4.8 Coordination

Phrasal coordination often is asyndetic in Udi. The standard pivot is S=A, cp.:

ta-q'un-sa mo-n-o-r is/a yaq'-en bia-bak-ama yaq'-q'un-tai-sa
go-3pl:s-$:pres prox-sa:abs-abs-pl near way-erg>instr dusk-be-cv:till way-3pl:s-go-pres
'They walk on the near road, they walk on the road until dusk'.

ta-q'un-sa bia-bak-ama furu-q'un-exa ek'a-l te-q'un bo?g/a?-b-sa
go-3pl:s-$.pres dusk-be-cv:till search-3pl:a>s-lv:pres what-foc neg-3pl:a find-lv-pres
'They walk until dusk, they look around, they don't find anything.'

pasc^'ag/-un g/ar-en gölö be?-ne-g/-sa me-t'a-laxo gädi-n-en ic^ yallug/-ax ... pi-n-e-bos^ c^ap-ne-xa
king-gen son-erg much observe-3sg:a-$-pres prox-sa:obl-gen-pp(on) boy-sa-erg refl hanky-dat2 ... blood-sa-gen-pp(in) dip=into-3sg:a-lv:pres
'The king's son looked at him; the boy ... dipped his hanky into the blood'.

In the subsequent clause, the center of the attention-information flow (AIF) which normally is S=A is referred to by agreement. The O-domain of a transitive structure in the first clause is often 'zero' in the second one, cp.:

nana-n s/um-ax aq'-i[-ne] a?il-ux ta-ne-d-i
mother-erg bread-da2 take-aor[-3sg:a] child-dat2 given-3sg:a-lv-aor
'Mother took to the bread and gave (it) to the child'.

Asyndetic coordination may be subjected to more integrating strategies with the help of two techniques:

a) Coordinating particles: va? 'and', os/(a) 'and then', amma 'but', focus particle -al :

s^o-t'-g/-o mec^it-un-bos^ bu-ne-i sa adamar tamiznut' elmug/-on biq'-i va? harai-ne-b-i [Mk 1:23]
dist-sa:obl-gen temple-gen-pp(in) be-3sg:s-past one man:abs unclean ghost-erg take-past and cry-3sg:s-lv-aor
'In their temple, there was a man whom an unclean ghost had taken, and he cried.'

so so pasc^'ag/-un g/ar-en zap-i lai-ne-sa os/a gädi-n-en gir-re-b-sa bütün ... dövlät-äx
one:abs one:abs king-gen son-erg pull=up-aor bring=up-3sg-$:pres then boy-sa:-erg gather-3sg:a-lv-pres all ... goods-dat2
'The king's son pulls one after the other out [of the hole] and then the boycollects all [...] goods.'


b) Verb serialization

The question of whether or not Udi possesses true 'serialized' structures still is a matter of discussion. The problem is mainly related to the semantics of the construction, whereas the structural aspects claerly support the serialization hypothesis. It should be noted that serialization in Udi represents an intermediate state between coordination and subordination.

Serialization normally appears with an S=A pivot. The verb of the first clause is les marked than the verb of the second clause with respect to two features: a) it lacks agreement, b) TAM formation is restricted to two domains: PAST vs. nPAST:

	First Verb	Second Verb
Agreement	missing	present (floating)
TAM	PAST vs. nPAST	Full paradigm

Cf.:

Rust'am-en aq'-i vu?g/ q'atir-ax tavar-ax k'iro-n-ax ta-ne-sa t'e c/a?la?g-i
Rust'am-erg take-past seven mule-dat2 axe-dat2 knife-sa-dat2 go-3sg:s-$:pres dist wood-dat1
Rustam takes [having taken] the seven mules, the axe [and] the knife [and] goes to that wood.'

va? s^o-n-o-al aiz-er-i q'adag/a-ne-b-i mus^-n-u
and dist-sa:abs-abs-foc raise-lv:past-past command-3sg:a-lv-aor storm-sa-dat1
'and he rised [and] controlled the storm.'

Note that case makring of the referent in S=A function is triggered by the first verb. In many cases the serialized verb has become semantically weak. This is especially true for verbs of motion. They tend to become grammaticalized as inchoatives, cf.:

tac-i sa dz^ähil g/ar-re biq'-sa
go:past-past one young boy hire-pres
'Having gone he hires a young boy' (> 'he starts to hire a young boy')

hame waxt'-a pasc^'ag/-un g/ar-al ar-i ba-ne-p'-esa
same time-dat1 king-gen son:abs-foc come:past-past arrive-3sg:s-lv-pres
'At the same time, the prince arrives.'

