CC BY
29
44
ф Ю.А. Ландер
| Кабардинские
| обобщенно-уступительные конструкции1
Статья описывает обобщенно-уступительную конструкцию в кубанском диалекте кабардино-черкесского языка (абхазо-адыгская семья). Рассматриваемая конструкция включает зависимое предложение, которое маркируется сочетанием условного и аддитивного маркеров и обязательно содержит вопросительное местоимение, имеющее семантику свободного выбора. Типологическое своеобразие кабардино-черкесской обобщенно-уступительной конструкции связано с тем, что у некоторых носителей квантифицирующее прочтение здесь может получить, помимо вопросительного местоимения, и анафорическое местоимение - при условии, что оно находится выше вопросительного местоимения в иерархии «эргативный актант > актант > абсолютивный актант > непрямой объект > посессор/объект послелога». В статье предлагается формально-семантический анализ кабардино-черкесской обобщенно-уступительной конструкции, который предполагает неселективную квантификацию (в том числе связывающую «дэвидсоновскую» событийную переменную) и позволяет описать особые эффекты, наблюдаемые в конструкции.
Ключевые слова: обобщенно-уступительная конструкция, универсальная уступительная конструкция, кабардино-черкесский язык, местоимения свободного выбора, дэвид-соновская семантика.
Yu. Lander
Kabardian universal concessive conditionals2
This paper provides the description of the universal concessive conditional (UCC) construction in the Kuban dialect of Kabardian, an ergative polysynthetic language of the
1 Автор признателен своим консультантам в ауле Ходзь за терпение и интерес к работе, участникам Северокавказского лингвистического семинара, на котором докладывались основные идеи данной работы, - за обсуждение, анонимному рецензенту - за ценные комментарии к ранней версии статьи.
Работа выполнена при финансовой поддержке РФФИ (проект № 15-06-07434a), РГНФ (проект № 14-04-00580) и Программы «Научный фонд НИУ ВШЭ» (проект № 15-09-0263).
Ответственность за все ошибки лежит на авторе работы.
2 I am grateful to my Kabardian consultants in the village of Khodz for their interest and patience, to the audience of the Northwest Caucaslian linguistic seminar where the ideas of this paper were first presented (Moscow, September 2015), for discussion, and to an anonymous reviewer for comments on an earlier version of the paper. This paper is mainly based on my research supported by Russian Foundation for Basic Research (grant No. 15-06-07434a) and Russian Foundation for Humanities (grant No. 14-04-00580) and made within the framework of the Academic Fund Program at the National Research University Higher School of Economics (HSE) in 2015 (grant No. 15-09-0263). All errors are mine.
Northwest Caucasian family. The Kabardian UCC construction involves a subordinate clause g cn which is marked with a combination of the conditional and additive markers and obligatorily ^ & contains an interrogative pronoun with free choice semantics. More surprisingly, for some ro ^ speakers, the subordinate clause in this pattern may also contain a definite pronoun cu ^ which receives quantificational reading if it is higher in the hierarchy ergatve > absolute ^ ve > indirect object > p0ssess0r/p0stp0siti0nal object than the free choice interrogative pronoun. It is suggested that this can be accounted with a semantic representation involving quantification over events.
Key words: universal concessive conditional, Kabardian, free choice, event-based semantics.
1. Introduction
In this paper, I discuss the Kabardian construction illustrated in (1).
(1) sat a-ba.m z'-je-s-?-a-m-j9, a-r ja-s-a
what that-oBL1 pREF-DAT-lsG.ERG-say-psT-coND-ADD that-ABs 3sG.ERG-do-psT
'Whatever I told him/her, s/he did that.'
This construction minimally consists of two clauses, of which one can be considered subordinate to another. In (1), sat abam z'jesPamja 'whatever I told him' is the subordinate clause, while ar jasa '(s/he) did that' is the matrix clause.
The pattern in (1) belongs to the family of universal concessive conditionals (UCCs). (2) shows examples of UCCs from languages close to Standard Average European:
(2) a. Spanish [Haspelmath, König, 1998, p. 610]
Donde-quiera que vaya, nunca la dejará where-want that go:suBj never her leave:Fuт:3sG
b. Dutch [Haspelmath, König, 1998, p. 611]
Waar ze ook maar heen gaat, hij zal haar nooit verlaten where she also only to goes he will her never leave
c. Russian
Kuda by ona ni posla, on ee nikogda ne pokinet where subj she not go:psT:F he her never not leave:FuT.3sG 'Wherever she goes, he will never leave her.'
