A grammatical sketch of Proto-Nostratic
Miguel Carrasquer Vidal, November 2013
Personal pronouns and personal affixes: Basque
While Basque is not usually considered to be related to Afro-Asiatic or a member of Nostratic, and
most of the effort has gone into proving (in my opinion, unconvincingly so far) that Basque is most
closely related to the (North) Caucasian languages, if we focus on the Basque personal markers, the
parallels with Afro-Asiatic are undeniable:
Basque person markers
indep. vb. abs.
1
ni
na2m hi
ha2f
1p gu
ga2
zu
za-
vb. erg.
-da-ga-na-gu-zu-
vb. past
n(en)h(en)genzen-
In the first person singular, we have ni as the independent form, for which we can compare e.g.
Hausa ni. The ergative suffix *-da- may have been denasalized from earlier *-na-, a common
phonological process in Basque, more so in view of the fact that none of the other person markers
show any evidence of suppletion. This implies that the 2 sg. independent pronoun hi is to be derived
from *ki (cf. Hausa ka, ki), and that the 2 sg. masc. ergative marker -ga- comes from *-ka-. Loss of
initial unvoiced stops and lenition in internal position are completely regular in Basque, so the
derivation is unproblematic. Perhaps most striking, in a language such as Basque that otherwise
shows no grammaticalization of gender at all, is the 2 sg. fem. marker -na- < *-ma-, which is similar to
Berber 2 sg. fem. -m and general Afro-Asiatic *kim ou fe . (Hausa kin, Tamazight šm:,
Egyptian - m all from *kim).
In the plural, the similarities in form are less striking. If we compare Basque gu, zu (=/su/) with, again,
Hausa mu, ku, there is, to be sure, a striking similarity in vocalism, but the consonants simply do not
match. The best we can do is perhaps hypothesize a 1pl. inclusive *ni-ku, *n-ku e- ou > gu, and,
somewhat less plausible, a 2pl. *ki-su, *k-su ou-them; vos-otros > *su.
The pattern singular *-i, plural *-u, if indeed shared between Basque and Afro-Asiatic, would have to
be very old. Another plural pattern (addition of *-n) is also seen in Basque (gen-, zen-) and e.g. Hausa
(mun, kun, sun). The Basque 2 sg. past tense form hen- may reflect *kim again, analogically
transferred to the 2 sg. masc. and eventually (not in all dialects) to 1 sg. ne(n)-. Cf. the Hausa markers
of the perfective verb (in Hausa, the -n has spread to the fourth perso i perso al a, pf. an):
Hausa person markers
indep. obj. dat.
1
ni
minì/mîn/mûn
2m kai
ka
makà/maː
2f k
ki
mikì
3m š
ši
masà
3f ita
ta
matà
1p m
mu manà
2
k
ku
mukù
3
su
musù
poss.
-wa
-kà
-kì
-sà
-tà
-mù
-kù
-sù
pf. vb.
k
kin
mun
kun
sun
Apart from the pattern sg. *i ~ pl. *-u(n), the most remarkable features of the Hausa (and other
Chadic) person markers are (1) the contrast between 1 sg. n- ~ 1pl. *m-, and (2) the suppletive 3 sg.
masc. marker on the perfective verb.
The latter marker (relatively isolated within Chadic) is derived from the PAA prefix conjugation prefix
*yV-, and will be discussed under the verbal paradigms.
The former contrast (to which the 1sg. possessive marker -wa may perhaps be added), is extremely
interesting. From an Afro-Asiatic perspective, one might speculate that *nu- became *mu- through
labialization caused by the following vowel *u. From a Nostratic perspective, however, the preferred
option would be to speculate that *mi became *ni in the singular by fronting/palatalization, and that
Chadic pl. *mu represents the original consonantism.
If we take the former option, we can reconstruct the following personal pronouns for the oldest layer
common to Afro-Asiatic and Basque:
*Proto-Afro-Asiatic-Vasconic person markers
Person 1.
2.
3.
sg.
*mi > *ni *ki
*si
(fem)
*kim
pl. 1
*mu
*ku *su
pl. 2
*mun
*kun *sun
Personal pronouns and personal affixes: Afro-Asiatic
We can now take a wider look at the person markers in Afro-Asiatic, in so far as they do not originate
in the verbal paradigms:
Afro-Asiatic person markers
Tamazight Beja Egyptian
Hausa
suffix
suffix suffix 1 suffix 2
-j
-wj
1 ni
-i
-ø
(*-ya?) (*-wVy)
-k
-k 2m ka
-š *-ka) -ka
(*-ku) (*-kuw)
-m, -šm
-m
2f ki
-ki
(*-kim)
(*-ki)
(*-kim)
-f
-sw
3m ši
-s
-s
(*-su) (*-suw)
-s
-sj, -st
3f ta
-s
-s
(*-si)
(*-siy)
-n
-n
1p mu(n) [-:γ]
-n
(*-ina) (*-ina)
2m ku(n) -:un
-kna
(*-kina) (*-kina)
2f ,,
-:
,,
,,
,,
-sn
-sn
3m su(n) -sn
-sna
(*-sina) (*-sina)
3f ,,
-snt
,,
,,
,,
indep.
Akkadian
oblique suffix
--
yâti
- -ya, -ni
k(u)âta
(*kuwat-)
m
k(i)âti
(*kimat-)
swt
š
(*suwat-)
stt
ši i
(*siyat-)
i i
--
-ka
-ki
-š
-ša -ši
- i - i i
--
k
i
--
ki
i
-kina
--
š
i
-š
--
ši
i
-kunu
-ši a
If we compare these forms with the hypothetical Proto-Afro-Asiatic-Vasconic ones given above, the
principal additional feature seems to be the addition of case markers to the personal pronouns, used
mainly to provide gender marking in the 2nd and 3rd person forms:
PAA case marked personal pronouns
1.
2.
absol
-ø
*ni
*ki
subject -u
(* i-u > *yu) *ki-u > *ku
object -a
* i-a > *ya
*ki-a > *ka
pl.
-ø
(*mun)
(*kun)
subject -u
(*mun-u)
*kun-u
object -i-, -a *nin-a
*kin-a
3.
*si
* i- *š
* i-a > *ša
(*sun)
*sun-u
*sin-a
The palatalizations * ia > *ya and * i , * ia > *š *ša (secondarily extended in Semitic to the plural
forms š
ši a), seem to confirm the primacy of the singular forms with *-i.
Subject forms of the 1st person seem to be largely absent from PAA. In the second person, there is
another wide-spread pattern, which opposes 2 masc. *ka to 2 fem. *ki. While the masculine might in
principle be derived from *ku + oblique -a > -ka, it seems more likely that this pattern in fact
predates the opposition *ku ~ *ki: we see it in Chadic, Berber, Cushitic, Semitic, and it also resurfaces
in the verbal paradigms. In any case, both patterns largely took the place of earlier *ki ~ *kim1, which
survives only in Chadic, Berber and Egyptian (and Basque).
Verbal paradigms: Afro-Asiatic
The prefix conjugation is widely attested in Afro-Asiatic (except Chadic):
PAA prefix conjugation
1 *ʔV22 *tV-a
2f *tV-i
3 *jV1p *nV2 *tV(pl)
3 *jV(pl)
Hausa, as we remarked before, may have preserved the 3rd person singular prefix in the
of the 3rd person perfective verb.
pronoun
That the prefix conjugation is the oldest verbal formation in Afro-Asiatic is shown by the fact that the
various Afro-Asiatic suffix conjugations can probably all be derived from verbal forms to which prefix
conjugated forms of an auxiliary have been added.
1
nd
One possible explanation for the peculiar 2 person feminine form is if we depart from a reduplicated form
*ki-k, which would regularly develop into *kiŋ, and then perhaps *kim in Afro-Asiatic. A more remote
possibility would be to link it to the feminine-diminutive suffix *-iq (PIE *-ih2): *ki-iq > *kiN > *kim.
