3.3.3 Verbal actancy: Agreement

The Udi verb morphology is characterized by a monopersonal agreement system that is basically derived from the clitization of personal pronouns and deictic structures. The grammaticalization process involved has led to minor changes of the personal pronouns whereas the deictic structures are essentially different from the corresponding demonstrative pronouns.

The main function of the agreement clitics is to establish correference with referents in subjective (S) or agentive (A) function (S=A); hence the system behaves accusatively . The underlying strategy of pronominal doubling   has an additional pragmatic function: The agreementclitic (in short AGR-clitic) focuses the constituent to which it is added.Note that Udi AGR-clitics are a) true clitics  (in that they are normally not sensitive to specific word classes), and b)are floating clitics which means thatthe speaker has different options to place a given clitic in a sentence.

NOTE: Alice Harris (2000, in press) has extensively dealt with the

questions of positioning AGR-clitics in Udi. The reader

should in any case refer to these sources to learn more about Udi

clitization strategies. The basic generalizations Harris has made are

paraphrased below.

Udi floating clitics may focus a constituent or the verbal action (see Harris 2000, in press), cp.:

me g/ar gölö-ne o?ne-xa
prox boy:abs much-3sg:s weep-lv:pres
'This boy is weeping very MUCH'.

me g/ar-re gölö o?ne-xa
prox boy:abs-3sg:s much weep-lv:pres
'This BOY is weeping very much'.

me gar gölö o?ne-ne-xa
prox boy:abs much weep-3sg:s-lv:pres
'This boy is WEEPING very much'.

Certain particles necessarily call for an AGR-clitic (because they are in 'natural' focus): te- (negation), gi- (conditional), q'a- (adhortative). Note that the resulting group may float, too.

The position of the clitic on the verb is governed by different options/constraints (see Harris 2000, in press for details). The major constraints are: A verb form marked by the present-future or the optative necessarily calls for the AGR-clitic to follow the tense marker, e.g.

s^e-t'-in me xe-n-ax u?g/-al-le    [no other position allowed]
dist-sa:obl-erg prox water-sa-dat2 drink-fut1-3sg:a
'(S)he will drink this water'.

s^e-t'-in me xe-n-ax u?g/-al-te-ne    [no other position allowed]

dist-sa:obl-erg prox water-sa-dat2 drink-fut1-neg-3sg:a
'(S)he will not drink this water'.

In case there is a (transparent or opaque) incorporated element, the most favored position of the clitic is between the INC element and the (light) verb (see Harris 2000, in press for a detailed description; I make use of the linguistic labels INC and LV applied by Harris 2000,  in press), e.g.:

s^e-t'-a xinär-en s^o-t'-xo xabar-re-aq'-i ...
dist-sa:obl-gen daughter dist-sa:obl-abl news-3sg:a-take-aor
'His daughter asked him...'

'(S)he went'

'(s)he stayed'

But note that the 'future/optative' constraint supercedes this option [ xabaraq'alle 'will ask'], see Harris 2000, in press.

Simple verb stems (without INC-element) allow endoclitization (that is, 'infixation'): the AGR-clitic is then placed before the final consonant of the root in (C)VC-root, or after the vowel in (C)V-roots, e.g.

'(S)he takes'

'(S)he seizes'

'(S)he goes'

This type of infixation is not allowed in case any kind of stem derivational element is present, cp.:

biq'-ne-c-i     [*bi-ne-q'-ec-i]

'(S)he was taken'

xabar-re-aq'-i   [*xabar-a-ne-q'-i]
'(S)he asked'

The functional distinction between infixation (
endoclitization ) and suffixation(enclitization) with simple verbs seems to be related to the question whetherit is the lexical (or semantic) aspect of the verb which is focused, or theTAM-domain [note that e.g. Alice Harris (Harris 2000, in press) disagrees to this assumption], cp.:

'(S)he took'


'(S)he DID take'

It follows that the endoclitization strategy represents the unmarked way of verbal agreement (with simple verbs) - which corresponds to the distribution of endoclitization and enclitization in texts.

The paradigm of AGR-clitics is dominated by case distinction. The accusative behavior of agreement in Udi is matched by the fact that (linguisticallyspeaking) all 'persons' are encoded for ABS and ERG alike. This includesthe third person which - in the (pro)nominal paradigm - is differentiatedby case. Historically speaking, the third person singular clitic (-ne) representsa generlized absolutive whereas the plural marker -q'un (N. -t'un) represents an old ergative case.

