The Finnish declension goes:

sg. pl.
nom. -0 -t
acc. -0 -t
gen. -n -jen
part. -a -ja
ess. -na -ina
iness. -ssa -issa
elat -sta -ista
illat -on -ihin
adess -lla -illa
ablat -lta -ilta
allat -lle -ille
transl.-ksi -iksi
abess -tta -itta

The nom. had no ending.

The acc. has lost its original ending *-m (which in Balto-Finnic had
merged as -n with the genitive ending).

The genitive ending was and remained *-n.

The partitive is derived from a Uralic ablative in *-ta ~ *-da.

The essive retains its Proto-Uralic form *-na.

The inessive, elative and illative cases are related. They consist
of an element *-s- ("inside"), followed by the essive *-na, the
ablative *-ta and the genitive *-n. The same structure is seen in
the elative, illative and ablative, which are based on an element
*-l- "outside", followed by the given case forms:

*-s-na > -ssa *-l-sa > -lla
*-s-ta > -sta *-l-ta > -lta
*-s-n > -h&n > -on *-l-n > -l&n > -lle

The translative and abessive are reconstructed as *-ksek and *-ttak.

Reflexes of the central cases (nom *-0, acc. *-m, gen. *-n, abl.
*-ta/*-da, loc *-na) can be found in most Uralic languages:

Southern Lapp dialects preserve the genitive in -n as well as
the accusative in -m. The essive is *-na, the locative is *-s-na
or *-s-ta in the singular, *-na in the plural. The acc.pl. is -d <
ablative (partitive) *(-j-)-ta/-da.

nom. -0
gen -(&)n
acc. -(&)m
inessive -nV or -s^-tV

nom -0
gen -n'
abl -do
all. -n'en'

nom -0
acc/lat/abl -n&
loc -t&

nom -0
acc -t < ablative
dat/gen -n-ak
superessive -(&)n

Yurak (Nenets):
nom. -0
gen. -?~ < -n
acc. -m(?)
loc. -xV-na, -ka-na
abl. -xV-d(&), -ka-d(&)

nom. -0
gen. -n/-t
acc. -m/-p
loc. -qI-nI
abl. -qa-n/t

The plural:

The Finnish nominative plural retains its PU shape *-t. The oblique
plural base was *-j.

The Finnish gen.pl. goes back to *-j-D-&n, i.e. the genitive marker
*-n added to the plural marker *-t ~ *-D, added to the oblique
plural in *-j.

In Lapp, the G.pl. is -i < *-j, while in in Northern Samoyed, *-j
is the acc.pl. The Nenets gen.pl. is built by adding plural *-t to
the acc.pl.

For Proto-Uralic, we can reconstruct a plural nominative *-t,
and an accusative/genitive/oblique base *-j.

New acc.pl. and gen.pl morphemes were built by combining *-j with
the plural morpheme *-t and with the singular acc/gen markers *m/*ta
and *n:

Samoyed Lapp BFinn VFinn Permic Ugric Hungarian
Nom.pl. -t -k < -t -t -t -- -t (-k)
Acc.pl. -j -j-ta (-j)
Gen.pl. -j-t -j -j-t-n

The dual marker has been preserved only in Samoyed and Ugric.
The Samoyed nominal dual marker is *-kan', Vogul has *-G-, and
Ostyak *-G&n.


The Finnish independent personal pronouns are:

1. minä
2. sinä < *ti-nä
3. hän
1. me
2. te
3. he

We have a suffix *-na which appears in the 1/2.sg.



The Finnish possessive:

1. -ni
2. -si
3. -nsä/-nsa
1. -emme
2. -nne
3. -nsä/-nsa

This goes back to the merger of separate Proto-Balto-Finic possessive
constructions with singular possessum and dual/plural possessum, which
can be reconstructed as follows:

"my house" "my houses"
sg.poss. pl.poss.
1. -mi -n-mi > -n(n)i
2. -ti, -Di -n-ti
3. -sen, -zen -n-sen
1. -mek -n-mek > -mmek
2. -tek, -Dek -n-dek
3. -sek, -zek -n-sek

The Finnish verb:

1. -n
2. -t
3. -:
1. -mme
2. -tte
3. -vat

Suhonen reconstructs for PBF:

1. *-m > -n
2. *-t > -t, -d, -D

3. [imperfect] *-0
[present] *-Ba, *-pa > *-Bi, *-pi
[passive] *-sen, *-zen

The particle *-Ba, *-pa is the ending of the present participle.