2
Different vowels (*a, *i, *u) appear in different, contradictory, contexts in the different Afro-Asiatic
languages. The question of their original distribution is interesting, but falls outside the scope of this paper.
A clear case is Beja (Northern Cushitic), where the present tense suffix conjugation is clearly derived
from the verbal stem + the past te se for s of the er to sa :
Beja prefix and suffix conjugations
to sa
past
pres
1
a-ni, a-di
a-ndi
2m ti-niy-a, ti-diy-a ti-ndiy-a
2f ti-ni-i, ti-di-i
ti-ndi-i
3m i-ni, i-di
i-ndi
3f ti-ni, ti-di
ti-ndi
1p ni-ni, ti-di
n-eeyad
2
t-ee-na, ti-dii-na t-eeyad-na
3
ee-n, i-dii-n
eeyad-na
to start
past cnt present
iiyid
yak-ani
t-iiyid-a yak-tiniya
t-iiyid-i
yak-tinii
iiyid
yak-iini
t-iiyid
yak-tini
n-iiyid
yak-nay
t-iiyid-na yak-teena
iiyid-na
yak-een
past
yak-an
yak-taa
yak-taayi
yak-iya
yak-ta
yak-na
yak-taana
yak-iyaan
past cnt.
yak-i
yak-tiya
yak-tiyi
yak-i
yak-ti
yak-ni
yak-tiina
yak-iin
prefix
*ʔa*ti- -a
*ti- -i
*yi*ti*ni*ti- -na
*yi- -n(a)
While the Beja past and past continuous forms cannot be derived from forms of the er to sa , the
endings certainly do seem to reflect suffixed auxiliaries of the prefix conjugation, with the verbal
stem of the auxiliary reduced to a vowel (past *aa) or nothing (past continuous).
Helmut Satzinger, in Semitic suffix conjugation and Egyptian stative: A hypothetic morpho-syntactic
scenario of its origin, makes the following remarks on the Cushitic suffix conjugations:
For long time it was thought that Cushitic has not any traces of the ancient Afro-Asiatic suffix
conjugation(1). Where its verb forms were conjugated by endings, these were obviously the
relics of an ancient prefix- o jugated au iliar er o para le to ‘ei is h er u
substantivum(2)).
Suffi Co jugatio i Ts a akko(3).
1sc
2sc
3sm
3sf
1pc
2pc
3pc
to dri k
úg-í
úg-dí
úg-í
úg-dí
úg-ní
úg-dè
úg-è
to eat
ží -ì
ží -tì
ží -ì
ží -tì
ží -nì
ží -tè
ží -e
Scheme:
verb * V-aux
verb *t V-aux
verb *y V-aux
verb *t V-aux
verb *n V-aux
verb *t V-aux-…
verb *y V-aux-…
However, Giorgio Banti has made plausible that there is in East Cushitic languages another
paradigm of suffix conjugation that goes back to the common origin of the Afro-Asiatic suffix
conjugation(4):
1sc
2sc
3sm
2sf
Saho
ba
(with tense marker, -ó/-á)
bi -ó
bi -ó
b-á
b-á
Somali cusub /
cúsbi(5) I
cúsbid
cusúb
cusúb etc.
úb/,
e
e
Banti reconstructs the personal morphs of this paradigm in this way:
.
1sc
2sc
1pc
2pc
*-i-yi (~ -i-yu)
*-i-tu
*-i-nu
*-i-tin
As for the use of this conjugation, he says,
In Saho-Afar and in Somali the reduced paradigm is characteristic of a lexically
defi ed group of er roots i ludi g to e or opula, to ha e a d to la k ,
emotion- og iti e er s su h as hate , lo e , k o , a d adje ti al o epts su h as
hite , red , e , lo g , ad .
(1)
Except Bedawye, with its conjugated adjectives ; cf. O. Rössler, ZDMG 100 [25]
(1950), 483–484.
(2)
L. Reinisch, Das persönliche fürwort und die verbalflexion in den chamitosemitischen
sprachen, Wien 1909, in particular, e.g., 308.
(3)
M. Mous, Cushitic, after G. Savà, A grammar of T amakko, (Kuschitische Sprachstudien,
22), 146.
(4)
G. Ba ti, Evidence for a Second Type of Suffix Conjugation in Cushitic, i Proceedings of
the Fourth International Hamito-Semitic Congress (Current Issues in Linguistic Theory 44),
123- 8; G. Ba ti, Ne perspe ti es o the Cushiti Ver al “ ste , i Pro eedings of the
twenty-seventh annual meeting of the Berkeley Linguistics Society, March 22-25, 2001: A.
Simpson (ed.), Special Session on Afroasiatic Languages, 1-48, in particular p. 6.
(5)
The ending of the 1st person is *- , like the suffix pronoun. Note that no Cushitic
language has an absolute pronoun in *-k, like Egyptian ink, Akkadian a k , Biblical
Hebrew a ōk , Berber nekk.
The ai differe e et ee Ba ti s two types of suffix conjugation in Cushitic is that in the first
type the suffixes are also added to the 3rd person, while the second type omits them there, as in the
Berber-Egyptian-Semitic stative-perfective. The endings are otherwise similar to each other, and
similar to the endings of the stative-perfective, except for the 1st person singular, where *-ku has
been replaced by the possessive ending *- .
If we equate the Cushitic suffix conjugations (especially type II), with the Berber-Egyptian-Semitic
stative-perfective, it stands to reason to derive all of these endings from a suffixed auxiliary or
copula, itself prefix-conjugated. For the copula, we can think of a form akin to Ugaritic kwn, Arabic
pf. كانk a, impf. يكونak
. That would explain the curious vocalism of the stative forms, with
*-u in the 1 sg. and the plural, *-a in the 2m, and *-i in the 2f. If *-a and *-i mark the masculine and
feminine, respectively, *-u must be part of the verbal root, which then must have been *ku. In the
2nd person, *ku-a and *ku-i give *ka and *ki (presumably by way of *kwa and *kwi). The first persons
contain the original unextended root vowel (*ʔV-ku, *nV-ku3). The plural forms may have lengthened
*ku or pluralized *kun.
The *k of the copula can be seen directly in Akkadian 1sg. - k Ge ez sg. -ku and all second person
forms (-ka, -ki, -kəm, -kən), Egyptian 1sg. -kw, and Tamazight 1sg. -γ. We can presume that in the
second person, the cluster *tk was reduced to *t outside of South Semitic (where it was reduced to
*k). In the 1 sg., as noted by Satzinger, Cushitic replaced the original ending with pronominal - . In
the 1st person plural, expected *-n-ku is not attested, but does appear (apparently metathesized to
*-knu) in the 1st person plural independent pronouns based on the stative endings (Tamazight : i,
3
It is a peculiarity of the prefix conjugation that in the 1 plural, the verbal root is not pluralized (because there
is no need to: the 1 sg. prefix ʔ- is already distinct from 1 pl. n-).
Rendille inno, Semitic * a
Egyptian forms: ntk, ,
). Note that in these forms, the cluster *-t(ə)k- is clearly present in the
(analogically also 3rd person ntf, nts, ntsn).
Additional evidence for *k (i.e. 2nd person *-tk, first person *-k ~ *-mk/*-km) comes from outside
Afro-Asiatic, as we shall see (e.g. for the 2sg.: Kartvelian *tki-, PIE *-th2, Eskimo *-ken ~ -ten, perhaps
Chukchi -tku).
Summarizing, we can give the following overview of the stative conjugation in Afro-Asiatic:
Afro-Asiatic stative endings
Akkadian Hebrew Arabic
1
- k
-tu
2m - a
-ta
2f - i
-t
-ti
3m -ø
-ø
-a
3f -at
-at
1p 2m -tem
-tum
2f - i a
-ten
-tunna
3m 3f -na
Ge'ez
-ku
-ka
-ki
-a
-at
-ən
-kəm
-kən
-u
-
Egyptian Tamazight
-kw
ø-γ
-tj
t-d
t-d
-j
i-ø
-tj
t-ø
-wjn
n-ø
t-m
-twjn
t-mt
-wj
ø-n
-tj
ø-nt
E Cushitic
*-i-yi, -yu
*-i-tu
*-i-tu
*-i-nu
*-i-tin
PAA
*-a-ʔ-ku
*-a-t-ku-a
*-a-t-ku-i
*-a
*-at
*-at-n-ku (?)
