wo(:)s (Skt. vas, OCS vasU)

3pl. only Greek spheîs, sphâs.


We see that the nominative shows two variants, one with final *-s (Arm -k`),
the other with final *-éy. The plural *-s cannot be the same *-s as the one in
the nom. and gen.sg., as that gives -0 in Armenian. We also know that /k`/ is
the regular reflex of *sw in Armenian, so we might want to explore the
possibility that the plural marker was *-sW (labialized *s). One piece of
evidence is the locative plural, where we have *-sú almost everywhere, but *-sí
in Greek, as well as *-i in
the singular. If plural *-s was actually *-sW, then loc.pl. *-sW-í could
easily have given *-sú.

If we go one step further and derive this *-sW from earlier **-tu, we can
further explain the 2sg. verbal ending *-s(W) as agglutinated from the 2sg.
personal pronoun *tu. The change /tW/ > /s/ is well attested in the prehistory
of e.g. Greek and Japanese.

The forms in *-ey- can then be explained as coming from *-ati (as in the
numeral *trey-es < *tVlati = Semitic Tala:T).

This gives the following plural forms:

1. 2. 3.
nom. **mu-átu **tu-átu **su-átu
abs. **mu-áti **tu-áti **su-áti
acc/gen **mu-átu-ma **tu-átu-ma **su-átu-ma
**mu-atu-má **tu-atu-má **su-atu-má

The tu..tu sequence in the 2pl. form was dissimilated to
**uátu, so that we have:

1. 2. 3.
nom. **mWésW **wésW **sWésW
abs. **mWéy **sWéy **sWéy
acc/gen **mWesWmé **wesWmé **sWesWmé

and further:

1. 2. 3.
nom. *més(W) *ús(W) --
abs. *wéy *sWéy --
acc/gen *nsmé *usmé *s(W)smé

The enclitic forms were built on the accusatives *nsmé, *usmé, *s(W)smé by
inserting *o and dropping final *-me (*n-o-s, *u-o-s, *s(w)-o-s).

The dual is analogous to the plural, except that the dual markers are
*-eh3 and *-eh1 instead of *-esW and *-ey.

1. 2. 3.
nom. **mu-íku **tu-íku **su-íku
abs. **mu-íki **tu-íki **su-íki
acc/gen **mu-íku-ma **tu-íku-ma **su-íku-ma
**mu-iku-má **tu-iku-má **su-iku-má

developing into:

1. 2. 3.
nom. **mWéxW **wéxW **sWéxW
abs. **mWéç **sWéç **sWéç
acc/gen **mWexWmé **wexWmé **sWexWmé

and further:

1. 2. 3.
nom. [*móh3] *úh3 --
abs. *wéh1 [*sWéh1] --
acc/gen *nh3wé *uh3wé *s(W)h3wé
encl. *noh3 *woh3 *s(W)oh3


PIE VERB

The Anatolian active conjugation is:

present: past:
Hitt. Luw. Hitt. Luw.
1. -mi -wi -un -ha
2. -si -si -s -s
3. -zzi -tti -t -ta
1. -weni ? -wen -man
2. -teni ? -ten ?
3. -anzi -nti -e:r -nta

This can be reconstructed as:

1. *-mW-i *-m.
2. *-s-i *-s
3. *-t-i *-t
1. *-mW-én-i *-mW-én
2. *-t-én-i *-t-én
3. *-én-t-i *-ér

The 1/2 singular forms are clearly derived from agglutination of the
personal pronouns **-mu, *-tu, while the 3rd. sg. ending can be explained as
coming from the oblique demonstrative root **ta (*to).

The plural forms contain the same **-mu, **-tu and **-ta, but combined
with a plural morpheme **-an (*-én-, *-ér in the Auslaut, except after
*-m-, as in the neuter nouns in *-r., *-n-os, but *-mn., *-mén-os).