1. *-k-me-k > *-mmek, *-me-k, *-me-n
2. *-k-te-k > *-ttek, *-te-k, *-te-n
3. *-Ba-t, *-ze-t

The plural morphemes are -k, -t and -n.

Lapp (Saami)

The possessive:

sg. du. pl.
1. *-mV *-mV-n *-mV-k (or *-mV-t ?)
2. *-tV *-tV-n *-tV-k (or *-tV-t ?)
3. *-sV *-sV-n, *-s-g&(n) *-sV-k (or *-sV-t ?)

E.g. for ac^'c^e "father":

ac^'c^am ac^'c^ame ac^'c^amek
ac^'c^ad ac^'c^ade ac^'c^adek
ac^'c^es ac^'c^eskâ ac^'c^esek

The verb is reconstructed as:

preterite present
1. *-m *-m
2. *-t > *-k *-t > *-k
3. *-0 *-ja [present ptc.]

1. *-mV-n > *-me *-ja-n
2. *-tV-n > *-de *-pa-tte
3. *-gâ *-pa-n

1. *-mV-k > *-mek *-pa
2. *-tV-k > *-dek *-pa-ttit'
3. *-en > *-e *-ja-t > *-ja-k

Sample paradigm from modern Saami ("I knew, I know"), to show the
flexive character of Saami:

preterite present
1. di:ttim dieDam
2. di:ttik dieDak
3. di:Dii dietta

1. di:Diime di:tti
2. di:Diide diettebät'te
3. di:Diigâ diettebâ

1. di:Diimek diettep
2. di:Diidek diettebe:ttit
3. di:tti di:ttik

Mari (Cheremis)

1. -(&)m
2. -(&)t
3. -s^&, -z^&

1. -(&)na
2. -(&)ta, -(&)da
3. -(&)s^t

1pl. -na is from the possessive with plural possessum, which
has otherwise been lost in Mari.

sg. pl.
1. -am, -em -na
2. -at, -et -ta, -Da
3. -es^, -a; -z^e -&t, -at; -s^t

In 3p.sg. -es^, /s^/ is the suffix of the present participle.


possessive (Erza):

sg.poss. pl.poss.
1. -m -n
2. -t -nt
3. -zo, -zE -nzo, -nzE

1. -mok, -mek -nok
2. -Nk -Nk
3. -st, -sk -st, -sk

Some confusion has taken place between these two paradigms. We
can reconstruct earlier:

sg.poss. pl.poss.
1. -m -n-m
2. -t -n-t
3. -zo, -zE -n-zo, -n-zE

1. -mVk -n-mVk
2. -tVk -n-tVk
3. -s-t, -s-k -n-s-t, -n-s-k

Mordvin has a true stative, i.e. nouns can take verbal endings
to express a state (e.g. E. sazorta-no "we are sisters").

1. -n
2. -t
3. -0

1. -ta-no
2. -ta-do
3. -t

The Mordvin verb distinguishes between an indefinite and a definite
conjugation. The definite conjugation shows a full grid of person/
number agreement with subject _and_ object. The forms are:

indef. (Erza/Moksha)
present past
1. -n -n'
2. -t -t'
3. -i -s'

1. -tano, -dano (M: -tama) -n'ek (M: -m'ä)
2. -tado, -dado (M: -tada) -d'e (M: -d'ä)
3. -it' -s't'

First sg. -n does not come from *-m, so is probably derived
from the forms containing *-n- < *-n-m- from the objective
(plural object) conjugation.

In the 3rd. person *-j is the ending of the present ptc.,
*-s' is the preterite marker. The stative has zero ending.

The plural forms go back to *-ta-mo-k, *ta-do-k, *-j-t,
*-s'-t, and *-t, where *-ta- is the stative plural suffix.

definite (Erza)
1. 2. 3. 1. 2. 3.
1. -- -tan -sa -- -tadiz' -sin'
2. -samak -- -sak -samiz' -- -sit'
3. -samam -tanzat -si -samiz' -tadiz' -sinze

1. -- -tadiz' -sin'ek -- -tadiz' -sin'ek
2. -samiz' -- -sink -samiz' -- -sink
3. -samiz' -tadiz' -siz' -samiz' -tadiz' -siz'

definite (Moksha)
1. 2. 3. 1. 2. 3.
1. -- -t'ä -sa -- -t'äd'äz' -sajn'ä
2. -samak -- -sak -samas't' -- -sajt'
3. -saman' -tanza -si -samaz' -t'äd'äz' -sin'ä

1. -- -t'äd'äz' -sas'k -- -täd'äz' -sas'k
2. -samas't' -- -sas't' -samas't' -- -sas't'
3. -samaz' -t'äd'äz' -saz' -samaz' -t'äd'äz' -saz'

It would go too far to analyze all these forms. The 2nd. person
suffix -k is probably from the imperative *-k.