*-at-t-ku-(a)n
*-at-t-k(u)-in
= pl.
= pl.
Afro-Asiatic independent pronouns based on the stative
Tamazight Rendille Egyptian
Akkadian
1 nk:
an(i)
jnk (*ʔanakV) a k
2m šg:
at(i)
ntk
a
(*[ʔa]ntaku)
2f šm:
a
(*[ʔa]ntaki)
3m nt:a
us(u)
ntf
(*[ʔa]ntasu)
3f nt:at
i
nts
(*[ʔa]ntasi)
1p
: i
inno
jnn (*ʔananV)
* a
2m ( : i)
atin
attunu
(*[ʔa]ntakina)
2f ( : i i)
attina
3m nitni
i o
ntsn
(*[ʔa]ntasina)
3f nitnti
If the stative/perfective is built on prefix-conjugation forms of the copula, that means that the prefix
conjugation is older than the suffix conjugation (stative/perfective), which in turn is older than the
Northern Nostratic conjugations with suffixed (possessive) pronouns, such as the PIE mi-conjugation.
This has interesting consequences for Proto-Nostratic word order, which may have originally been
SVO or VSO instead of SOV.
Personal pronouns and personal affixes: Kartvelian
We will start by listing the person markers in the different Kartvelian languages:
Old Georgian person markers
indep.
poss.
1
me
m
2
š
š
3
1p
v
v
2
tkven
tkven
3
obj..
mgxgwgx-
subj.
vxøv-t
x-t
ø-en
Zan person markers
indep.
poss.
1
ma(n)
kim
2
si(n)
skan, ckan
3
1p
ki
ki
2
tkva(n)
tkvan
3
obj.
mk/g/røm-(r)t
k/g/r- -(r)t
ø-
subj.
vøøv-(r)t
ø-(r)t
ø-en
Svan person markers
indep.
poss.
1
mi
mišg -wi
2
si
isgu, -wi
3
1x näj
išg
1i
g šg
2
sgäj
ig
3
obj.
mxngwx-
subj.
xx(l-)
xwlx(l-)
-šd
-šd
-šd
-x
Kartvelologists reconstruct three different series of s(h)ibilants, the reflexes of which are the
following:
Proto-Kartvelian s(h)ibilants
Proto-Kartvelian Georgian
I s z c
ʒ s z c
II
ʒ s z c
III š ž
š ž
Zan/Svan
ʒ s z c
ʒ š ž
šk -- k
sk -- ck
The reflexes of *š etc. before *w are Zan/Svan sk, et .
ʒ
g
ʒg
Ga
reliʒe s rule .
This is why Geo. morrespo ds to )a kim- and Svan -šg i, and why Geo. š - th
corresponds with Zan skan-/ckan- and Svan (i)sgwi.
However, the traditional reconstruction is problematical. Why would plain *š have given /šk/ i
three out of four la guages
h are the i ila ts series III so ofte fou d efore *w?
A better solution seems to me to allow only plain sibilants (traditional
ʒ) and shibilants
(traditional
ʒ) in Proto-Kartvelian, and to explain the (less frequent) third series as arising
from earlier labialized *sw/*šw, *zw/*žw, *cw/* w, * w/* w and *ʒw/* w, with delabialization in
Georgian, and *w > g in Zan and Svan (as in neighbouring Armenian).
If we further assume that the - - series arose out of palatalization of earlier k- -g, we can rewrite
the personal pronouns as Proto-Kartvelian:
Proto-Kartvelian person markers
indep.
poss.
obj.
1
*mi
*kwi-m*m2
*si
*tki(n)*g3
--*x1x *naj (?)
*kwin*m-, *n-?
1i
(*kwan)
*kwin*gw2
*tkwan
(*tkwin-)
*g3
--*x-
subj.
*xw*x*ø*xw*xw*x*ø-
-t
-t
-t
-en
We seem to have two sets of person markers: one based on the pronouns *mi, *ki (and *si?),
reflected in the singular independent pronouns (and perhaps the 1 pl. exclusive of Svan?) and most
of the object prefixes, another apparently based on the stative endings (*ku-, *tk-), presumably by
way of independent personal pronouns patterned like the ones we saw above in Afro-Asiatic
(*ʔa k etc.): these are found in the plural independent pronouns, and in the possessive and
subject prefixes of Kartvelian. Note that Kartvelian shares the vocalism of the stative first person
singular *ku (perhaps, as stated above, simply the original shape of the copula) with Semitic (e.g.
Akkadian šarr k I a ki g , P“e . *ʔa k I .
It would seem that the object prefixes were once in a voicing context (*k > *g, *kw > *gw), while the
subject prefixes were in a spirantizing environment (*kw > *xw, *tk > *x). Both sets of prefixes
occupy the same slot in the verbal chain in the attested Kartvelian languages, so no obvious
explanation is available.
The second person independent pronoun *si could in principle be derived both from original *ti and
from *ki by assibilation (to *ci or * i) and subsequent simplification to *si. Given the object prefixes
*m- and *g- < *k-, which are based on prefixation of the personal pronoun, it seems preferable to
opt for *ki, making Kartvelian truly intermediate between Southern Nostratic Nikia *ni-*ki and
Northern Nostratic Mitia *mi-*ti.
Personal pronouns and personal affixes: Indo-European
In my articles on the Pre-PIE personal pronouns and the Pre-PIE verbal endings (in two parts), I
reconstructed the Pre-PIE inventory as:
Pre-PIE and PIE person markers
indep.
vb.
vb. stat.
*mú
*-mu
*-k(a)
1
2
3
1
2
3
*ma
*mu-m á
*tú
*ta
*tu-m á
-*sa
*su-m á
*m -átu
*m -áti
*m -atu-m á
* -átu
* -áti
*[t]u-atu-m á
-* -áti
*su-atu-m á
vb. dat.? PIE
*-m
*-tu
*-tk(a)
*-n
*-ø-t(a)
*-(a) (+tu)
*-ø
*-m -átu
*-m -án
*-mtku
*-m
*- -á[tu]
*- -án
*-tku
?
*-án-t(a)
*-án(+tu)
*-ati ?
(*h1 g
*me
*me, *meme
*tú
*te
*twe,*tewe
-*se
*swe, *sewe
w
*més
*wéj* -mé ~ *nos
w
*(j)ús
w
*s éj*us-mé ~ *wos
--*smé ~ *swos
w
*-m (i)
*-h2e
*-m
*-s (i)
w
*-th2e
*-r
*-t(i)
*-ø , *-s
*-ø
*-més (i)
*-wén(i)
*-me, *-médh
*-m
*-té
*-tén(i)
*-e, *-dhwé
?
*-ént(i)
*- r *-r *- r
*-j
w
Compared to the PAA and Kartvelian person markers, Indo-European (and Northern Nostratic in
general) shows complete case marking on the personal pronouns, with subject *-u and object *-a in
the singular, and in the plural abs. *-án, subject *-átu, and object/oblique *-áti (the plural forms
being built on the old plurals in *-u). The corresponding dual forms (not shown above) were *-íŋ,
*-íku, *-íki.
Important differences are of course second person *tu, *ta (as opposed to PV, PAA and PK *ki) and
first person *mu, *ma (as opposed to PV, PAA *ni, but like PK *mi). In Pre-PIE, and Northern
Nostratic in general, the case-marked second person pronouns *ki-u, *ki-a apparently underwent a
forward shift to (*k , *ka > ) *tu, *ta, followed analogically by the dual/plural forms (* át-, * ík-).
The third person pronoun survived in PIE as the reflexive pronoun, which appears only in the oblique.
