[tied] Morphology (13/20)

From: Miguel Carrasquer
Message: 14994
Date: 2002-09-03

II. NOUNS

13. General introduction

At the most remote stage, pre-PIE may have been an ergative language, with a
triptotic case marking ergative **-u, absolutive **-a, and genitive **-i. The
plural and the dual were made with affixes (nominal plural **-abh-, pronominal
plural **-at-, dual **-ik-), plus case markers (diptotic ergative **-u, oblique
(absolutive-genitive) **-i), giving **-abhu ~ **-abhi, **-atu ~ **-ati and
**-iku ~ **-iki. A collective, similar to the Semitic "broken plural", could
also be made (by shifting the stress one syllable to the end [or failing to
retract the stress one syllable to the front], and by lengthening this stressed
vowel). There was also a collective suffix **-ka, and a diminutive (later
feminine) suffix abs. sg. *-ik(a).

In the singular, the old genitive and ergative fell into disuse, and scarcely
any trace of them can be recovered, except perhaps what can be inferred from the
Caland system (see below under n-stems). A new case system was built using a
new set of case endings. As in the verb, we can distinguish between syllabic and
non-syllabic endings, causing Ablaut variations in the declension (in the usual
case: strong e-grade stem with the asyllabic endings, weak zero-grade stem with
the syllabic endings). The endings can be reconstructed as follows:

Nominative *-z
Accusative *-m
Absolutive *-0 (= neuter / Vocative)

Genitive *-as
Dative-Locative *-a
Instrumental-Ablative *-at

We can distinguish between root nouns, consisting of a single (monosyllabic)
root, and derived (compound) nouns, which have a root and a (syllabic) suffix.
Both root and derived nouns can further be divided into three groups, depending
on the structure of the root: light roots (ending in a single consonant or
resonant), heavy roots (ending in two or more consonants, resonants, or
resonant+consonant), and long roots (with original long vowel in the root).
Most suffixes are light, but the common suffix **-ant is heavy. Derived nouns
could take the accent either on the root (proterodynamic class) or on the suffix
(hysterodynamic class).

For simple consonant stems, this results in a set of paradigms which can be
given as follows. The symbol â (= /a:/) stands for an unaccented final syllable
vowel (followed by a consonant), affected by "svarita lengthening" after a light
syllable:

ROOT NOUNS:
LIGHT HEAVY LONG
N CáR-z CáRC-z Cá:R-z
A CáR-m CáRC-m Cá:R-m
Vn CáR CáRC Cá:R
G CaR-ás CaRC-ás Ca:R-ás
DL CaR-á+i CaRC-á+i Ca:R-á+i
IA CaRát CaRC-át Ca:R-át

PROTERODYNAMIC NOUNS:
LIGHT HEAVY LONG
N CáR-CâR-z CáRC-CaR-z Cá:R-CaR-z
A CáR-CâR-m CáRC-CaR-m Cá:R-CaR-m
Vn CáR-CâR CáRC-CaR Cá:R-CaR
G CaR-CáR-âs CaRC-CáR-âs Ca:R-CáR-âs
DL CaR-CáR-a+i CaRC-CáR-a+i Ca:R-CáR-a+i
IA CaR-CáR-ât CaRC-CáR-ât Ca:R-CáR-ât

HYSTERODYNAMIC NOUNS:
LIGHT HEAVY LONG
N CaR-CáR-z CaRC-CáR-z Ca:R-CáR-z
A CaR-CáR-m CaRC-CáR-m Ca:R-CáR-m
Vn CaR-CáR CaRC-CáR Ca:R-CáR
G CaR-CaR-ás CaRC-CaR-ás Ca:R-CaR-ás
DL CaR-CaR-á+i CaRC-CaR-á+i Ca:R-CaR-á+i
AI CaR-CaR-át CaRC-CaR-át Ca:R-CaR-át

COLLECTIVES:
LIGHT/HEAVY/LONG
n CaR(C)-Cá:R-h2
G CaR(C)-Ca:R-ás
DL CaR(C)-Ca:R-á+i
AI CaR(C)-Ca:R-át

After the working of zero grade, we get:

ROOT NOUNS:
LIGHT HEAVY LONG
N CéR-s CéRC-s CóR-s
A CéR-m CéRC-m CóR-m
Vn CéR CéRC CóR
G CR-és CRC-és CeR-és
DL CR-é+i CRC-é+i CeR-é+i
IA CRét CRC-ét CeR-ét

PROTERODYNAMIC NOUNS:
LIGHT HEAVY LONG
N CéR-CoR-s CéRC-CR-s CóR-CR-s
A CéR-CoR-m CéRC-CR-m CóR-CR-m
Vn CéR-CoR CéRC-CR CóR-CR
G CR-CéR-os CRC-CéR-os CeR-CéR-os
DL CR-CéR+i CRC-CéR+i CeR-CéR+i
IA CR-CéR-ot CRC-CéR-ot CeR-CéR-ot

HYSTERODYNAMIC NOUNS:
LIGHT HEAVY LONG
N CR-CéR-s CRC-CéR-s CeR-CéR-s
A CR-CéR-m CRC-CéR-m CeR-CéR-m
Vn CR-CéR CRC-CéR CeR-CéR
G CR-CR-és CRC-CR-és CeR-CR-és
DL CR-CR-é+i CRC-CR-é+i CeR-CR-é+i
AI CR-CR-ét CRC-CR-ét CeR-CR-ét

COLLECTIVES:
LIGHT/HEAVY/LONG
n CR(C)-CóR-h2
G CR(C)-CeR-és
DL CR(C)-CeR-é+i
AI CR(C)-CeR-ét

And finally, after Rasmussen's initial accent rule <> and Szemerényi lengthening
before *-s and *-h2 {}, we get:

ROOT NOUNS:
LIGHT HEAVY LONG (Static)
N {Cé:R-(s)} CéRC-s {Có:R-(s)}
A CéR-m. CéRC-m. CóR-m.
Vn CéR CéRC CóR
G CR(.)-és CR.C-és <CéR-s>
DL CR(.)-é+i CR.C-é+i <CéR+i>
IA CR(.)-éh1 CR.C-éh1 <CéR-t>

PROTERODYNAMIC NOUNS:
LIGHT HEAVY LONG (Static)
N {CéR-Co:R-(s)} CéRC-CR.-s CóR-CR.-s
A CéR-CoR-m. CéRC-CR-m. CóR-CR-m.
Vn CéR-CoR CéRC-CR. CóR-CR.
G CR.-CéR-os CR.C-CéR-os <CéR-CR.-s>
DL CR.-CéR+i CR.C-CéR+i <CéR-CR+i>
IA CR.-CéR-ot CR.C-CéR-ot <CéR-CR.-t>

HYSTERODYNAMIC NOUNS:
LIGHT HEAVY LONG (Static)
N {CR.-Cé:R-(s)} {CR.C-Cé:R-(s)} CéR-CR.-s
A CR.-CéR-m. CR.C-CéR-m. CéR-CR-m.
Vn CR.-CéR CR.C-CéR CéR-CR.
G CR.-CR-és CR.C-CR-és <CéR-CR.-s>
DL CR.-CR-é+i CR.C-CR-é+i <CéR-CR+i>
AI CR.-CR-éh1 CR.C-CR-éh1 <CéR-CR.-t>

COLLECTIVES:
LIGHT/HEAVY/LONG
n {CR.(C)-Có:R-(&2)}
G <CR.(C)-CéR-s>
DL <CR.(C)-CéR+i>
AI <CR.(C)-CéR-t>

As most animate nouns were either root nouns or hysterodynamic compound nouns,
there developed a tendency to differentiate what were in origin merely phonetic
variants of the oblique cases (G. -és ~ -os, DL. -éi ~ -i, IAb -ét ~ -ot) as if
they were separate cases, eventually resulting in a split between local cases
(generally used with inanimate nouns) on the one hand (locative *-i, ablative
*-ot), and grammatical cases (generally used with animate nouns) on the other
hand (dative *-éi, comitative/instrumental *-ét ~ *-éh1). No such split
occurred in the genitive, where different languages chose either *-és or *-os
(and consequently also *-es and *-ós) as the generalized genitive marker (in
some cases, as in Sanskrit, we cannot tell which). Animated nouns which were
originally in the proterodynamic class formed new secondary datives and
instrumentals, accented on the ending, which, due to their higher frequency
compared to the original locative(/dative) and ablative(/instrumental) cases,
became one of the sources of the so-called amphidynamic declension, with the
strong cases accented on the root, and the weak cases on the ending.

