The particle ki is almost never used in Azerbaijani finite relative clauses, at least in written texts. We found only two instances in a corpus of some 2000 pages, and they are typically descriptive, not restrictive RCs, as in (67) and (68):
(67)
|
iç-ə-k
|
həmin
|
o
|
məhəbbət
|
körpü-sü-nün
|
sağlığ-ı-na,
|
|
|
drink-opt-1pl
|
foc.prox
|
dist
|
love
|
bridge-pos3-gen
|
health-pos3-dat
|
|
|
|
|
|
|
|
|
|
|
hansı-nın
|
ki,
|
bir
|
uc-u
|
bu
|
dam-da-dır,
|
bir
|
uc-u
|
|
|
which-gen
|
ki
|
one
|
end-pos3
|
prox
|
roof-loc-cop3
|
one
|
end-pos3
|
|
|
|
|
|
|
|
|
|
|
|
|
o
|
dam-da
|
oğlan
|
ev-i-ynən
|
qız
|
ev-i-nin
|
|
|
|
dist
|
roof-loc
|
boy
|
house-pos3-with
|
girl
|
house-pos3-gen
|
|
|
|
|
|
|
|
|
|
|
|
|
ara-sı-nda!..
|
|
|
|
|
|
|
|
|
interval-pos3-loc
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
‘Let’s drink to this bridge of love, [I mean] whose one end is on one of the roofs, the other end on the other roof, between the house of the groom and the roof of the bride.’ (S. Qədirzade, Bənövşə, 1969)
|
|
(68)
|
əlifba-mız
|
(hansı
|
ki,
|
Türk
|
#əlifba-sı-na
|
|
alphabet-pos-1pl
|
which
|
ki
|
Turk
|
alphabet-pos3-dat
|
|
|
|
|
|
|
|
yaxın
|
idi)
|
dəyişdir-il-di.
|
|
|
|
close
|
coppst
|
change-pass-pst
|
|
|
|
|
|
|
|
|
|
‘Our alphabet (which was close to the Turkish alphabet) was changed.’ (F. Ələkbərov, Əlyazmalar Institutu Azərbaycanda erkən əlifbalar, Azerbaijan International jurnalı, 2000)
|
The preceding examples should be considered derivates of the hansı construction without ki, but they seem to have moved one step forward in the direction of forming European-type relative pronouns. This is not directly comparable with the use of ki in Tati head-internal restrictive relative clauses.
The presence of an attributive participial form capable of relativising subjects, objects and most oblique positions in Tati is doubled by the non-participial, head-internal strategy. It can be headed by a pronoun in initial position (cf. [69]), or by a NP made up of the head-noun preceded by the adjective kitam meaning ‘which’ (cf. [70]).
(69)
|
ki
|
ki
|
burvar-tü-ne
|
küş-de-s,
|
|
who(nom)
|
ki
|
brother-pos2-acc
|
kill-perf-3
|
|
|
|
|
|
|
adam
|
di-imu
|
nisdü.
|
|
|
person
|
village-1pl
|
negcop3
|
|
|
|
|
|
|
|
‘The one who killed your brother is not a person from this village.’ (fn)
|
|
|
(70)
|
kitam
|
adam
|
ki
|
burvar-tü-ne
|
küş-de-s
|
|
which
|
person
|
ki
|
brother-pos2-acc
|
kill-perf-3
|
|
|
|
|
|
|
|
ez-i
|
di
|
nisdü.
|
|
|
|
abl-prox
|
village
|
negcop3
|
|
|
|
|
|
|
|
|
|
‘The person who killed your brother is not a person from this village.’ (fn)
|
As in Azerbaijani, the main asset of this strategy is the finiteness of the verb form employed, which allows for a switch in tense and mood categories (cf. [71]).