Another significant loss was that of 1 sg. *mu, replaced by *h1 g, a form which is best explained as
deictic *h1e- followed by the stative 1sg. suffix *-k4. The peculiar lenition in the Auslaut is comparable
to what we also see in the pronominal n. sg. *to-d < * -t, the nominal nom. sg. *-(o)-z(w) < *- -t(w),
but not in any verbal forms (presumably because of the SOV character of PIE).
For a discussion of the phonetic developments leading to the PIE forms, I refer to my articles
mentioned above. One additional comment may be added here: in my article on the personal
pro ou s, I said: Equally unclear is the ultimate origin of the dative enclitic forms (*moj, *toj, *soj).
The ending is *-oj, not *-oi as e.g. in the locative, as shown by Slavic mi, ti, si, OLith. mi, ti, si and
Greek
,
. Perhaps we can think of pre-forms *ma i * a i a i .
The structure is therefore the same as that of PSem. *ʔa - k I . In my article on the personal pronouns, I
derived *h1 g instead from deictic *h1e- followed by the emphatic particle*-g.
4
Personal pronouns and personal affixes: Uralic and Altaic
Uralic inherited an inventory of person markers roughly similar to that of PIE.
Saami-Baltic Finnic person markers
Saami
Estonian
indep. poss. vb.
indep.
1
mon
-n
-n
ma, mina
2
don
-t
-t
sa, sina
3
son
-s
-ø
ta, tema
1d moai -me -(m)me
2
doai
-de
-(d)de
3
soai
-ska -ga
1p mii
-met -(m)met me, meie
2
dii
-det -(d)det
te, teie
3
sii
-set
-je
nad, nemad
indep. obl.
minusinuta-, tema-
vb.
-n
-d
-(b)
Finnish
indep.
minä (mä)
sinä (sä)
hän (se)
mei-me
me
tei-te
te
nende-, nei- -(va)d he (ne)
vb.
-n
-t
-ø
poss.
-ni
-si
-nsä
-mme -mme
-tte
-nne
-(va)t -nsä
The Baltic Finnic independent pronouns *minä, *tinä > sinä, with oblique *minu-, *tinu-, are
divergent, compared with the usual set *mon, *ton, *son (presumably from *mu, *tu, *su with the
pronominal oblique (< genitive?) marker *-n, or *-in).
The plurals (mii, meie, me; dii, teie, te; sii) reflect the original plural oblique in *-ati > *-ej.
The verbal plurals -mme, -tte (in part also transferred to the possessive) do not reflect the normal
pronominal/possessive conjugation, but are instead from stative-intransitive *-(t)mək, *-(t)tək (in
Saamic, we have -mmet, -ddet in the plural, while the dual forms are said to reflect *-men, *-den <
*-miŋ, *- iŋ).
The Finnish possessive represents the merger of two originally distinct paradigms: one with the
possessed in the singular, the other with the possessed in the plural:
1
2
3
1
2
3
*-mi
*-ti
*-sen
*-mek
*-tek
*-sek
*-n-mi > -(n)ni
*-n-ti
*-n-sen
*-n-mek
*-n-dek
*-n-sek
-ni
-si
-nsä
-mme
-nne
-nsä
Volga-Finnic person markers
Mordvin (Erzya)
indep. poss.sg. poss.pl.
1
mon
-m
-n
2
ton
-t
-nt
3
son
-zo
-nzo
1p mi
-mok
-nok
2
i
-nk
-nk
3
i
-st
-st
Mari
indep.
myj
tyj
tudo
me
te
nuno
poss.
-m
-t
-žo
-na
-ta
-š
vb.
-m
-t
-0,-š
-na
-da
-t
The Mordvin independent pronoun plurals may reflect the original duals *miŋ, * iŋ, * iŋ. The
possessive markers still distinguish the forms with singular possessum from those with plural
possessum. As in Finnish, the stative plural endings *-(m)mək and *-(t)tək have extended from the
verbal into the possessive domain, both in Mordvin (-mok, -nok; -nk) and in Mari (-na; -ta).
The Mordvin verbal system distinguishes intransitives from transitives with 1-2-3 singular and plural
objects, forming a conjugational grid of high complexity (despite obvious syncretisms):
Mordvin person grid
subj 1 obj
1
-an
2
-at
-samak
3
-i
-samam
1p -tano
2
-tado - ami
3
-i
- ami
2 obj
-tan
3 obj
-sa
-sak
-tanzat -si
- ad i
-i k
-sink
- ad i
-i
1pl obj
- ami
- ami
- ami
- ami
2pl obj 3pl obj
- ad i
- i
- i
- ad i
- i
-tad i
- i k
- i k
- ad i
- i
Figuring out the prehistory of this grid is left an exercise for the reader.
Permian person markers
Komi
indep. poss. vb.
1
me
-ë(j) -ø
2
te
-ïd
-n
3
sijö
-ïs
-ø
1p mi
-nïm -am
2
ti
-nïd
-ad
3
najö
-nïs
-nï
Udmurt
indep.
mon
ton
so
mi
ti
soos
poss.
-e,-ï
-ed,-ïd
-ez,-ïz
-mï
-tï,-dï
-sï,-zï
vb.
-ø, -m
-d
-ø, -z
-mï
-dï
-ø,-zï
The most remarkable aspect of the Permian person markers is the Komi 2 sg. ending -n, which it
shares with Ob-Ugric and Samoyed.
Ugric person markers
Proto Ob-Ugric (Honti)
indep.
poss./vb.
1
*ääm, *mää *-m
2
* äγ * γ
*-n
3
*θöγ
*-ø
*-θ
1d *miin
*-miin
2
*niin
*-niin
*-tən
3
*θii
*-γ
*-θii
1p *mään
*-γw
2
*nään
*-n(Vn)
3
*θVVn
*-t
*-θ Vn)
Hungarian
indep. vb. intr.
én
-k -m
te
-sz -l
-ø,-n
vb. tr.
-m
-d
-ja
mi
ti
k
-juk, -unk -unk
-játok
-tok
-ják
-juk
-unk
-tok
-nak,-k
poss.sg.
-m
-d
-ja
poss.pl.
-im
-id
-i
-ink
-itok
-ik
In Ob-Ugric, the second person is characterized by *n instead of general Uralic *t, not only in the
Auslaut (as also seen in the Komi and Samoyed 2sg. verbal endings), but in the Anlaut of the
independent pronoun and in the du. and pl. suffixes as well (a form *-tən is however preserved in the
2 du.). Perhaps this was triggered by the phonetical development *s > *θ > t in Ob-Ugric, which
would have caused a merger of the 2nd and 3rd person pronouns, had it not been for nasalization in
the 2nd person pronouns, perhaps initially in the dual (*tiin > *niin) and plural (*tään > *nään), finally
also, and unphonetically, in the singular form (* äγ * äγ). The existence of the 2sg. verbal (stativeintransitive) ending *-n might also have played a role in the latter analogical development.
The origin of 1pl. *-γw is obscure. Perhaps it reflects the original stative-intransitive ending PUr
*-(m)mək.
Hungarian has -sz (/s/) instead of *-n where we would have expected it in the stative-intransitive
paradigm. The development is obscure. The regular development of 2sg. *-t seems to be -l, as
witnessed by the intransitive ending -l, and the special form for 1st person subject acting on 2nd
person object -lak (= 2 obj. *-la- + 1 subj. *-k).
Apart from Selqup, Hungarian is the only Uralic language to have preserved the stative-intransitive
1sg. ending *-k. Since *k is lost intervocalically in Hungarian, we can be certain that the ending was
originally *-k, not *-kV.
The intransitive plural endings -unk, -tok may well continue PUr. stative-intransitive *-(m)mək,
*-(t)tək, but since Hungarian has generalized -k as the plural marker everywhere (in the verb, in the
noun and in the pronoun, except for 1-2pl. mi, ti), there is no way to tell.
The Hungarian possessive paradigm differs from the Finnic one in that the plural possessed is marked
by *-i- rather than *-n-, although 2sg. -d might conceivably continue plural possessum *-nt-. More on
this below, after due attention to the Samoyed forms.