1. *-mu > *-mW
2. *-tu > *-sW
3. *-0 > *-0
1. *-mu-án(a) > *-mWén
2. *-tu-án(a) > *-tWén
3. *-án(a) > *-én

At this point, *t(a) was added to the 3rd. person forms, and *-i was
added to make a present continuous:

1. *-mWi *-mW
2. *-sWi *-sW
3. *-ti *-t
1. *-mWéni *-mWén
2. *-tWéni *-tWén
3. *-énti *-én(t)

The *-n > *-r soundlaw comes later:

1. *-mi, *-wi *-m.
2. *-si *-s
3. *-ti *-t
1. *-méni, *-wéni *-mén, *-wén
2. *-téni *-tér (anal. *-tén)
3. *-énti *-ér, *-ént

[The form *-te:r (analogical after 3pl. *-é:r) is found in Toch.B.
-cer]

The thematic stems are analogous, except that 1sg. -omWi was affected
by Umlaut of final *-i, and the resulting *-owu became *-o:u (as in
the u-stem loc.sg. *-ou-i > *-o:u), still attested in Toch.B. -ew.
Elsewhere *-o:u was further reduced to *-o:.

In the 2nd.sg. thematic, intervocalic *-sW- is reflected regularly as
*s^ in Slavic (-e-s^-I).

In the 3pl., the past tense always has *-ont (*-o-ent), never +-ór.

Besides this model, as reflected most faithfully in Anatolian, there is a
different scheme, which can be reconstructed as:

1. *-mu > *-m(W)
2. *-tu > *-s(W)
3. *-su > *-s(W) or: *-0 + *tu > *-s(W)
1. *-mu-átu > *-mWésW
2. *-tu-átu > *-t(W)é
3. *-su-átu > *-s(W)ésW or: *-0-átu + *-tu > *-éssW

As in the 2pl. pronoun, the form *-tuátu was dissimilated to *-tuá (> *-té).

Third sg. *-s occurs in the Hittite and Tocharian preterite, and probably lies
at the origin of the s-aorist (*-s-m, *-s-s, *-s-t, etc., with Szemerényi
lengthening of the root).

Third plural *-(s)és(W) is not attested anywhere, but was combined with *-ér of
the first scheme as *-ér-s > *-é:r (Av. -r.s^, Skt. -ur, Hitt. -é:r, etc.) or
with *-ént as *-énts (Osco-Umbrian -ens).

For the dual, only such combined forms are attested:

1. **-mu-íka > **-mWáh2 + ésW > **-mWh2ás > *-wás
2. **-tu-íka > **-t(W)áh2 + ésW > **-th2ás > *-thás
3. **-0-íka > **-yáh2-t + ésW > **-yh2tés > *-(i:)tés

The difference between the two conjugation models seems to reflect
an old transitive / intransitive distinction: while the first form
has 3rd. sg. *-t(a) and pl. *-án(a), the second form has 3rd. sg.
*-sW (either from 3rd. sg. *-su, or from *so, the nominative of
the demonstrative pronoun *to) and plural *-átu, which reflect old
"ergative" forms. The 1/2 sg. show no such distinction: this would
indicate that the old absolutives *ma and *ta (which survive as
accusative enclitics) had been already replaced by ergative *mu and *tu in a
new subject case (nominative). This would also explain the necessity for new
accusative forms based on the ergative (**mu-ma, **tu-ma, **su-ma > *mé, *twé,
*swé) in the personal pronoun.

Besides the active conjugation and its two subtypes, PIE also had
two other conjugation types based on the stative: the perfect (Hittite
hi-conjugation) and the middle.

The PIE perfect is reconstructed as (Hitt. hi-conjugation forms to the right):

perfect present past
1. *-h2a -hi -hun
2. *-th2a -ti -t(a)
3. *-e -i -s(ta)
1. *-mé -wéni -wén
2. *-té -téni -tén
3. *-é:r -ánzi -é:r

The middle seems to have originally been a verbal form incorporating
the indirect object:

1sg/pl 2sg 3sg/pl 1du 2du 3du 2pl
*-m(W) *-r(i) *-(i) *-h2m *-h2r(i) *-h2(i) *-dh(W)(i)
1. -- -h2ar(i) -h2a(i) -- -h2ah2r(i) -h2ah2(i) -h2adh(i)
2. -th2am -- -th2a(i) -th2ah2m -- -th2ah2(i) --
3. -(t)om -(t)or(i) -(t)o(i) -(t)ah2m -(t)ah2r(i) -(t)ah2(i) -(t)odh(i)

1. -- -h2mWor(i) -h2mWo(i) -- -h2mWah2r(i) -h2mWah2(i) -h2mWodh(i)
2. -h2tWom -- -h2tWo(i) -h2tWah2m -- -h2tWah2(i) --
3. -h2tom -h2tor(i) -h2to(i) -h2tah2m -h2tah2r(i) -h2tah2(i) -h2todh(i)

1. -- -mWor(i) -mWo(i) -- -mWah2r(i) -mWah2(i) -mWodh(i)
2. -dhWom -- -dhWo(i) -dhWah2m -- -dhWah2(i) --
3. -ntom -ntor(i) -nto(i) -ntah2m -ntah2r(i) -ntah2(i) -ntodh(i)
-rom -ron(i) -ro(i) -rah2m -rah2n(i) -rah2(i) -rodh(i)


The original shape of the subject markers would have been:

1. *-ka > *-h2-
2. *-tka > *-th2-
3. *-a > *-0- [~ *-t-]

(1. *-k-mu-) > *-h2m(W)-
(2. *-k-tku-) > *-h2t(W)-
(3. *-k-) > *-h2- [~ *-h2t-]

1. *-mu > *-m(W)-
2. *-tku > *-dh(W)-
3. *-u > *-0- [~ *-nt- ~ *-r-]

The dative markers were:

1. *-ma > *-m-
2. *-na > *-r-
3. *-a > *-0-

1. *-k-mu > *-h2m(W)-
2. *-k-na > *-h2r-
3. *-k-a > *-h2-

1. *-mu > *-m(W)-
2. *-tku > *-dh(W)-
3. *-u > *-0-


PIE NOUN

For athematic stems, the nominal declension in the singular can be
reconstructed as follows:

C-stems i-stems u-stems
nom. *-s *-i-s *-u-s
acc. *-m *-i-m *-u-m
voc. *-0 *-ey-0 *-ew-0, *-ow-0
n. *-0 *-i-0 *-u-0
gen. *-és, *-os, *-s *-ey-s, *-y-os *-ew-s, *-ow-s, *-w-os
dat. *-éy *-ey-ey *-ew-ey, *-ow-ey
loc. *-i *-ey-i > *-e:y *-ew-u > *-e:w, *-ow-u > *-o:w
ins. *-éh1 *-y-éh1 *-w-éh1
(abl. *-ot)

The thematic stems follow a slightly different model:

nom. *-o-s
acc. *-o-m
voc. *-e
gen. *-o-syo
dat. *-o-ey > *-o:y
loc. *-o-i
ins. *-o-eh1 > *-o:h1
abl. *-o-ot > *-o:t

The fundamental division is between the strong cases (nom., acc., voc., neuter
NVA) with asyllabic endings, and the weak cases (G, D, L, I, Ab) with syllabic
endings.

In the weak cases, the sylable of the desinence caused the accent to move one
syllable to the end, so that we have (for nouns with initial accent):

N *h2ák^-mo:n "stone"
A *h2ák-mon-m
(V *h2ák-mon)

G *&2k^-mén-os
(D *&2k^-m(e)n-éy)
L *&2k^-mén-i
(I *&2k^-mn-éh1)

Nouns with final accent have:

N *p&2té:r "father"
A *p&2tér-m
V *p&tér

G *p&2tr-és
D *p&2tr-éy
(L *p&2tr-í)
I *p&2tr-éh1

It can be seen that the dative (*-éy) probably arose as an end-stressed variant
of the locative (*-ey > *-i), while the instrumental, before the soundlaw *-ét

> *-éh1, looked like an end-stressed variant (*-ét) of the ablative (*-ot).