(71)
|
kitam
|
adam
|
ki
|
burvar-tü-ne
|
kuf-den
|
|
which
|
person
|
ki
|
brother-pos2-acc
|
beat-prs3
|
|
|
|
|
|
|
|
ez-i
|
di
|
nisdü.
|
|
|
|
abl-prox
|
village
|
negcop3
|
|
|
|
|
|
|
|
|
|
‘The person who is beating up your brother is not a person from this village.’ (fn)
|
Tati usually adds the particle ki after the head-noun marked for its syntactic function and before the verb (cf. [70] and [71] above). The matrix clause may contain a resumptive pronoun, in which case we are dealing with a correlative construction, as in (72).
(72)
|
[kitam
|
seg
|
ki
|
dendu
|
bzeren]
|
bo-u
|
sugum
|
medi !
|
|
which
|
dog
|
ki
|
tooth
|
evt.strike.3
|
dat-dist
|
bone
|
proh.give
|
|
‘Do not give a bone to the dog which bites.’ (fn)
|
Most frequently, however, and systematically if the antecedent head has subject function in the matrix clause, the latter contains a gap (cf. [73] and [74]):
(73)
|
kitam
|
nozu
|
ki
|
şir-e
|
xar-de-s,
|
ez-i
|
xune
|
nisdü.
|
|
which
|
cat
|
ki
|
milk-acc
|
eat-perf-3
|
abl-prox
|
house
|
negcop3
|
|
‘The cat who has drunk the milk is not from this house.’ (fn)
|
|
|
(74)
|
kitam
|
xune-re
|
(ki)
|
merd
|
üsde-re-s (...)
|
|
which
|
house
|
ki
|
man
|
buy-perf-3
|
|
‘The house that her husband has bought (does not please his wife).’ (fn)
|
In some instances, a shared argument other than the head noun is repeated (e.g. xune ‘house’ in [75]).
(75)
|
kitam
|
merd
|
ki
|
xune
|
üsde-re-s,
|
xune-re
|
|
which
|
man
|
ki
|
house
|
buy-perf-3
|
house-acc
|
|
|
|
|
|
|
|
|
be-zen
|
xaşden
|
volen-den.
|
|
|
|
|
dat-woman
|
self
|
show-prs3
|
|
|
|
|
|
|
|
|
|
|
|
‘The husband who has bought a house shows it to his wife.’ (fn)
|
Possessors can be relativised, whether or not their possessa are subjects in the relative clause (cf. [75] for a subject possessum and [76] for a non-subject possessum).
(75)
|
e-kitam
|
gede
|
ki
|
piyer-i
|
mür-de-s
|
herey
|
ze-ren.
|
|
abl which
|
boy
|
ki
|
father-pos3
|
die-perf-3
|
shout
|
strike-prs3
|
|
‘The boy whose father has died is crying.’ (fn)
|
|
|
(76)
|
e-kitam
|
kile
|
ki
|
xuniyi-re
|
di-rini
|
e-mü
|
xuvar-mü-nü
|
|
abl-which
|
girl
|
ki
|
house
|
see-prs2
|
abl-1
|
sister-pos1-cop3
|
|
‘The girl whose house you see is my sister.’ (fn)
|
If the head-noun is the possessor of an argument of the matrix clause (e.g. seg ‘dog’ and guşyeyi ‘ear’ in [77]), it appears in the case corresponding to its function in the RC, and is cross-referenced by a possessive suffix on the possessum in the matrix clause:
(77)
|
[kitam
|
seg
|
ki
|
rousden]
|
guş-yeyi-re
|
me-keş !
|
|
which
|
dog
|
ki
|
bark-PRS.3
|
ears-pl.pos3-acc
|
proh-pull
|
|
‘The dog which barks, do not pull his ears!’ (fn)
|
As can be seen in (77), in such head-internal relative clauses ki is used not as a linker morpheme, like in Persian, but as a marker of focus on the preceding word, head of the RC. Without this particle the clause takes an indefinite, more general meaning, thus compensating for the absence of a conditional form (cf. [78] and [79]):
(78)
|
kitam
|
seg
|
rousden...