Both A and S can be marked in the verb with respect to S-split and A-split. The general technique is to demote A to IO which is echoed in the verb wioth the help of 'dative1'-clitics. Though S=A>IO demotion is in many cases fossilized, some verbs still reflect the old opposition 'control' vs. 'non-control', e.g.

bez baba-x be?-zu-g/-sa
my father-dat2 see-1sg:a-$-pres
'I see (observe) my father'.


bez baba-x a-za-k'-sa
my father-dat2 see-1sg:io-$-pres
'I see my father' (i.e., 'I realize my father')

Some verbs such as ak'sun 'see', va?baksun 'believe', buq'sun 'love' etc.(verba sentiendi) normally call for dative clitics (but notethat esp. inthe Nidzh dialect, the use of dative clitics becomes more andmore obsolete):

t'eg/i bu-za-q'-sa s^ähär-ä es-ax
today want-1sg:io-$-pres town-dat1 go:inf-dat2
'Today I want to go to town'.

'I am cold'

s^o-t'-g/-o qe.?-q'o-b-i k'ua es-ax
dist-sa:obl-pl-dat1 fear-3pl:io-lv-aor home go:inf-dat2
'They were afraid of going home'.

The Udi agreement clitics:







-zu-, -z-, -e.z-

< *-zu




-nu-, -n-

< *-un




-ne- (N. -e-) / -a

< *-n(e)-o





< *-yan





< *-va?n




-q'un (N. -t'un-)

< *-t'g/on



The possessive clitics are mainly used to encode a possessive structure ('to have'), e.g.:

sa eg/el ba-vi-k-i
one sheep be-2sg:gen-$-aor
'You (sg.) had one sheep'.

But note that some speakers prefer the io-clitic instead of the possessive one [sa eg/el bavaki].


Udi has a proper clitic to denote a questioned third person singular (sg.only!) which is –a (see Harris 1992). This element is clitisized to the element under focus:

ema-a bu-t'uq-sa ?
how=much-3sg:q want-3sg:io-pres
'How does (s)he want?'

s^e-n-o-a k'ua ?
dist-sa:abs-abs-3sg:q home
'Is (s)he at home?'

s^in-a zax k'ua ec^-al ?
who:erg I.dat2 home carry-fut1
'who will carry me home?'

3.3.4 Participles and converbs

There are two particples in Udi (both coreference a referent in S=A function). The present-future particple corresponds to the present-future tense marker -al and is always 'agentive' (or 'active'). The past particple corresponds to the aorist tense marker -i and is labile as for its orientation in transitive structures:

besb-i adamar
kill-part:past man
'The man who has killed' or 'the man who was killed' 

In rare cases, both the o-optative and the e-perfect may be used in the sense of a particple, too. Participles can occour both in attributive function or as non-finite forms in subordination. In fact, participles in soibordination share a number of properties with non-finite verbs in serialized sequences. It is important to note that the case of a referent is determined by the valence of the first verb in such a serialized sequence, cp.:

s^ägird-ux ar-i p-i-q'un        [*s^ägirdg/on ari piq'un]
disciple-pl:abs come:past-part:past say-aor-3pl:a
'The disciples came [and] said...'


ar-i s^ägird-g/-on p-i-q'un
come:past-part:past disciple-pl-erg say-aor-3pl:a
'Having come the disciples said...'

Participles normally maintain their verbal valence even if they are nominalized, cp.:

zu arum-ax biq'-i-t'-ux be?g/-i-zu
I:abs wheat-dat2 take.Part:past-sa:obl-dat2 observe-aor-1sg:a
'I watched the one who took the wheat'.

In case the nominalized participle denote a specific referent, it can loose its nominalization marker and may switch to the standard noun inflectiontype:

c^äli-biq'-al-o (pl. c^äli-biq'-al-o-r)
'one who catches fish'


c^älibiq'al (pl. c^älibiq'alux)

Converbs normally are coupled with switch-reference nexcept for telic converbs. The following forms apply:

Final / telic



Final / telic


INF + -an(k') [= BEN?]

Anterior action


Azeri -dan ?

Parallel action

-(a)xun (N. -(a)t'an)

Former locative ?

Posterior action

-am(a), -min,-t'ama, -t'amin
(N. -ma, -me.n, -me.ndz^a)



ta-q'un-sa mo-n-o-r bia-bak-ama

go-3pl:s-pres prox-sa:abs-abs-pl dusk-become-cv:post
'They go until dusk'

t'ia ba-ne-k-i Irod bi-ama

there be-3sg:s-$-aor Herode dead-cv:post
'There he stayed till Herode was dead'.

zu gärämzalug/-axo c^'ebak-axun be?g/-sa-z...

I:abs grave-abl pass=by-cv:par see-pres-1sg:a
'Passing by a grave I see...'

In many instances, converbs are replaced by 'participle + postposition',e.g.:

bütün mal-ux laic/-er-i-t'-uxo os/a
all goods-pl:abs load-lv:past-part:past-sa:obl-abl pp(after)
'After all goods had been loaded...'

In such constructions, coreference is preferred.