The genitive also shows stress-induced variants (stressed *-és, unstressed
*-os, static *-s), but they never crystallized into separate case-forms. The
explanation is that end-stressed (hysterodynamic) nouns were almost always
animate, so that an opposition came to be felt between animate dative *-éy and
inanimate locative *-ey, and between animate comitative *-ét and inanimate
ablative *-ot.

So for an earlier stage, I would reconstruct:

nom. *-z
acc. *-m
voc. *-0
gen. *-as
DL *-a
IAb *-at

Nominative *-z is required by thematic *-os (nom. +-s would have given thematic
+-es), and by certain voicing effects in laryngeal stems (Skt. sanaj- "old man"
< *sén-ah2-z, etc.). The same voicing occurs in the inanimate neuter marker
*-d (< *-to), and perhaps in collective marker *-h2 (if **-G). It applied then
to any voiceless consonant in an asyllabic ending (except when sentence final,
as in the verbal endings). At an even earlier stage, we can thus reconstruct
*-tu, the animate/ergative form of the demonstrative pronoun, and nominative
*-z is parallel to neuter pronominal *-d < **-ta, the inanimate/oblique form of
the demonstrative pronoun.

Accusative *-m is the same genitive/accusative marker *-ma that we saw in the
personal pronouns, except that it was agglutinated before the loss of final
vowels.

Both the nominative (article) and the accusative were added to the bare stem.

The vocative/neuter in *-0 can be derived from an absolutive in **-a.

The weak cases are built on the absolutive. The genitive suffix was *-sV
(probably *-si, cf. thematic *-sy-o), the instrumental/ablative suffix was
*-ta, the dative/locative suffix was *-a.

nom. **R-tu
acc. **R-ma
voc. **R-a
n. **R-a / **R-ta (pronouns)
gen. **R-a-si
DL **R-a-a
IAb **R-a-ta

The plural forms are:

C-stems i-stems u-stems o-stems
nom. *-es *-ey-es *-ew-es, *-ow-es *-oy, *-o-es, *-os-es
acc. *-n.s *-i-ns *-u-ns *-o(:)ns
n. *-&2 *-i-h2 *-u-h2 *-&2
gen. *-om *-y-om *-w-om *-o-om, [*-oy-s-om]
dat. *-bhios *-i-bhios *-u-bhios *-oy-os, *-oy-bhios
loc. *-su, *-si *-i-su *-u-su *-oy-su
ins. *-bhi(:)s *-i-bhis *-u-bhis *-o:y-s

Only the nom.pl.is a strong case, the others are weak (except that
the acc.pl. has become a strong case in some branches, by analogy
from the singular).

My basic assumptions are:

1. The nom.pl. is *-esW (Arm. -k`), and the *-s that we see in the
acc.pl., dat/abl.pl., loc.pl. and ins.pl. is a shortned form of this
nominative ending *-es(W) (which itself is not subject to zero-grade),

2. The pronominal oblique plural is *-ey-/*-oy- (we saw it in the
p.pronouns *wey-es, *swey-es). The corresponding nominal oblique is
*-bhi-, which in the Auslaut (after the loss of final vowels) gives *-m.
This explains the hesitation between *-bhi-, *-bh- and *-m- in the
plural forms. Plural *bhi is not connected to the postposition *bhi,
which makes instrumentals in Slavic (*-mI), Armenian (-w, -b) and
Greek (*-phi).

At the most remote stage, we would have had a nominal plural:

nom. **-abhu
acc./gen. **-abhi

besides pronominal **-atu, **-ati.

The nominative was eventually replaced by the sg.absolutive (C-stem *-0,
i-stem *-ey, u-stem *-ew/*-ow) followed by the pronominal plural
article *ésW (< **átu).

The oblique developed stress-induced variants **-ém, **-om, **-m.
The second one (*-om) was generalized as the gen.pl., the third one
(*-m), reinforced by *ésW, as the acc.pl. (*-m-esW > *-ms > *-n.s).

New oblique cases were formed by adding singular **-a and **-a(:)t
to the oblique base in *-abhi (in part after it had already become
*-bh(^) and subsequently *-m(^)).