As in the verb, a further category, generally with fixed columnar accent and
without Ablaut, is formed by the thematic nouns and adjectives, for which see
below.

In the plural, the differences between the proterodynamic and amphidynamic types
are less pronounced. Taking two simple (light root) PD and HD noun as examples,
we have:

N *h2ák^-mon-es *p&2-tér-es
A *h2k^-mén-n.s *p&2-tr-n'.s
G *h2k^-mén-om *p&2-tr-óm
D *h2k^-mn.-bhiós *p&2-tr.-bhiós
I *h2k^-mn.-bhí(:)s *p&2-tr.-bhí(:)s
L *h2k^-mn.-sú *p&2-tr.-sú

We see that only the N.pl. is a strong form, but the ending is invariably
unaccented *-es (there are no variants *-s or *-os). The Acc. and Gen. pl. are
weak forms, but here too we see no variation in the endings, which are always
*-ns (< *-ms) [accented even in the HD declension!] and *-om (the common variant
*-o:m presumably imported from the o-stems). The Dat./Abl., Ins. and Loc. are
"doubly weak" and also have the same endings for all declensional categories
(*-bhiós ~ *-mós, *-bhís ~ *-mís [*-bhi:s ~*-mí:s in Avestan and Slavic], and
*-sú [*-sí in Greek]).

The element *-bhi- occurring in the plural cases is mostly identified with the
postposition *bhi "with, by", which is kown to make a (secondary) Ins. sg. in
e.g. Slavic (-mI), Armenian (-w, -b) and Greek (-phi). However, it seems
strange that a postposition should be *followed* by what is clearly a case
ending in *-bhi-ós. Furthermore, in the thematic declension, the place of
*-bhi- is taken by the element *-oi- (*-oios [secondarily *-oi-bhios], *-o:is,
and also Loc.pl. *-oisu), which suggests that both *-bhi- and *-oi- have the
same function, namely that of oblique plural markers.

In my opinion, *-bhi- is indeed originally the oblique nominal plural marker (<
**-abhi), just like *-oi- ~ -ei- (< **-ati) is in origin the oblique pronominal
plural marker. The unextended acc./gen. plural **-abhi would have given PD
*-om, HD **-ém, the former being precisely the PIE Gen.pl. ending. Presumably,
the HD/Root ending *-ém was everywhere replaced by the HD ending *-om or by the
thematic ending *-o:m (unless Gothic *-e:m is a cross between *-ém and *-o:m).

The oblique cases were built on **-abhi, giving:

PD HD
L **-abhí-a D **-abhi-á
Ab **-abhí-ât I **-abhi-át,

later redundantly extended with the plural article *átu (of pronominal origin),
giving:

PD HD
L **-abhí-a-atu D **-abhi-á-atu
Ab **-abhí-ât-atu I **-abhi-át-atu, with haplology:
> **-abhí-âtu > **-abhi-átu

and then:

PD HD
L **-bhí-osW D **-bhy-ósW
Ab **-bhí-osW I **-bhy-ésW > *-bhísW

This explains the merger of Dat. and Abl. in the plural. The Loc.pl. should
have merged as well, but was rebuilt on the model of the singular, where it has
a zero ending, to which the *-sW plural marker was added (abstracted from the
*-sW ending of the Dat/Abl and the Ins.), and further extended, as in the
singular, with the suffix *-i (giving *-0-sw-í > *-sí ~ *-sú).