Samoyedic person markers
Proto-Nenets
vb.
poss.sg. poss.pl
1
-m
-mə
-inə
2
-n(tə) -rə
-itə
3
-ø
-ta
-ita
1d -mi
-mi
-i i
2
-i
-ri
-i i
3
-kə
-i
-i i
1p -mat -mat
-inat
2
-tat
-rat
-itat
3
-t
-ton
-iton
Nganasan
indep. vb.
mənə -m
tənə
-ŋ
sïtï
-ø
mi
-miˁ
ti
-riˁ
sïti
-kəj
mïŋ
-muʔ
ïŋ
-ruʔ
ï ïŋ
-ʔ
Selqup
p. sg. p. pl. indep.
-mə
-jnə man
-rə
-jtə
tan
-tu
-jtu
tëp
-miˁ - iˁ mee
-riˁ
- iˁ
εε
- iˁ
- iˁ
tëpääqI
-muʔ -jnuʔ mee
-ruʔ -jtuʔ εε
- ŋ - ŋ tëpït
poss.
-mï
-lï
-tï
-nII
-lII
-tII
-mït
-lït
-tït
intr.
-k
-ntï
-ø
-mII
-lII
-qII
-mït
-lït
-tït
tr.
-m
-l
-tï
-mII
-lII
-tII
-mït
-lït
-tït
The Samoyedic intransitive verb (the transitive conjugations are identical to the possessive) shows
two forms which go back to the old stative: 1st person *-k (in Selqup), and 2nd person *-ŋ (Nganasan,
e te ded ith or al possessi e *-tə in Nenets and Selqup).
The forms with dual and plural possessum (c.q. verbs with dual and plural object) show a number of
anomalies when the possessum/object is in the plural:
1. first person *-m- becomes *-n-;
2. second person *-t- does not weaken to -r- or -l-, but remains as -t-;
3. the third person suffix shows different vocalism from the 1-2 forms (*-ta < *-sa; *-ton < *-son
vs. *-mə, *-rə; *-mat, *-rat);
4. the plural possessum/object is reflected as -i- (< *-j) before the suffixes of the
possessor/subject.
We saw similar phenomena above, in the cases of Baltic/Volga-Finnic (where 1st person *-n- has been
generalized) and Hungarian (which has generalized *-j-, like Samoyed).
Everything can be explained if we depart from the oblique plural *-ati of the possessum/object,
which develops regularly to *-j in the Auslaut, but remains as *-t- in the Inlaut. In the 1-2 persons, the
suffixes of the possessor/subject were added at an early stage, causing reduced vocalism in the
suffix, and assimilation of the cluster *-tm- > *-(n)n- (2nd person *-tt- remained). In the third person,
the suffix was added later, with full vocalism, and after the development *-ati > *-ðj > *-j.
The PUr. reconstruction is:
Uralic plural possessum/object
PUralic
Fennic
Hungarian
1 *-t-mə
*-ni
-i-m
2 *-t-tə
*-nti
-i-d
3 *-j + *sá
*-n-sen
-i
1 *-t-mat
*-m(m)ek -i-nk
2 *-t-tat
*-ntek
-i-tok
3 *-j + * án *-nsek
-i-k
Samoyed
*-i-nə
*-i-tə
*-i-ta
*-i-nat
*-i-tat
*-i-ton
In Fennic, 1st person *-n- was generalized. In Samoyed and Hungarian, *-j- was generalized, but the
effect of earlier *-t- is still (partially) apparent on the consonant (*-n-, *-t- instead of *-m-, *-r- in
Samoyedic; -d instead of *-l in Hungarian).
In summary, the Uralic person markers can approximately be reconstructed as:
Proto-Uralic person markers
possessive sg.
indep. tr. sg. object
1
*mon *-mə
2
*ton
*-tə
3
*son
*-sa
1p *mej
*-mat
2
*tej
*-tat
3
*sej
*-son
1d *miŋ *-miŋ
2
* iŋ
*- iŋ
3
* iŋ
*- iŋ
possessive pl.
tr. pl. object
*-t-mə
*-t-tə
*-j-sa
*-t-mat
*-t-tat
*-j-son
*-t-miŋ
*-t- iŋ
*-j- iŋ
possessive du.
tr. du. object
*-k-mə
*-k-tə
*-k-sa
*-k-mat
*-k-tat
*-k-son
*-k-miŋ
*-k- iŋ
*-k- iŋ
stative
intr.
*-k
*-ŋ
*-ø
*-tmək
*-ttək
*-t, *-n
*-kmək
*-ktək
*-k
The only form that requires some additional commentary is 2sg. *-ŋ. If this form is correctly
reconstructed as the original stative-intransitive ending (and it indeed does occur almost exclusively
as an intransitive verbal marker: Komi -n, Ob-Ugric *-n, Nenets -n(tə), Selqup -n-tï, Nganasan -ŋ), then
it stands in place of expected *-tk. A development *-tk > *-nk > *-ŋ in the absolute Auslaut would be
rather unremarkable. If the -r that we see in the IE middle endings is in origin also a 2nd person
(dative) marker *-n < *-ŋ, then IE shows reflexes of both -tk > *-ŋ and *-tk-> *-th2- (and *-tku- >
*-dhw-). A second person ending -ŋ or -n also occurs in Turkic. The Eskimo-Aleut 2sg. ending -n,
however, simply reflects *-t by regular soundlaw5.
A language which some researchers have grouped together with Uralic as a sister language (UralicYukaghir) is the East Siberian Yukaghir. If we look at the Yukaghir person markers, we find the
following data:
5
The final nasalization of voiceless stops is (in Selqup) or relatively recently was (in Eskimo-Aleut) a synchronic
sound law. I believe it coincidentally was also a soundlaw much much earlier, in Pre-Proto-Nostratic.
Yukaghir person markers
indep. vb. intr
1
met
-ŋ
-m
2
tet
-k
-k
3
tude
-ø
-ø
1p mit
-l i
-uok
2
tit
-mut
-mut
3
titte
ŋ -, -ŋi ŋ -
vb. tr
-ø
-k
-ø
-j
-mk
ŋ -
-ŋ
-k
-ø
-j
-mk
-ŋa
The independent pronouns indeed look as if they might be related to Uralic (or any other Northern
Nostratic language), although I see no obvious comparandum for the element -t. The verbal endings,
however, except for intr. 1sg. -m (and perhaps -ŋ) and 1pl. -uok (?), look nothing like the endings we
would expect to see (3sg. -ø does t ou t .
Based solely on the verbal endings, therefore, I would have to conclude that Yukaghir is not closely
related to Uralic, nor even a Nostratic language. The independent pronouns, of course, may have
been borrowed from a Uralic-like language.
e o tur to the Altai personal pronouns, a thorny subject to which I have devoted a separate
paper (The Altai perso al pro ou s). To make a long story short, my conclusions in that paper
were that Turkic, Mongolian and Tungusic share a sound law *mw- > b- (except when another nasal
follows). In Turkic, as in Basque, the sound law has a wider scope and applies to unlabialized *m- as
well. Turkic and Tungusic (but not Mongolian), additionally share a second sound law (with IndoEuropean, Japanese and Ancient Greek): *tw- > s-.
Turkish person markers
poss.
verb I verb II
1
-(I)m
-m
-(y)Im
2
-(I)n
-n
-sIn
3
-(s)I(n) -ø
-ø
1p -(I)mIz -k
-(y)Iz
2
-(I)nIz
-nIz
-sInIz
3
-lArI(n) -lAr
(-lAr)
vb. opt.
-yIm
-sIn
-sIn
-lIm
-sInIz
vb. imper. indep
ben
-ø, -sAnA sen
-sIn
o(n)
biz
-(y)In(Iz),
siz
-sAnIzA
-sIn(lAr) -sIn(lAr)
onlar
I the possessi e a d the perfe ti e er al e di gs, the old stati e-intransitive ending *-ŋ has
completely replaced *-t, as e.g. in Ob-Ugric. In the 2pl., *-ŋ has replaced both the original possessive
and the original stative ending, by adding pronominal plural -Iz (< *-ati) to the ending (some dialects
add the nominative ending -Ar (< *-atu) instead (-γar, -ŋar)).