The possessive:

Zyryene (Komi) Votyak (Udmurt)

1. -Ej, -E -e, -I
2. -Id -ed, -Id
3. -Is -ez, -Iz

1. -nIm -mI
2. -nId -dI, -tI
3. -nIs -zI, -sI

We can reconstruct the following paradigms:

sg.poss. pl.poss.
1. *-mV *-nV-mV
2. *-tV *-nV-tV
3. *-sV *-nV-sV

1. *-mVk *-nV-mVk
2. *-tVk *-nV-tVk
3. *-sVk *-nV-sVk

The verb:

Zyryene Votyak
1. -0 -0
2. -n -d
3. -s, -0 -z

1. -m, -mnIm -m(I)
2. -d, -dnId > -nnId -dI, -tI
3. -nIs > -nI, -snI -zI, -sI

1st.p. sg. regularly *-m > -0. Note 2sg. Zyryene *-n.

The plural forms have been influenced by the possessive
(Z. -m-nIm, -d-nId, -s-nIs).

The Ob-Ugric languages:

Vogul (Mansi)

I will follow László Honti's abstract presentation, which is
closer to the historical structure than to the actual phonetic
shape of the forms (see below for an example of an actual Vogul


sg.poss. du.poss. pl.poss.
1. -m -G-m -n-m
2. -n -G-n -n-n
3. -t& -G-n -n-n

1. -ä:m -G-ä:m -n-ä:m
2. -ä:n -G-ä:n -n-ä:n
3. -ä:n -G-ä:n -n-ä:n

1. -w -G-w -n-w
2. -ä:n -G-ä:n -n-ä:n
3. -ä:n -G-ä:n -n-ä:n

subj. sg. obj. du. obj pl.obj
1. -m -(l)-m -G-m -n-m
2. -n -(l)-n -G-n -n-n
3. -0 -t& -G-t& -n-t&

1. -m&n -l-m&n -G-m&n -n-m&n
2. -n&n -l-n&n -G-n&n -n-n&n
3. -G& -t&n -G-t&n -n-t&n

1. -w& -l-w& -G-w& -n-w&
2. -n& -l-n& -G-n& -n-n&
3. -t -t&n -G-t&n -n-t&n

Ostyak (Xanty)

sg. obj du.obj. pl.obj.
1. -Vm -G&L-am -L-am
2. -V(n) -G&L-a -L-a
3. -VL -G&L -&L-L

1. -(V)mVn -G&L-&m&n -L-&m&n
2. -in -G&L-&n -L-&n
3. -in -G&L-&n -L-&n

1. -VGW -G&L-&GW -L-&GW
2. -in -G&L-&n -L-&n
3. -iL -G&L-aL -L-aL

indet. det.sg.obj det.du.obj det.pl.obj
1. -&m -em -G&L-äm -L-äm
2. -&n -e -G&L-ä -L-ä
3. -0 -t&G -G&L -L

1. -m&n -t&m&n -G&L-&m&n -L-&m&n
2. -t&n -t&n -G&L-&n -L-&n
3. -G&n -t&n -G&L-&n -L-&n

1. -&GW -t&GW -G&L-&GW -L-&GW
2. -t&G -t&n -G&L-&n -L-&n
3. -t -iL -G&L-äL -L-äL

Hungarian (Magyar)

sg.poss. pl.poss.
1. -m -i-m
2. -d -i-d
3. -0 -i

1. -nk -i-nk
2. -tok -i-tok
3. -k -i-k

Verb (present):

subj. obj.
1. -Vk -Vm
1>2 -lak
2. -sz -Vd
3. -0 -ja

1. -Vnk -juk
2. -tok -játok
3. -nak -ják

Verb (preterit:)

subj. obj.
1. -ta-m -ta-m
1>2 -ta-lak
2. -tá-l -ta-d
3. -(V)tt -ta

1. -tu-nk -tu-k
2. -ta-tok -tá-tok
3. -ta-k -tá-k


Yurak (Nenets)

1. -w, -m'i
2. -r, -l
3. -da, -ta
1. -m'i?~
2. -r'i?~, -l'i?~
3. -d'i?~, -t'i?~
1. -wa?, -ma?
2. -ra?, -la?
3. -do?, -to?