The grammaticalization of such serialized verbs becomes more transparant ifthe case marking trigger switches to the valence pattern of the second verb:

gädi-n-en ar-i pasc'ag/-un g/ar-ax mog/or-re-b-esa
boy-sa-erg come:past-past king-gen son-dat2 wake=up-3sg:a-lv-pres
'The boy now wakes up the prince.'


4.9 Subordination

Udi knows three basic types of subordination:

a) converbs;
b) clausal subordination,
c) participle constructions

Converbs as described in section 3.3.4 often appear in switchreference (with  respect to the S=A pivot), cp.:

g/ar-en gädi-n kex biq'-axun ta-ne-st'a xibq'o manat-ax
son-erg boy-sa:gen hand:dat2 seize-cv:par give-3sg:a-lv:pres 60 rubel-dat2
'When the son takes the boy's hand, he (the boy!) gives him 60 rubel.'

ta-q'un-sa mo-n-o-r is/a yaq'-en bia-bak-ama
go-3pl:s-pres prox-sa:abs-abs-pl close way-erg>instr sunset-be-cv:until
'They walk on the close way till sunset.'

häzir-b-a zenk’ena biasun-un s/um uk-san va? t’ak’-p-i q’ullug/-b-a za, uk-ama va? u?g-/ama, va? os/a uk-a ug?-a un [Lk 17:8]
prepare-lv-2sg:imp I:ben evening-gen bread-dat2 eat-cv:fin and gird-lv-part:past service-lv-2sg:imp I:dat1 eat-cv:until and drink-cv:until and then eat-2sg:imp and drink-2sg:imp
'Prepare [it] so that I can eat the evening meal and having girded yourself, serveme when I eat and when I drink; and then you shall eat and drink.'       

Coreference, however, is not unfrequent:

eg/a-xun c^'e-ne-bak-sa sa gärämzalug/-un bos^t'an
go:fut-cv:par pass=by-3sg:s-lv-pres one grave-gen garden
'While going he passes by a graveyard.'

va? ba-ne-k-i beivan ga[n]-mxox t’e g/e-n-al-cirik’ ic^ ak’-eg/-ama Izrail-a. [Lk 1:80]
and be-3sg:s-$-aor wild place-pl dist day-sa-super-until refl see-pass:fut-cv:until
'And he stayed in wild places until the days when he showed himself to Israel.'

Converbs may be replaced by a nominalized participle + postposition, e.g.

k'ul cip-i-t'uxo os/a p'a?len-al furu-q'un-exa
earth pour=out-part:past-sa:obl-abl pp(after) they=twio-foc walk=around-3pl:s-lv:pres
'After having poured out the earth they two walk around'.

Note that this construction does not necessariliy imply coreference.

Clausal subordination is a typical feature of Udi which can be related tolanguage contact especially with ClassicalArmenian. The Armenian subordinatoret^ce 'that' has be bporrowedinto Udi (te) and hasbecome the general means to signal clausal subordination (thus copying the'Oriental' subordinator ke / ki etc.). te has produced a large variety of subordinators, among them:

manote	relative pronoun	mano 'which one' + te
s^et'abaxt'inte; et'abaxt'inte	because	s^e-t'-a-baxt'-in-te 'dist-sa:obl-gen-sake-erg>instr' + te e-t'-a-baxt'-in-te 'what-sa:obl-gen-sake-erg>instr' + te
s^ute	who	s^u 'who' + te
eq'arate	how much	eq'ara 'how many?' + te
evaxte	when	e-vaxt'-te 'which time' + te
orte	because	or 'which one?' (Armenian) + te
maate	where	maa 'where?' + te
s^ite	whose	s^i (who:gen) + te
mat'abos^te	in which	ma-t'-a-bos^-te 'where-sa:obl-gen-in' + te
etärte	as	e-tär-te 'which kind' + te
uk'ante	as if	uk'-a-nu-te 'say:fut-opt-2sg' + te ('so to say')
andz^ag/te	where	an-dz^ag/-te 'that (Pers.) place (Iran.)' + te
ägänä(te)	if	Persian ägär 'if' + te

The subordinator te also serves to introduce direct speech(indirect speech is uncommon in Udi), cp.:

gädi-n-en exne te zu vaxo kala-zu
boy-erg say:pres-3sg:a sub I:abs byou(sg.):abl big-1sg:s
'The boy says: 'I'm bigger than you.'