The first plural perfe ti e e di g -k is unexpected. A regular development from stative *-mmək
seems difficult to justify phonetically. Perhaps we can compare the development of the
corresponding 2pl. form, where singular *-ŋ replaced the original ending(s). If the same happened in
the 1 pl., with 1sg. stative *-k replacing *-mmək, and *-k was subsequently lost as a 1st sg. ending
(replaced by *-m from the possessive), then unadorned *-k may have remained as a 1pl. marker,
without the need to pluralize it, as happened with 2pl. *-ŋ > *-ŋi ~ *-ŋar.
Proto-Mongolian person markers
1 *bi, *min-, *nama2 *ti, *tin-, *tama3 *i, *in-, *ima1x *ba, *man1i *bida, *bidan2 *ta, *tan3 *a, *anThe 1st person sg. oblique form *nama-, for expected *mama-, shows nasal dissimilation (in IE, the
accusative 1sg. *meme was likewise dissimilated everywhere except in Indic, but instead to *mene).
The 1st person plural inclusive *bida clearly reflects *bi I + *ta ou pl. .
Unlike Turkic, but like Tungusic, the plural forms are based on the inherited plural absolute in *-án.
Following the example of singular *bi, obl. *min-; *ti, oblique *tin-, the old plurals *man and *tan (as
well as *bidan) acquired new unetymological casus recti *ba, *ta and *bida.
Evenki person markers
indep
vb I
1 b mi -m
2
i -nni, -ndy
3
ŋa
-n
1x b m - -v
1i mit
-p
2
-s
3
ŋar
-ø
vb II
-v
-s
-n, -ø
-vun, -bun
-t
-xun,-lun,-run
-tyn
sg. poss. / (n)
-v
-mi
-s
-ny,-dy
-n
-n
-vun -mun
-t
-ty
-xun -nun
-tyn -tyn
pl. poss.
-v -b
-l -r
-n
-vun, -bun
-ty
-lun, -run
-tyn
The 1st person plural inclusive mit at first sight seems to contradict both soundlaws that we
established for Proto-Tungusic: we have m- instead of *b-, and -t instead of *-s. Initial m- is of course
easily explained: the form is based on oblique *min- e , as sho
the refle es i the other
Tungusic languages: Udehe minti, Solon miiti, Even mut. Other forms are based on nom. *bi(i): Oroch
biti, Negidal bit(ta). The development of *-tw- to *-t- in the Inlaut (as opposed to *s in the Anlaut and
Auslaut) is completely parallel to what we see in Indo-European: *swej-es ou pl. , *-s nd person
sg. suffi vs. *-te nd perso pl. suffi .
Unlike Turkic, but like Mongolian, the plural forms are based on the inherited plural absolute in *-án.
Following the example of singular *bi, obl. *min-; *si, oblique *sin-, the old plurals *mun and *sun
acquired new unetymological casus recti *bu and *su.
Old stative 2sg. *-ŋ, extended with *-t-, as in Samoyedic, appears to be reflected in the verbal
suffix -nni, -ndy.
Proto-TMT person markers (?)
indep.
possessive
1
*m i-n
*-mu
2
* i-n
*-tu
3
* i-n
*-ø
1p *m adi,* m an *-m an
2
* adi, * an
*- an
3
* an
*-t, *-n (?)
stative
*-k
*-ŋ
*-ø
Personal pronouns and personal affixes: Nivkh, Chukchi, Eskimo-Aleut
Nivkh (Gilyak), is spoken on the northern half of Sakhalin Island and in the region of the Amur River
estuary.
Nivkh person markers
Аmur
E. Sakhalin N. Sakhalin
1s
i
i
1d megi ~ mege m ŋ
1x
ŋ
1i mer ~ mir
i
memak
i
ŋ
m r ~ mir
mer ~ mir
mir
2s ʰi
ʰi
ʰi
2p ʰ
ʰi
ʰi
3s if, i
aŋ
3p imŋ ivŋ imɣ ir i
i
in
The pronominal system can easily be derived from a Northern Nostratic prototype, with trivial
developments *m i > ni, * i > * i. The dual forms of the 1st person reflect expected *m iŋ ~
*m iki (>m ŋ , megi), while the plurals can be derived from *m a (> i ) and * a (> ʰi ). First
perso i lusi e
e a d ou
ight o tai the e pe ted ele e ts st person *me- and 2nd person
*-t- > -r-, and a plural marker *-n (mer, m r ).
Chukchi, together with Koryak, Kerek and Alyutor forms the Chukotkan family of languages, spoken
in the extreme NE of Siberia. Itelmen (Kamchadal) is still spoken on the Kamchatka peninsula, and
may be distantly related to the Chukotkan languages.
Chukchi person markers
indep.
S prefix
1
γəm
t-, m2
γət
ø-, q-, n3
ətɬon, ən- ø-, n-, ne1p muri
mət-, mən2
turi
0-, q-, n3
ətri, ər0-, n-, nen-
S suffix
-k
-ø, -tku
-ø, -n
-mək
-tək, -tku
-t
O suffix
-γəm
-γət
-n
-mək
-tək
-net, -t
The 1st and 2nd person suffixes -m, -t are added in the singular to an element γə-. In the plural,
Chukchi reflects inherited *m adi, * adi as muri, turi.
The 1st person markers have been preserved in the subject prefixes m-, mət-, mən-.
The stative is well preserved in the Chukchi subject suffixes: 1sg. -k, 1pl. -mək, 2pl. -tək (besides
3sg. -ø and 3pl. -t). Only the 2nd person sg. affix has been lost. Chukchi does have two special affixes
for 2nd person subject acting on a plural object (-tku) and a singular object (ine-). Both suffixes also
play a role as intransitivizing (antipassive) markers. Whether there is an etymological connection
between the marker -tku and the inherited stative endings sg. *-tk, pl. *-tku remains to be seen.
All in all, Chukchi preserves the inherited Nostratic person markers with remarkable accuracy.
The Eskimo-Aleut languages are spoken from the tip of Eastern Siberia across the Aleutian islands,
Alaska, Northern Canada to Greenland. Aleut is a separate sub-branch, Eskimoan is subdivided into
Yupik (Siberia and SW Alaska) and Inuit (Alaska to Greenland).
Proto-Yupik person grid
O
S
intr.
1s
2s
3s
1d
2d
3d
1p
2p
3p
1
-ŋa
--
-mken
-ka
--
-mtek
-γka
--
-mci
-nka
2
-ten
-p ŋa
--
-n
-pekuk
--
-γk
-pekut
--
-ten
3
-ø
-aŋa
-aten
-a
-inkuk
-atek
-k
-inkut
-isi
-i
1d
-kuk, -k ŋ
--
-m γ en
-vuk, -puk
--
-m γtek
-γp k
--
-m γci
-puk
2
-tek
-petŋa
--
-jek, -tek
-p γk k
--
-γ k
-ptekut
--
-tek
3
-k
-aγ ŋa
-aten
-ak
-inkuk
-atek
-γk
-inkut
-isi
-ik
1p
-kut
--
-m γen
-vut, -put
--
-mtek
-γp
--
-mteci
-put
2
-ci
-peciŋa
--
-si, -ci
-pecikuk
--
-γ i
-pecikut
--
-ci
3
-t
-a ŋa
-aten
-at
-inkuk
-atek
-γk
-inkut
-isi
-it
There are three basic formations: the intransitive, the transitive/possessive with 3rd person
possessum/object, and the transitive with 1-2 person object.
The intransitive reflects the inherited stative-intransitive:
*-k > -ŋ + -a > -ŋa,
*-tk > -t + -t > -ten,
*-ø
*-kmek > -kuk > -kuŋ
*-ktek > -tek [> -teŋ]
*-ku > -g > -k
*-tmek > -tuk => -kud
*-ttek > -tek => -ci
*-tu > -d > -t
The 1pl. has added plural -t, perhaps by metathesis of original *-tuk. In the 2pl., the ending is
everywhere -ci.