du.poss. -xVju-/-kaju-
pl.poss. = Acc.pl. (*-j-)

The Nenets "predeterminative" (e.g. ngano-da-r "the boat for you", as
opposed to ngano-r "your boat"), is built according to the structure
ROOT + 3sg.possessive + nom/acc/dat case + 1/2/3 sg/du/pl. possessive

nom. 1. -da-m'i/-ta-m'i, -du:-w/-tu:-w
2. -da-r, -ta-r
3. -da-da, -ta-ta
acc. 1. -da--m'i/-ta--m'i, -du:--w/-tu:--w
2. -da-m-t, -da-m-d, -ta-m-t
3. -da-m-ta, -ta-m-ta
dat. 1. -da-n-/-ta-n-, -da-n-'i/-ta-n'-i
2. -da-n-t, -da-n-t, -ta-n-t
3. -da-n-ta, -da-n-da, -ta-n-ta

Nenets has a true stative, the forms of which are:

present: preterite:
1. -dm, -tm -dams', -tams', -mans'
2. -m, -t -nas', -tas'
3. -0 -s', -c'
1. -n'i?~ -n'ins'
2. -d'i?~, -t'i?~ -d'ins', -t'ins'
3. -xV?~, -k(a)?~ -xVns', -kans'
1. -wa?, -ma? -wac', -mac'
2. -da?, -ta? -dac', -tac'
3. -?, -t -c'

The verb has indeterminate and determinate inflections, the latter
distinguishing the number of the object. There is also a reflexive

intr./indet. det.sg. det.du/pl. refl.
1. -dm?, -m? -w -xVju-/-j(a)-n -w?
2. -n -r -xVju-/-j(a)-d -n
3. -0 -da -xVju-/-j(a)-da -?
1. -n'i?~ -m'i?~ -xVju-/-j(a)-n'i? -n'i?
2. -d'i?~ -r'i?~ -xVju-/-j(a)-d'i? -d'i?
3. -xV?~ -d'i?~ -xVju-/-j(a)-d'i? -xV?
1. -wa? -wa? -xVju-/-j(a)-na? -na?
2. -da? -ra? -xVju-/-j(a)-da? -da?
3. -? -do? -xVju-/-j(a)-do? -d?


In the following, N2 = n/t, M2 = m/p, K2 = k/ng:


1. -mI, -M2
2. -lI, -l
3. -tI, -t

1. -mI:(j)
2. -lI:(j)
3. -tI:(j)

1. -mIT2
2. -lIT2
3. -tIT2

dual possessed: *-kVj-
plural possessed -i:-


1. -K2
2. -ntI
3. -0
1. -mIT2
2. -lIT2
3. -T2

intr. tr.
1. -K2 -M2
2. -ntI -l
3. -0 -tI, -t
1. -ej, -mI:(j) -ej
2. -lI:(j) -lI:(j)
3. -qI -tI:(j)
1. -mIT2 -mIT2
2. -lIT2 -lIT2
3. -tIT2 -tIT2

The Proto-Samoyed forms can be reconstructed as follows:


1. *-mV (Nenets -m'i < 1du.)
2. *-lV/-rV, *-tV
3. *-ta

1. *-min'
2. *-lin'/-rin', *-tin'
3. *-tin'

1. *-mat
2. *-lat/-rat, *-tat
3. *-tVn

The dual possessed was expressed as *-kVj-, and the plural as *-j-.

For the verb, we can reconstruct:

intransitive transitive
1. *-k (Selqup) *-mV
*-tV-mV (North)
2. *-n (Nganasan) *-rV/-lV
*-n-tV (Nenets, Selqup)
3. *-0 *-tV

1. *-min'
2. *-rin'/-lin'
3. *-kan' *-tin'

1. *-mat
2. *-rat/-lat
3. *-t *-tVn, -tVt

The dual object is expressed as *-kVj-, the plural object as *-j-.


Taking all of this together, I arrive at the following Proto-Uralic


1. *-k
Based on the Hungarian objective and Selqup stative.

2. *-n
Based on Zyryen, Ob-Ugric, Magyar (-sz < -*n?) and Samoyed.

3. *-0
Reflected in all branches.

1. *-(g)-m&-n'
2. *-(g)-d&-n'
There is no clear evidence for dual (g), but logically it must have
been there.

3. *-g&n'
This is just the nominal dual ending (Ostyak -G&n, Vogul -G, Samoyed

1. *-d-m&-k
Plural -d- based on Mordvin -ta-mk.