1 Abbreviations: abs - absolutive, acc - accusative, add - additive, adj - adjectivizer, adv - adverbial, all - allative, aor - aorist, Aux - auxiliary stem, ben - benefactive, caus - causative, car - caritive, comp -comparative, cond - conditional, dat - dative, dem - demonstrative, dir - directive, dyn - dynamic, erg -ergative, f - feminine, fut - future, imp - imperative, ins - instrumental, io - indirect object, ipf -imperfec-tive, lnk - linker, loc - locative preverb, mal - malefactive, mod - modal, neg - negation, obl - oblique, pl - plural, poss - possessive, pot - potential, pred - predicative form of pronouns, pref - prefixal element of the verb 'say', pst - past, rcp - reciprocal preverb, re - refactive/reversive, rec - reciprocal cross-reference, rfl - reflexive, sg - singular, subj - subjunctive.
^ In this paper, I provide a description of the UCC construction in the Kuban dia-
1 lect of Kabardian. The constructions like (2) have been sometimes treated as relative
ru
Fo clause constructions (see, for instance, [van Riemsdijk, 2006]). However, the cha-=§ racterization of UCCs as relatives was cast in doubt in a number of works (see oS especially [Haspelmath, König, 1998; Gawron, 2001]), and I will argue against this analysis too. Further, I propose a semantic analysis of Kabardian UCCs which is based on the Davidsonian approach to the semantics of clauses. I will show that this representation explains not only the free choice semantics of the interrogative phrase in UCCs but also unexpected patterns like the possible use of a definite pronoun carrying an indefinite meaning (3):
(3) a-ba.m xet j9-XaKw-m-j9 selam ja-rj-e-x
that-oBL who 3sG.ERG-see-coND-ADD greeting DAT-3sG.ERG-DYN-carry 'No matter who saw whom, s/he greets him/her.'
The paper has the following structure. Section 2 provides the necessary background on Kuban Kabardian. Section 3 describes the definitional characteristics of Kabardian UCCs. In Section 4 I informally discuss the semantic contribution of the subordinate clause. Section 5 is devoted to the referential properties of the interrogative pronoun in UCCs. Section 6 presents unusual data on multiple quantificational elements within the subordinate clause. Section 7 provides a formal semantic account of various properties of the UCC construction in Kabardian. The last section presents conclusions.
2. Background on Kuban Kabardian
The data presented in this paper are from the Kabardian variety spoken in the village of Khodz (the Republic of Adygea, which is a part of the Russian Federation). Kabardian (or East Circassian) together with West Circassian (or Adyghe) constitutes the Circassian branch of the Northwest Caucasian family, which also includes Abkhaz, Abaza and the now extinct Ubykh language. Kuban Kabardian is an outlier Kabardian dialect, which is separated from other dialects of the language. Its choice for this paper is not motivated by its specific properties, though. In particular, my preliminary data suggest that many statements given below hold for other Circassian varieties as well, but more research should be done to make any concrete statements.1
Northwest Caucasian languages are highly polysynthetic: the word may have a very complex structure and contain multiple prefixes which cross-reference its
1 The only published description of Kuban Kabardian I am aware of is [Kumakhov, 1969]. Standard Kabardian is described in a number of reference grammars (mostly published in Russian) including [Turchaninov, Tsagov, 1940; Jakovlev, 1948; Abitov et al., (eds) 1957; Bagov et al. (eds), 1970; Colarusso, 1989, 1992, 2006; Kumakhov (ed.), 2006]. Although some of these works mention UCCs, none of them include any detailed description of this construction.
arguments. Cf. the following sentence where the first verb cross-references the | cn indirect object introduced with the possessive applicative1 and the second verb cross- ^ ^ references the absolutive and the ergative arguments: g 5
cD ^
(4) as'xe 0-j9-?e-te-q9m krasnodar s-ja-Ke-kwe-n-u £
money 3sG.io-poss-be-ipF-NEG Krasnodar 1sG.ABs-3sG.ERG-cAus-go-MoD-ADv 'She had no money to send me to Krasnodar.'
Like many other polysynthetic languages (cf. [Baker, 1996; Jelinek, Demers, 1996]), the Circassian languages pose a problem of what is to be considered arguments: full NPs or cross-reference markers (for some discussion of West Circassian see [Lander, 2005; Lander, Testelets, to appear]). Without insisting on any concrete syntactic structure, I will assume here, however, that NPs at least sometimes do not enter the core of the clause and may be placed somewhere in the clausal periphery (see [Lander, 2010, to appear] for some specific evidence). This assumption will become important in the discussion of some patterns below.