The forms with 3rd person object/possessum are, as was pointed out years ago by Uwe Seefloth,
strikingly similar to the forms we saw earlier in Uralic. The number of the possessum is expressed as
du. *-k- > *-γ, in the plural as *-j before a third person possessor/subject, *-t- elsewhere. The
parallelism with the Samoyedic forms is remarkable, and strongly indicative of a genetic link between
Uralic and Eskimo-Aleut.
The subject endings are -ka, -n, -a; -Muk, -Dek, -ak; -Mut, -ci and -at, probably from *-k (+-a), *-t,
*-(s)a, *-mug, *-teg, *-(s)ag; *-mud, *-ted6 and *-sad. Except for the replacement of *-m (which still
remains as the subject marker in pluripersonal forms) by *-k from the stative-intransitive, the rest of
the endings reflect the PN possessive suffixes faithfully.
The forms with non-third person object are based on the passive participle of the verb + ergative
suffix + possessive suffix + the stative endings expressing the object, where the possessive suffix
fuses with the ergative as 1st person *-m-m- > -m-, 2nd person *-m-t- > -p-.
6
The aberrant development *-ted > *-ci in Eskimo-Aleut can be compared with similar irregularities in the 2pl.
in Indo-Europea , hi h I alled the TUATU-pro le i
paper o the perso al pro ou s.
Confirmation for the 2sg. stative suffix as original *-tk- (ultimately, as we recall, from the 2sg. prefix
*t- plus an auxiliary root *k(u)) is provided by the forms with 2sg. object, which show variation
between -ten and -ken, depending on the preceding sound. We can reconstruct *-tket at the ProtoEA level (cf. the Aleut independent personal pronouns 1. tiŋ < *tik, 2. txin < *tkit).
The Nostratic plural formations
There are a number of pluralizing processes which can be traced to at least a subset of the Nostratic
languages.
One pattern for which there is evidence from at least Afro-Asiatic and Indo-European is the process
which is k o as that of the roke plurals i “e iti , a d as the olle ti e root shape i I doEuropean.
Arabic broken plurals
C1 C2 C3
Type
š á k - l
1
á
- n
2
5
š á k - w a
7
má k - t a b
9
g á m a l
11
12
w á l a d
3
4
8
10
6
13
14
š á
k
-
C1
C2
š - k
š
k
k -
š
k
a g - w í b a
-
k
C3 meaning
“ hap
“pla
“ orkma
“ hair
“offi
“ am l
“bo
k ú t u b
m ú d u n
“bag
“big
“a
“ p
r
“ora or
“book
“ i
Of the fourteen common Arabic patterns given above, seven involve lengthening of the final root
syllable (if necessary (7), shortening the first syllable). The stress shift is automatic in Arabic. The
other patterns (except 3), involve forms where the singular base form is already in the shape
. Here, the reverse process may take place (6, 13, 14), the long vowel being shortened and
the stress shifting to the first syllable.
The first seven patterns find an exact parallel in the IE collectives, which are made by lengthening the
suffix vowel (if the stem vowel is long, it is shortened). Paradebeispiel is the PIE ord for ater :
sg.
gen.
* d-an > * dr
* d-án-âs > *udnós
coll. * d-an > *udór-h2 > d r.
* d- -ás > *udéns
It is e tirel possi le that tra es of the roke plural or olle ti e patter are to e fou d outside
of Afro-Asiatic and Indo-European in the lexicon of other Nostratic languages, if one were to look for
them. I have not done so.
The plural suffix *-ab- has left clear traces in Indo-European and Kartvelian. The original paradigm
can be reconstructed as follows:
The PN *-ab- plural
abs *-am
nom *-ab-u
obl
*-ab-i
In Kartvelian, the suffix -eb- occurs in the indefinite plural inflection:
Georgian noun
sg.
pl. indef.
nom -ø, -i -eb-i
voc -o
-eb-o
erg
-m(a) -eb-ma
dat -s
-eb-s
gen -is
-eb-is
ins
-it
-eb-it
adv -ad
-eb-ad
pl. def.
-n-i
-n-o
-t
-t
-t
---
In IE, too, the suffix occurs only in nominal forms, and is absent (except for late analogical extensions)
from the pronominal paradigms. It is found in the Apl. *-m-s, the Gpl. *-õm (from the thematic
inflection), the DAb pl. *-bhio , *-bhos, *-mos, and the Ins pl. *-bhi(i)s, *-mis (see my paper on the
PIE nominal plurals).
PIE noun
nom
acc
obl
gen
dat/loc
abl/ins
pl.
pl. obl.
Pre-PIE
*-tw
*-mw, *-mwá (pron.)
*-a*-a-sj
*-a+i
*-a-t
*-atw, *-atj (pron.)
*-am, *-abhj-
PIE
*-z > *-s
*-m, *-mé ~ *-wé
*-s, *-es, *-os
*-i, *-ei
*-t, *-eh1, *-ot
*-es, *-s; *-ej, *-j
*-m-, *-om, *-bh(i)- ~ *-m-
There is no clear evidence for this plural suffix outside of Kartvelian and Indo-European. One might
think, for instance, of the Egyptian nominal plural in *-aw, but there are other possible comparanda
for that suffix (e.g. Akkadian - ith le gthe ed ase o el a ordi g to the roke plural patter .
Another possibility is the common Yukaghir plural -pe, but only if that language is indeed Nostratic.
One possible set of cognates involves an important language family which I have not mentioned so
far, and which is traditionally included among the Nostratic languages. The reason why I did not
mention Dravidian in my discussion of the Nostratic personal markers is apparent from the following
table of independent personal pronouns:
Proto-Dravidian personal pronouns
1
1x
m
2
1i
m 2
m
3
3
m
On the surface, these forms do not suggest any close connection with either Nostratic set (*mi ~ *ni,
*ki ~ *ti). The plural marker *-m, however, might suggest a connection with the Kartvelian/IndoEuropean plurals discussed above. Of course, in those languages the suffix is specifically nonpronominal, while in Dravidian the suffix *-m occurs only on pronominal plurals (the nominal plural is
*-ar(u), which can be linked to the PN plural morpheme *-at-, discussed below). We would have to
envisage a scenario whereby the personal pronouns acquired nominal plural endings, and the nouns
subsequently adopted pronominal plural endings (e.g. from a suffixed definite article).
The most widely attested plural morpheme in Nostratic is *-at-, which is present in virtually all
Nostratic languages. In fact, it can be used (even more effectively than the personal marker sets) as a
first-approximation test for the Nostratic affiliation of a language.
The Proto-Nostratic paradigm can be reconstructed as follows:
The PN *-at- plural
abs *-an
nom *-at-u
obl
*-at-i
Besides the many examples that we have already seen above in the plurals of the personal markers,
we can add some remarks about the following plural formations:
In Afro-Asiatic, the definite plural is marked by nunation/mimation *-n (see my paper on Semitic
nunation and mimation). In Akkadian, some nouns have the plural suffix , obl. - i, adjectives
form their plurals with - tu, obl. - ti.
The Kartvelian definite plural has -n-i in the nominative, -t in the oblique, as can be seen in the table
above.
The Indo-European (originally pronominal) plural is marked by *-esw, oblique *-ej < *-atu, *-ati.
In Uralic, the universal plural marker is *-t, oblique *-j.
In Saamic, unadorned *-j still functions as the plural genitive marker. In other Uralic languages, a new
gen. pl. form has been created by adding *-ten (= pl. marker *-t + gen. sg. marker *-n) to the old form
*-j , giving *-jten (> Finn. -jen).
Plural *-n occurs in the pronouns, as we have seen. A peculiar phenomenon in the Finnic
demonstrative pronouns based on the stem *t- is that the initial consonant turns into n- in the plural
(e.g. Finn. tämä this , tuo that , se it , pl. nämä, nuo, ne, obl. *näj-, *noj-, *nij-). The prehistory of
these forms is not clear, but it would be surprising if there was no connection with the n-plural,
perhaps through a prefixed deictic base *V pl. *Vn (sg. *V-ta, pl. *Vn-tan, *Vn-tati > *(V)ta-, pl.