3.3.5 Localizing the center: preverbs

The technique of preverbial marking in order to localize the relational center with respect to some landmark is rather restricted in actual Udi. Preverbs that once must have played a major role in this context have nowadays become strongly fossilized. Contrary for instance to Tabasaran, they do not play a role in the TAM-paradigm (e.g., to encode a perfective aspect). The following preverbs can be identified:


into, in


up, on


out, away, around


down, below, under



Some of these preverbs can be linked to either adverbial structures or locative case forms.

3.3.6 Negation

The general technique to encode negation is the clitic te which also denotes 'no!'. It always calls for a AGR-clitic to follow it. te can negate both the relational center (verb), qualifications of this center via adverbs, and referents in the referential center, cp.:

zu me c^äli-n-ax te-zu biq'-e
I.abs prox fish-sa-dat2 neg-1sg:a catch-perf
'I did not catch THIS FISH'.

zu me c^äli-n-ax biq'-e-te-z
I:abs prox fish-sa-dat2 catch-perf-neg-1sg:a
'I DID not catch this fish'.

s^ähär-ä t'eg/i te-z eg/-al
town-dat1 today neg-1sg:s go:fut-fut1
'TODAY, I won't go to town'.

For modal negation see the section below.

3.3.7 Modality

The egenral means to express modaility in Udi is the use of the optative. Additionally, Udi has two modal particles. Both have to be followed by AGR-clitics becuasethey are 'naturally' focused. q'a- denotes an adhortativeand is alwayslinked to the aorist (-i). In negation, it is preceded by the element  ma-, cp.:


'(s)he has to do...'

'they have to praise'

The group (ma-)-q'a-AGR behaves like a floating clitic, cp. 3.3.3.

A conditional is signalized by the element gi- (often > i- ). Its negation isnä-gi- > nä-i- (esp. in Nidzh). It is often copuled with theaorist or the perfect tense. Its favorite place is before the verbal rootor between INC and light verb, cp.:

'if he would not take'

'if they praise(d)'


The imperative paradigm
is restricted to three persons: 2sg, 1pl, and 2pl.and to the light verb esun < 'go'. The following suffixes apply (all of them linked to the suppletive stem ek-):

ek-e        'go!'
ek-en      'we should go!; Let's go!'
ek-nen    'go (plural!'

cp.: take 'go there!' < *ta-ek-e (inf. taisun); q'us^ bak-ek-e 'become a bird!' (said by a sorceress) < bak-esun 'become:pass'; c^ur-eke 'stand!' (inf: c^uresun) etc.

With other verbs, the optative is used to encode the imperative. But note that the second person singular lacks agreement:

a'q-a    'take!' (cp. bakan(u) 'you (sg.) should take')
bak-a    'be!' (cp. bakan(u) 'you (sg. should become')

vi yoldas^-g/-o günäh-g/-ox bag/is^lamis^-b-a
you:gen friend-pl-gen sin-pl-dat2 forgive-lv:trans-imp
'Forgive the sins of your friends!'.

3.3.8 Copula

Udi makes little use of its copula bu. Instead, the personal agreement clitics are used. There are only two tense forms (present marked by zero, and past marked by -i) both for AGR clitics in copula function and the 'true' copula bu, cp.:

me k'odz^ kala-ne
prox house:abs big-3sg:s
'This house is big'

t'e adamar haq'ullu-ne-i
dist man:abs clever-3sg:s-past
'That man was clever'.

ic ^s^ägird-ux t'ia bu-q'un-i
refl:exoph disciples-pl there cop-3pl:s-past
'His disciples were there'.

The AGR-clitic may occasially float even if functioning as a copula, cp.:

zu-z kala
I:abs-1sg:s big
'As for me, I am (the) old(est)'

bez vic^i k'ua-ne
my brother:abs at=home-3sg:s
'my brother is at home'.

Occasionally, the copula bu can appear without AGR-clitics, e.g.:

dostlug/ s^el a^ bu, dus/manlug/ te
friendship:abs good think cop enmity:abs neg
'Friendship is a good think, enmity isn't'.

This example also illustrates the use of the negative aprticle te as a negative auxiliary. te is normally followed by AGR-clitics, e.g.

me k'odz^ kala te-ne amma s^avat'-t'e
prox house big neg:3sg:s but nice-3sg:s
'This house is not big but it is nice'.

The two particles q'a (adh) and gi- (cond) can also be used in copula function, e.g.

un bez yoldas^ q'a-n(u)
you(sg) I:gen firend cop:adh-2sg:s
'Be my friend!'

bez g/ar kala gi-n // bez g/ar-gi-n kala
I:poss son:abs big cop:cond-3sg:s // I:poss son:abs-cop:cond-3sg:s big
'When my son is old...' // when my SON is old...'

The negation of these copula-like elements is canonical (ma-q'a- resp. nä-gi- > näi-). The following table summarizes the copula functions of Udi:









Often nominal predicates