In general, all languages of the Northwest Caucasian family are morphologically ergative. Cross-reference within the predicate is divided into three series, namely the absolutive series, the ergative series, and the indirect object series (introduced by applicatives within the predicate). Circassian languages distinguish between the absolutive case which marks the (3rd person) intransitive subjects and transitive undergoers and the oblique case which, roughly, marks non-absolutive arguments. None of these cases necessarily receives overt expression, though.2
The syntactic characterization of Northwest Caucasian languages in general and Circassian languages in particular is less clear. Lander [2010] and Letuchiy [2012] presented various arguments from syntax and morphology which evidence that erga-tivity in West Circassian is not only a surface phenomenon, and Lander [2014] even argued that Circassian languages belong to a specific kind of syntactically ergative languages. However, there are some processes in Circassian which at first glance are organized on a nominative/accusative basis. The most prominent of them is reflexive binding. In Circassian, reflexives are normally expressed morphologically, by prefixes which belong to the same paradigm as other cross-reference prefixes. The controller is usually the ergative in transitive verbs (5) and the absolutive in multivalent intransitive verbs, i.e. with intransitive verbs that contain indirect object morphology (6).
1 The 3rd person singular indirect object is cross-referenced by a null prefix. Note that null morphemes are not shown in most examples given below.
2 In addition to the absolutive and oblique cases, there are forms that are usually described as the "adverbial case" and the "instrumental case". However, since these forms display quite specific behavior, it is unlikely that they should be included into the same paradigm as absolutive and oblique.
(5) s9-xw9-te-cpm za-q-a-z-Ke-watapsa-z'a-n-ew-ja
iSG.ABS-happen-IPF-NEG REL.ABS-DIR-3PL.I0-1SG.ERG-CAUS-releaSe-RE-M0D-ADV-ADD
'I was not able to make them release myself.'
(6) se Kwgnj'e-m-c'e sa-z-o-pXa-z'
I mirrOr-OBL-INS 1SG.ABS-RFL.I0-DYN-l00k-RE
'I am looking at myself in the mirror.'
However, there are intransitive verbs, called inverse verbs, which violate this. For them, a more agentive indirect object may control the reflexive prefix appearing in the position intended for the absolutive cross-reference. Examples include the verb 'be forgotten', with the agentive indirect object argument introduced with a locative applicative (7),1 and the "potential" forms of verbs with transitive stems (i.e. stems allowing the appearance of the ergative cross-reference prefix), where the most agentive argument appears as an indirect object introduced with the benefactive applicative (8).
(7) za-s-sa-Kwgpse-z'-a
RFL.ABS-iSG.IO-LOC-be.forgotten-RE-PST
'I forgot myself.'
(8) z9-s-xw-je-Ke-3'e-z'-q9m jastoijaja-m
RFL.ABs-1sG.Io-BEN-DAT-cAus-study-RE-NEG history-oBL
'I cannot teach myself history.'
The basic word order of Kabardian is usually claimed to be SOV (Actor -Undergoer - Verb), see [Colarusso, 1992, 169ff]. Yet in reality almost no strict restrictions on the order of constituents is found. Nonetheless, it is natural to describe this language as left-branching, since it has postpositions rather than prepositions and normally places syntactically autonomous attributes before their heads.
3. UCCs: the first acquaintance
The defining properties of the UCC in Kabardian are listed below:
- the subordinate clause contains an interrogative pronoun,
- the subordinate clause is marked by a sequence of morphemes -m-ja, the result of combining the conditional -me and the additive -ja.
Typologically, both properties are quite widespread for UCCs [Haspelmath, Konig,1998], and combining the conditional and additive markers is typical for concessives in general [Xrakovskij et al., 2012].
1 Curiously, this verb also allows the appearance of the indirect object reflexive prefix controlled by the absolutive argument, and in fact, this form is even highly preferred by some speakers:
(i) ss-z-ss-KWgpse-z'-a
iSG.ABS-RFL.IO-LOC-be.forgotten-RE-PST 'I forgot myself.'
See also [Letuchiy, 2013] for discussion of similar facts in West Circassian.
49
fC CN S 01
CU Z
VD
(9)-(14) illustrate the use of various interrogative pronouns in UCCs:
(9) ma-c'e xet sat1 qa-s-ja-Kwet-a-m-ja, se
that-iNs who what DIR-LOC-3sG.ERG-find-PST-COND-ADD I t= o
qa-z-a-t sz
DIR-1SG.IO-3PL.ERG-give
'No matter who finds what here, they give it to me.'
(10) st-xwe.de-rje he qa-z-e-3eq-a-m-ja, what-like-aDJ dog DIR-1SG.IO-DAT-bite-PST-COND-ADD a-ba.m sa-s-o-s'ane
that-oBL 1sG.Aвs-Loc-DYN-be.afraid
'Whichever dog bit me, I am afraid of it.'