*(V)nan, *(V)naj.
The Turkic suffixes -iz (< *-ati), -lar (< *tatu ?) were already discussed above.
In Mongolian, the plural is generally -d, in Tungusic -l (< -r < -d). Both language groups have *-n in the
plural of the personal pronouns.
The Chukchi noun shows reflexes of *-an and *-ad(u/i). The inanimate plural has -t < *-d, the animate
has -n in the nominative, -r- < *-d- in the oblique.
Chukchi noun
sg. (anim)
nom -ø
erg
-te -ənə
loc
-k(ə) -ənə
dat -γ ə -əna
abl
-γəpə
or
-γ i
pl. anim.
-nti
-ərək
-ərək
-ərək
-rγəpə
-ərəγ i
pl. inan.
-t, -ti
= sg.
= sg.
= sg.
= sg.
= sg.
In Eskimo-Aleut, the regular and universal plural is *-d (> -t), and oblique *-j is attested in the
possessive paradigm.
The dual
For obvious reasons, the dual is less well attested than the plural. We have evidence for the following
paradigm from Indo-European, Uralic, Nivkh and Eskimo-Aleut:
PN (?) dual suffix
abs *-iŋ
nom *-ik-u
obl
*-ik-i
Indo-European neuter NA du. *-ih1 reflects *-iki. The o-stem NA dual ending -oh3 reflects nominative
*-iku. The PIE dual and plural verbal endings *-muán, *-tuán and *-m íŋ, *- íŋ merged as *-wén and
*-tén, but the verbal endings *-muíku, *-tuíku, with delabialized second element, gave rise to the
dual endings *-wáh2/*-wh2-ás, *-táh2/*-th2-ás.
The Uralic dual ending is *-g > *-k. In the pronouns, we have already seen forms with -iŋ > -i . Double
marking of the dual (*-g-iŋ) occasionally occurs (e.g. Khanty -γən, Nganasan -kəj).
The Nivkh 1du. pronoun megi was already mentioned. In Eskimo-Aleut the dual ending is *-g > -k.
Afro-Asiatic also has a dual number, but its formation is completely different:
Afro-Asiatic dual
nom. obl.
-ay
It is therefore doubtful whether the dual suffix *-ik- can be reconstructed for Proto-Nostratic, or
merely for Proto-Northern-Nostratic.
The accusative *-mwa
Another grammatical marker to which at least Northern Nostratic status can be assigned is the
accusative *-m a, which is attested in Indo-European, Uralic and Tungusic.
The form is probably to be derived from the accusative of the demonstrative/interrogative root *mu,
*mu-a.
In Indo-European, the suffix is usually *-m, but in the personal pronouns, it appears as *-mé or *-wé
(sg. *m(w)é, *twé, *swé, du. * h3wé, *uh3wé, *sh3wé, pl. * m , *usmé, *smé).
In Uralic, the accusative marker is *-m.
In Tungusic, the accusative marker is *-ba < *-mwa.
References
Abdel-Massih, 1971, A reference grammar of Tamazight, Ann Arbor.
Abondolo, Daniele (ed.), 1998, The Uralic Languages, London-New York.
Bulatova, N. Ya., 1999, a k a ali ki v kov, St. Petersburg.
Campbell, George L., 1991, omp di m of h World La g ag , London-New York.
Caron, Bernard, 2013, Hausa Grammatical Sketch, CorpAfroAs.
Carrasquer Vidal, Miguel, 2013a, The Pre-PIE personal pronouns, Academia.edu.
Carrasquer Vidal, Miguel, 2013b, PIE nominal plurals, Academia.edu.
Carrasquer Vidal, Miguel, 2013c, The PIE verbal endings, Academia.edu.
Carrasquer Vidal, Miguel, 2013d, The PIE verbal endings, part II, Academia.edu.
Carrasquer Vidal, Miguel, 2013e, Proto-Vasconic phonological system, Academia.edu.
Carrasquer Vidal, Miguel, 2013f, The Basque verb, Academia.edu.
Carrasquer Vidal, Miguel, 2013g, Nunation and mimation in Semitic, Academia.edu.
Carrasquer Vidal, Miguel, 2013h, Th
Al ai
p r o al pro o , Academia.edu.
Erdal, Marcel, 2004, A grammar of Old Turkish, Leiden.
Georg, Stefan, 1991, Some thoughts on the etymology of the Turkic plural suffix -lar/-ler, in: Proceedings of the
Permanent International Altaistic Conference, Oslo.
Georg, Stefan, 2001, Cross-Bering comparisons, Leiden.
Göksel, Aslı a d Celia Kerslake,
, Turkish: a comprehensive grammar, London-New York.
Greenberg, Joseph H., Indo-European and its closest relatives: the Eurasiatic language family, Volume 1.
Grammar, Stanford, 2000.
Jacobson, Stephen, 1979, A grammatical sketch of Siberian Yupik Eskimo, Fairbanks.
Janhunen, Juha , 1982, On the structure of Proto-Uralic, in: Finnisch-ugrische Forschungen, 44 (1982):23-42.
Karlsson, Fred, 1999, Finnish: an essential grammar, London-New York.
Kraft, Charles H. and Anthony Hamilton Millard Kirk-Greene, Teach Yourself Hausa, London.
Kümmel , Martin Joachim, 2009, r a d d I dog rma i h ? ) r Frag d „E ra ia i h “ d
anderer Makrofamilien, Würzburg.
Levitskaja, L.S., 1976, I ori ka a morfologi a vaš kogo a ka, Moscow.
ipi ski, Edward, 1997, Semitic languages: outline of a comparative grammar, Leuven.
Loprieno, Antonio, 1995, Ancient Egyptian: A Linguistic Introduction, Cambridge.
Maslova, Elena, 2003, Tundra Yukaghir and its speakers, München.
Miller, Douglas B. and R. Mark Shipp, 1996, An Akkadian handbook, Winona Lake.
Mous, Maarten, 2007, Cushitic typology, Lyon.
Muravëva, I.A., Daniel, M.A., Ždanova T.Ju., 2001, Chukchi language and folklore in texts collected by
V.G.Bogoraz. Part two: Grammar, Moscow.
Poppe, Nicholas, 1987, Introduction to Mongolian comparative studies, Helsinki.
Rasmussen, Jens. E., 1987, Zur Typologie der Eskimosprachen, in: APILKU 6.
Róna-Tas, András, 2007, Nutshell Chuvash.
Rounds, Carol, 2001, Hungarian: an essential grammar, London-New York.
Satzinger, Helmut, 2004, Statuses and cases of the Afroasiatic personal pronoun, in: Gábor Takács (ed),
Egyptian and Semito-Hamitic (Afro-Asiatic) studies in memoriam W. Vycichl, Leiden.
Satzinger, Helmut, 2012, Semitic suffix conjugation and Egyptian stative: A hypothetic morpho-syntactic
scenario of its origin.
Seefloth, Uwe, 2000, Die Entstehung polypersonaler Paradigmen im Uralo-Siberischen, in: Zentralasiatische
Studien 30, 163-191.
Sinor, Denis (ed.), 1988, The Uralic Languages, Leiden.
Steever, Sanford B., 2006, The Dravidian Languages, London-New York.
Tuite, Kevin, 1998, Kartvelian morphosyntax, München.
Vovin, Alexander, 2011, Fir a d
o d p r o i g lar pro o : a pillar or a pillor of h Al ai
hypothesis?, i : T rk illeri Ara tır aları, . :
-278.
Wedekind, Klaus and Charlotte, and Abuzeinab Musa, 2008, Beja pedagogical grammar.
Zaborski, Andrzej, 2011, The Morphological Status of Verbal Prefix Vowels in Hamitosemitic, in: Aula Orientalis
XXIX-1, 155-176.