|
|
which
|
dog
|
bark-PRS.3
|
|
‘Whichever dog barks...’ (fn)
|
|
|
|
(79)
|
be-kitam
|
kar
|
des
|
şen-dunum,
|
be-u
|
qadeğe
|
na-ren.
|
|
|
dat-which
|
work
|
hand
|
throw-prs.1
|
dat-dist
|
interdiction
|
put-prs.3
|
|
|
‘Whatever work I take on, he forbids it.’ (fn)
|
The strategy illustrated above is used whenever the head-noun is highly topical and has to be preposed to the RC, a feature quite systematically associated with the semantic role of (transitive) Agent, and the fronting movement involved by this strategy is a way to mark the head of the RC as a topic. It is, however, not restricted to filling any functional gap left by the disappearance of the Indo-European active participle, since it is also found in relative clauses in which the syntactic position of transitive subject may be relativised.
..2.Pragmatic and (non factual) modal motivation
Grammatically, the question-word using strategy may be called a “primary strategy”, since it allows for the relativization of all syntactic positions including subjects, but pragmatically, it is not dominant and highly marked for an additional value:
The use of the head-internal strategy allows speakers to
-
topicalize the head and the function relativized;
-
express non-factual modality in the relative clause.
Non-factual modality is expressed on the verb, usually in the ‘eventual’14 form; recall that the particle ki is always omitted in this case:
(80)
|
be-kitam
|
ölke
|
m-yo
|
tü-ne,
|
b-re-m.
|
|
dat-which
|
country
|
evt-come.3
|
you-acc
|
subj-go-1pl
|
|
‘Let’s go to whatever country you want.’ (fn)
|
|
|
(81)
|
e-kitam
|
beg
|
kiliyi-re
|
m-xas-di
|
|
abl-which
|
lord
|
girl.pos3-acc
|
evt-want-pst-2
|
|
|
|
|
|
|
bu-ra-m
|
xas-d-um
|
bere-tü.
|
|
|
subj-go-1
|
ask-pst-1
|
for-2
|
|
|
|
|
|
|
|
‘Whichever lord’s daughter you wish I will ask for you.’ (fn)
|
..3.Restrictions on the participial strategy
Three-place predicates give rise to rather heavy participial clauses, and even if the relativised position is that of a subject, the participial construction is accepted as grammatically correct when it is suggested to informants, but the translation first given by our informant uses the head-internal equivalent construction (cf. [82]).
(82)
|
be-ayol
|
muş
|
de-re
|
nozu-re
|
girind
|
|
dat-child
|
mouse
|
give-part
|
cat-acc
|
catch.imppl
|
|
‘Catch the cat which has given a mouse to the child!’ (fn)
|
|
|
(83)
|
kitam
|
nozu
|
ki
|
be-ayol
|
muş
|
de-re-s,
|
u-re
|
girind.
|
|
|
which
|
cat
|
ki
|
dat-child
|
mouse
|
give-perf-3
|
dist-acc
|
catch.imppl
|
|
|
= (82)
|
|
In other cases, the motivation for choosing this strategy was the categorical inability of the Tati De-participle to relativise functions lower down on the accessibility (or ‘Keenan-Comrie’) scale. The participial strategy was rejected, and the head-internal one was judged the only possibility. The functions concerned are those of
-
a Recipient marked by the dative-directive pro-clitic be- (cf. [84]);
-
a Beneficiary introduced by the preposition bere ‘for’ (cf. [85]);
-
animate argument of a verb “take” marked with the ablative proclitic (cf. [86]);
-
comitative with preposition boş (cf. [87]);
-
and possessor of a noun which is not subject of the RC (cf. [88]).