The same plural suffix *-sW was also used to differentiate the Acc.pl. from the
Gen.pl., which originally were identical (**-âm/*-ám < **-abhi). At the same
time, the Acc.pl., with its ending in *-Vm, was analogically drawn toward the
model of the Acc.sg. in *-m., which resulted in the forms:

(PD) **h2k^-mén-om + -sW -> **&2k^-mén-m. + -sW > *&2k^ménn.s
or: **h2k^-mén-om + -sW -> **h2ák^-mon-m. + -sW > *h2ák^monn.s

(HD) **ph2-tr-ém + -sW -> **p&2-tr-m.' + -sW > *p&2trn.'s
or: **ph2-tr-ém + -sW -> **p&2-tér-m. + -sw > *p&2térn.s

The second variant is found in e.g. Greek, where the acc. pl. is synchronically
a strong acc. sg. with *-s added. There can be little doubt that this is an
analogical simplification, given the existence of the first variant, found in
e.g. Sanskrit, where the Acc.pl. was synchronically an Acc.sg. with stress
shifted one syllable to the end ("weak accusative") and extended with plural
*-s.

The nominative plural (perhaps originally **-abhu > **-om/**-ém or **-ou/**-éu)
was replaced by the absolutive singular (i.e. the vocative/neuter form) followed
by the plural article *átu, which, like the thematic vowel (of which it is in
fact the nom. plural), has no effect on the position of the stress, and is
itself not subject to zero grade or "svarita lengthening" (only "thematic vowel
lengthening" before a voiced segment). Hence the PD and HD Nom. plurals:

**hák^-mon + esW > *h2ákmones
**p&2-tér + esW > *p&2téres

Neuter (inanimate) nouns originally had no plural as such, but used the
collective suffix *-h2, attached either to the collective root (with lengthened
suffix vowel) or to the normal root. In the latter case, we can again
distinguish the following cases:

ROOT
LIGHT HEAVY LONG
CáR-h2 > Cé:R(&2) CáRC-h2 > CéRC(&2) Cá:R-h2 > Có:R(&2)

PD
LIGHT HEAVY LONG
Cár-CâR-h2 CáRC-CaR-h2 Cá:R-CaR-h2
> Cér-Co:R(&2) > CéRC-CR.: > CéR-CR.:

HD
LIGHT HEAVY LONG
CaR-CáR-h2 CaRC-CáR-h2 Ca:R-CáR-h2
> CR.-Cé:R(&2) > CR.C-Cé:R(&2) > CéR-CR.:

The two alternative endings -Co:R&2 and -CR.: are seen in e.g. Sanskrit n.pl.
kárma:ni (< -mo:n&2) and kárma: (< *-mn.:).


The dual

The dual suffix in the oblique (accusative) was *-iki (the erg. ~ nom. *-iku >
**-ih3 is only reflected in the pronominal declension, and in the non-neuter
o-stems). This developed into **-ih1, which, in light PD stems, should have
given *-(y)êh1 > *-yeh1, and in HD stems *-yéh1 > *-íh1. The attested forms of
the consonant stems are:

Skt. mf [-a:(u) < o-stems]
n. -i:
OCS mf. -i
n. [-e^ < o-stems]
Lith. -e
Grk. -e
OIr. *-E

Greek and Lithuanian *-e (*-ye in a form like *okWye > osse, Arm. ac^`k` <
*okWye + sW) can go back to a sandhi form of the proterodynamic ending with loss
of the laryngeal (Lithuanian -e can also represent *-(y)e: directly). The
Sanskrit and Slavic forms point to hysterodynamic *-ih1 > -i:.

In the oblique cases of the dual, *-h3(W) performs the same function as *-sW in
the plural. It comes from a dual pronominal "article" **-iku > **-ih3, appended
to a *plural* oblique (with *-bhi- in the C stems, *-oi- in the o-stems). We
have:

Dat/Abl. *-bhiâ-h3(w) > *-bhyo:(u)
(Ins. *-bhi(:)-h3(w) > *-bhy&3(u), *-bhi:(u) > *bhyou, *-bhi:)
Loc. *-h3(w)-i > *-&3u > *-ou
Gen. *-h3(w)-s > *-&3us > *-ous

Apparently *h3(w) vocalized everywhere as *o(u) in the Loc. and Gen., not just
in Greek. Perhaps we should reconstruct **R?

The genitive dual was formed in a manner similar to the Loc. du. and plural, by
analogy with the singular (ending *-s). We have:

G L DAbIns
Skt. -o:s = gen -bhya: + m
Avest. -å -o: -bya
OCS. -u = gen -ma
Lith. = loc -au -m (= ins)
Grk. = loc -oiiun = loc ( -oiiun < o-stem Loc. *-oi-&3u + -m)


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