2. *-d-d&-k
Plural -d- based on Mordvin -ta-dk

3. *-t
This is just the nominal plural ending.


1. *-m&
Attested in all branches.

2. *-t&, *-d&
Attested in all branches (Magyar 2sg. obj. -l < *-D).

3. *-sa, *-za
Attested in all branches. Note the different vocalism of the 3rd. sg.
in Samoyed, (Magyar *-a?), Baltic Finnic (*-sen) and Volga Finnic.

1. *-m&-n'
2. *-d&-n'
3. *-s&-n'
Dual endings (Samoyed -n', Ob-Ugric -n, Lapp -n).

1. *-m&-k (*-m&-t)
2. *-d&-k (*-d&-t)
3. *-s&-t (*-s&-n)
Plural endings patterning as -k, -k, -t (BFinnic, Lapp?, VFinnic),
-k, -k, -n (Vogul) or -t, -t, -n (Samoyed). This probably reflects
the same phenomenon as the difference in vocalism in the singular.
A possible scenario is that the 3rd. person endings were agglutinated
at a later time than the 1/2 person endings.


1. *-g&s^-m&
2. *-g&s^-d&
3. *-g&j-sa

1. *-g&s^-m&-n'
2. *-g&s^-d&-n'
3. *-g&j-s&-n'

1. *-g&s^-m&-k (*-g&s^-m&-t)
2. *-g&s^-d&-k (*-g&s^-d&-t)
3. *-g&j-s&-t (*-g&j-s&-n)

The distinction between a 1/2p. infix -g&s(^)- (i.e. Ostyak -G&L-) and
a 3p. infix -g&j- (i.e. Samoyed -kVj-) is purely hypothetical, and
based on a possible such differentiation of the plural infix. See


1. *-d-m&
2. *-d-d&
3. *-j-sa

1. *-d-m&-n'
2. *-d-d&-n'
3. *-j-s&-n'

1. *-d-m&-k (*-d-m&-t)
2. *-d-d&-k (*-d-d&-t)
3. *-j-s&-t (*-j-s&-n)

The distinction between 1/2p *-d- and 3p *-j- is not actually seen in
any of the forms with plural object/possessum, where the paradigms
have been completely regularized in different ways (Samoyed and Magyar
*-j-, Ostyak -L- < *-D-, Vogul, Permic, Volga-Finnic, Baltic-Finnic,
Saami *-n-, from first person *-d-m- > *-n(n)-).

However, the distinction survives where it had been analogically
extended to, namely in the Ugric forms with singular and dual object.

We have:

Vogul sg.obj. present / past (kit- "ask")

1 kiti:-m(&) ~ kiti:-l-&m kit&s-m& ~ kit&s-l-&m
2 kiti:-n(&) ~ kiti:-l-&n kit&s-n& ~ kit&s-l-&n
3 kiti:-t& kit&s-t(&)

1 kiti:-l-eEm kit&s-l-eEm
2 kiti:-l-eEn kit&s-l-eEn
3 kit&j-eEn kits-eEn

1 kiti:-l-&w kit&s-l-&w
2 kiti:-l-eEn kit&s-l-eEn
3 kit&j-eEn kits-eEn

In the singular 1/2 subject forms -l- is optional, but it's
required in the 1/2/ dual & pl. subject forms. We have a

sg. 0 - 0 - 0 (or: l - l - 0)
du. l - l - 0
pl. l - l - 0

In the present this appears to be a pattern 0 - 0 -j, but that's
rather because of the present tense formant -j-/-G-.

If we look at the forms with dual and plural object (originally
distinguished by the infixes -G- and -n-, but now completely merged
in spoken Vogul), we see the patterns:

sg. 0 - 0 - 0
du. n - n - n
pl. n - n - 0

where -0- can represent earlier -G- or -j-.

The case is much clearer in Ostyak, where the forms with sg.
object are (w&(j)- "to take"):

1. w&L-em
2. w&L-e
3. w&L-t&G (pattern: 0 - 0 - 0)

1. w&L-t-&m&n
2. w&L-t-&n
3. w&L-t&n (pattern: t - t - 0)

1. w&L-t-&GW
2. w&L-t-&n
3. w&L-i-L (pattern: t - t - i)

Finally, the Hungarian suffix for 1st. person acting on 2nd. person,
-lak, surely also contains an infix *-D- > -l-, which is related to
the -t- in Ostyak and the -l- in Vogul.

Miguel Carrasquer Vidal