(84)
|
be-kitam
|
ayol
|
ki
|
nozu
|
muş
|
de-re-s,
|
be-u
|
deyş.
|
|
|
dat-which
|
child
|
ki
|
cat
|
mouse
|
give-perf-3
|
dat-dist
|
look.imp
|
|
|
‘Look at the child to whom the cat has given a mouse!’
|
|
(85)
|
bere
|
kitam
|
kile
|
ki
|
piyer-i
|
nüg-e
|
lupke
|
üsde-re-s
|
|
for
|
which
|
girl
|
ki
|
father-pos3
|
new-attr
|
dress
|
buy-perf-3
|
|
|
|
|
|
|
|
|
|
|
be-u
|
deyş.
|
|
dat-dist
|
look.imp
|
|
|
|
|
‘Look at the girl for whom her / whose father has bought a new dress.’ (fn)
|
(86)
|
e-kitam
|
nozu
|
ki
|
ayol
|
muş
|
üsde-re-s,
|
u-re
|
girind.
|
|
abl-which
|
cat
|
ki
|
child
|
mouse
|
take-perf-3
|
dist-acc
|
catch.imp
|
|
‘Catch the cat from whom the child has received (=taken) a mouse!’ (fn)
|
|
(87)
|
boş
|
kitam
|
parşah
|
davo
|
s-um,
|
ez-u
|
xak-i-re
|
|
|
with
|
which
|
king
|
fight
|
do.subj-1
|
abl-dist
|
earth-pos3-acc
|
|
|
|
|
|
|
|
|
|
|
|
be turbi
|
es
|
bere
|
keşi-ren
|
avar-den
|
zür-mü
|
m-ras-ü,
|
|
|
dat-bag.ez
|
horse
|
for
|
pull-inf
|
carry-inf
|
strength-pos1
|
evt-reach-3
|
|
|
|
|
|
|
|
|
|
|
|
‘Whichever king I make war with, I have the power to carry his earth away in a bag.’ (example translated from Azerbaijani)
|
(88)
|
e-kitam
|
eyol
|
ki
|
mumser-i-re
|
pare
|
sax-dem
|
|
abl-which
|
child
|
ki
|
hair-pos3-acc
|
cut
|
do-perf-1pl
|
|
‘The child whose hair we cut / have cut.’ (fn)
|
In some interesting cases (e.g. [89]), the markers of the two different syntactic relations (possessum and Patient argument) can scramble (the e- genitive proclitic goes with the possessive suffix in the main clause, while the –e accusative enclitic is governed by the main verb):
(89)
|
[e-kitam
|
seg-e
|
ki
|
guşye-yi-re
|
keş-ren-im]
|
mo-kuf !
|
|
of-which
|
dog-acc
|
ki
|
ears-pos3-acc
|
pull-prs-1-pl
|
proh-beat
|
|
(lit.) ‘The dog whose ears we pull, do not beat it!’ (fn)
|
..4.Double relativisation
In the case of double relativisation, the head-internal strategy has to be used in Tati. The double RCs provided above from our Azerbaijani corpus can be translated into Tati with a parallel construction:
(90)
|
ki-re
|
e-kitam
|
gede
|
xoş
|
ma-ren,
|
|
who-acc
|
abl-which
|
boy
|
well
|
come-prs3
|
|
|
|
|
|
|
|
sib-e
|
be-u
|
b-şən-ü
|
|
|
|
apple-acc
|
dat-dist
|
subj-throw-3
|
|
|
|
|
|
|
|
|
|
‘Which boy pleases to whom”, let her throw the apple to him.’ (fn)
|
|
|
(91)
|
her
|
şexs
|
e-kitam
|
mezheb
|
be-kitam
|
mezheb
|
|
each
|
person
|
abl-which
|
religion
|
dat-which
|
religion
|
|
|
|
|
|
|
|
|
xas-d-ü
|
tan-ü
|
güdoş-den.
|
|
|
|
|
want-pst-3
|
can.subj-3
|
pass-inf
|
|
|
|
|
|
|
|
|
|
|
|
‘Let everyone have the possibility to abide by the faith he wishes.’
|
|