Relative clauses (introduced by manote) are not restricted to a specific type of head, cp.:

adamar manote t'ya-ne bak-e
man rel:abs there-3sg:s be-perf
'the man who has been there...'

adamar-en manote t'ya-ne bak-e ic^ as^ s^el-le-b-e
man-erg rel:abs there-3sg:s be-perf refl work good-3sg:a make-perf
'The man who had been there had done his work well.'

adamar-a manote t'ya-ne bak-e bez xunc^i-n-ax bu-t'u-q'-e
man-dat1 rel:abs there-3sg:s be-perf my sister-sa-dat2 love-3sg:io-$-perf
'The man who had been there loved my sister.'

zu adamar-ax be?-zu-g/-sa ma-t'-in-te ic^ as^-n-ux s^el te-ne b-esa
I:abs man-dat2 see-1sg:a-$-pres rel-sa:obl-erg-sub refl work-sa-dat2 neg-3sg:a mak-pres
'I see a man who does not do his work well.'

Likewise, the relative pronoun can encode any kind of relational (or locative/instrumental) behavior, cp.:

adamar, ma-t'-ux-te be?g/-sa-zu 
man rel-sa:obl-dat2-sub see-pres-1sg:a
'The man whom I see.'

adamar ma-t'-u-te bu-za-q'-sa
man rel-sa:obl-dat1-sub olve-1sg:io-$-pres
'The man whom I love.'

pak ma-t'-u-te furu-q'un-p-esa
park rel-sa:obl-dat1-sub walk=around-3pl:s-lv-pres
'The park in which they walk around'

mex ma-t'-in-te s/um-ax k'ac'-zu-exa
knife rel-sa:obl-erg>instr-sub bread-dat2 cut-1sg:a-lv:pres
'The knife with which I cut the bread'

yapug/ ma-t'-a-laxo-te c^ur-re-p-i
roof rel-sa:obl-gen-pp(on)-sub stand-3sg:s-lv-aor
'The roof on which (s)he stood.'

adamar ma-t'-ust'a-te k'odz^-ne    [preferred: ma-t'-ay-te (rel-sa:obl-gen-sub)]
man rel-sa:obl-adess-sub house-3sg:s
'The man who has a house'


Relative constructions may be replaced by particple constructions:

zu beg/-al(a) adamar
I:abs see-part:pres man
'The man I see'

za buq'-i adamar
I:dat1 love-part:past man
'The man loved by me'

t'e s^u mand-i ga-l-a-q'un mand-esa
dist night stay-part:past place-sa:obl-dat1-3pl:s stay-pres
'They stay at that place where they had stayed at night.'

is/a yaq'-en tag/-al adamar-ux te-q'un qai-bak-sa
close way-erg>instr go:fut-part:pres man-pl:abs neg-3pl:s back-be-pres
'The men who take the close way do not come back.'

The examples show that the there is no obvious internal pivot (S=A or S=O) in such constructions. The head may be associated to any kind of relational behavior within the participle construction, e.g.

[ZERO]:S_i VERB:PART HEAD_i

[ZERO]:A_i - O_j VERB:PART HEAD_i

A_j - [ZERO]:O_i VERB:PART HEAD_i

etc.

In transitive structures, the embedded verb is necessarily 'labile': If an A-referent in switch reference is present [O in coreference is ZERO], the participle becomes 'passive'-like, if the A-referent is in coreference [O in switch reference is ZERO], the structure is 'active'-like, cp.:

adamar-en ögmis^-b-i g/ar
man-erg praise-lv-part:past boy
'The boy who has been praised by the man'

adamar-ax ögmis^-b-i g/ar
man-dat2 praise-lv-part:past boy
'The boy who had praised the man'

The general scheme is:

	Embedded A	Embedded O
'Active'	ZERO	OVERT (switch)
'Passive'	OVERT (switch)	ZERO

Note that this scheme only holds for past participles (-i). As expected, present participles are always of the 'active' type [S=A]., cp.:

tag/-al adamar
go:fut-part:pres man
'The man who goes'

c^äli-n-ax biq'al adamar
fish-sa-dat2 seize-part:pres man
'The man who catches fish'

*adamar-en k'al-uk'-al g/ar [recte: g/ar mat'uxte adamaren k'aluk'alle ]
man-erg call-lv:fut-part:pres boy
'The boy who will be called by the man'

If we interpret 'labile' constructions as being typical for an ergative behavior (at least with respect to East Caucasian languages), we arrive at the following gerenal distribution:

nPAST	-al	S=A	accusative behavior
PAST	-i	S=O(=A)	ergative behavior

It should be noted that the -al-participle is sometimes followed by a yet not identified element -a in attributive function,e.g.

me as^-n-ux b-al-a adamar
prox work-sa-dat2 make-part:pres-? man
'The man who does this work...'

As far as I can see, there is no obvious functional difference to the standard type me as^nux bal adamar. It may, however, be speculated that the element -a is related to the AGR-clitic -a (in questions, cf. 3.3.3).