?? LOC *-abhí-a + *-asW -> *-&bhíosW > *-bhiós
DAT *-abh(y)-á + *-asW -> *-&bh(y)ósW > *-bh(i)ós
*-amá + *-asW -> *-&mósW > *-mós
?? ABL *-abhí-a:t + *-asW -> *-&bhíot&sW > *-bhió(t)s

INS *-abh(y)-át, *-am(y)-át + *-asW -> *-&bhyét&sW > *-bhí(t)s
*-am(y)-át + *-asW -> *-&myét&sW > *-mí(t)s

(I withdraw earlier, more complex explanations for these forms)

The amalgamation of dat. and abl. pl. was allowed to exist,
but a new loc.pl. was made by suffixing *-sW to the endingless loc.sg.,
and adding stressed *-í. The resulting forms are *-sí and *-sú.

For the thematic stems, I reconstruct:

Nom. *-á-atu > *-o-sW (> *-os-es in Gmc. *-osiz and Skt. *-a:sas)
Obl. *-á-ati > *-oy

The plural in *-o-es > *-o:s is an analogical formation, made by adding
*-es to the thematic vowel *-o-.

The acc.pl. and gen.pl are analogical after the athematic stems (*-o- + *-ns
and *-o- * -om).

The plural oblique *-oy- is the basis for Dat/Abl *-oy-os (later displaced
by *-oy-bhios or *-o-bhios), the Ins.pl. *-oy-s > *-o:y-s (Szemerényi
lengthening?) and the Loc.pl. *oy-sW-i > *-oy-s-u.


PIE also had two ways to make collectives. One was simply by adding the
collective marker *-h2. The other (mostly used together with *-h2),
consisted in shifting the accent one syllable to the right and lengthening
the accented vowel. Because of the rule prohibiting two consecutive long
vowels, if the original stressed vowel had been long, it was shortened.

CV'C-CVC -> CVC-CV':C
CV':C-CVC -> CVC-CV':C


Some examples:

"water"
**wá:d-an, **wa:d-án-âs > *wód-r, *wédnos
coll.
**wad-á:n+h2, *wad-a:n-ás > *udó:r, *udéns

"dawn"
**h2áws-as-z, **h2áws-as-m, **h2aws-ás-âs > *h2áuso:s, *h2áussm., *h2ussós
coll.
**h2aws-á:s+h2, **h2aws-a:s-ás > *h2usó:s, *h2uséss

"winter"
**gh^áy-âm-z, **gh^ay-ám-âs > *gh^éyo:m, *gh^yémos
coll.
**gh^ay-á:m+h2, **gh^ay-a:m-ás > *gh^yó:m, *gh^yéms

"earth"
*d(h)í:gh^-am, **d(h)i:gh^-ám-âs > *d(h)é:gh^m, *d(h)gh^mós
coll.
*d(h)ígh^-á:m-h2, **d(h)igh^-a:m-ás > *d(h)gh^ó:m, *d(h)gh^éms

"dog" (coll.)
**k^aw-á:n-z, *k^aw-a:n-ás > *k^wó:n, *k^wéns ~ *k^úns

Finally, the dual can be reconstructed similarly to the plural, except that
the plural article *esW (*sW) becomes the dual article *oh3(W) (*h3(W)).
The dual oblique stems (*-oy-, *-bhi-) were borrowed from the plural.

nom (anim.) (o-stem *-o-h3(W))
nom (inanim.) stressed *-íh1, unstressed *-ye(:)h1, static *-yh1
gen. *-h3(W)-s > *-&3ws ~ *-ous
loc. *-h3(W)-í > *-&3w ~ *-ou
dat. *-oy-o-h3(W)(+m), *-bhi-o-h3(W)(+m) > *-bhio:(m)
ins. *-oy-h3(W)(+m) > *-oyyum, *-bhi-h3(W)(+m)

The ultimate pre-PIE forms can be given as:

du. nom. **-iku
du. obl. **-iki


=======================
Miguel Carrasquer Vidal
mcv@...