(11) sat-sa-Kwe-c'-ja zabKe-sxo q-je-pse-m-ja,
what-LOC-time-INS-ADD wind-AuG DIR-DAT-blOW-COND-ADD
(a waxte-m) shan»wabz'e-r ze-xwa-z-o-s
that time-oBL window-ABs rec.io-ben-1 sG.ERG-DYN-do
'Whenever the strong wind blows, I close the windows.'
(12) sata-m she.c'e a-ba.m sa-ma-?-a-m-ja, what-oBl for that-oBL Loc-NEG-be-psT-coND-ADD a-ba.m par-ja xwa-ze-fe-c'-qam
that-oBL nothing-ADD ben-rfl . io -mal -go.out-NEG 'Whatever s/he is there for, nothing comes (of it).'
(13) ta-xwe.de-rje txaX-xe-r f-u s-Xa»w-m-ja, which-like-ADJ book-pl-ABs well-ADv 1sG.ERG-see-coND-ADD a-p-xwe.de-xe-m se s-ja-3'e-n-u s-f^-f that-oBL-like-pi-oBL I 1sG.ABs-3pL.io+DAT-read-MoD-ADv 1sg.io-mal-desired 'Whichever books I like, I prefer to read them.'
(14) dapse jo-ma-t-m-ja xwa-rja-qwa-ne-qam how.much DAT+2sG.ERG-NEG-give-coND-ADD BEN-Loc-enough-MoD-NEG 'However much you will give him, this will not be enough for him.'
The interrogative pronoun is usually fronted (possibly because of fronting of its correlate in the Russian stimulus), but it need not do so, as the following example demonstrates:
(15) nac'al'nike-r xet je-ma-pX-m-ja,
chief-ABs who DAT-NEG-look-coND-ADD
z.e.we ja-ze ze-tr-a-pe-t
right.away 3pi.io+poss-mouth REc.io-Loc-3pL.ERG-close-ipF
'Whoever the chief looked at, they closed their mouths straight away.'
1 Interestingly, although sat 'what?' (or some cognate of sat) serves as the basic inanimate interrogative pronoun in Circassian, it is not actively used in Kuban Kabardian questions, being replaced with another pronoun, Ac. The latter, however, cannot be used in the UCC construction.
The presence of an interrogative pronoun is necessary, and it cannot be replaced, for example, by an indefinite NP (cp. the infelicitous (16a) with (16b)).
(16) a. *za-he-gwere-r qa-z-e-3eq-a-m-ja,
one-dog-some-ABs DIR-1sG.Io-DAT-bite-psT-coND-ADD
a-ba.m sa-s-o-s'ane
that-oBl 1sG.Aвs-Loc-DYN-be.afraid
b. st-xwe.de-rje he qa-z-e-3eq-a-m-ja,
what-like-ADJ dog DIR-1SG.IO-DAT-bite-PST-COND-ADD a-ba.m sa-s-o-s'ane
that-oBl 1sG.Aвs-Loc-DYN-be.afraid
'Whichever dog bit me, I am afraid of it.'
The subordinate clause normally occurs in the left periphery, but it can be embedded, as in (17), or appear in the right periphery, as in (18):
(17) 3'a.r-ja ne-s, xet sa-d-je-ma-3'-a-m-ja,
now-ADD DIR-reach who 1SG.ABS-COM-DAT-NEG-Study-PST-COND-ADD zec'e-m-ja s-a-?o-se
all-oBl-ADD 1SG.ABS-3PL.IO-LOC+DYN-knOW
'Till now, whoever studied with me, I meet with all of them.'
(18) s-ja-Ke-gwaf-a, sat sa-s-XeKw-a-m-ja
1sG.ABs-3sG.eRG-cAus-glad-psT what LOC-1SG.ERG-See-PST-COND-ADD '(Everything) made me happy, whatever I saw there.'
As noted above, UCCs are marked with a combination of the conditional marker and the additive morpheme. The same sequence of suffixes is used for scalar concessive conditionals with the meaning 'even if' (19) or 'though' (20) and for alternative concessive conditionals (21):
(19) ja3'a.r-ja bz-ja-x z-e-b-Ke-caxw-ne-m-ja,
now-ADD language-lnK-six RFL.IO-DAT-2sG.ERG-CAUS-knOW-FUT-COND-ADD adaga-bze-c'e deKw-u wa-pseXe-fa-ne-qam
Circassian-language-INs good-ADv 2sG.ABs-speak-poT-FuT-NEG 'Even if you learn six languages more, you will not be able to speak Circassian well.'
(20) a-ba ja-maXxe-m, masane-m kwed.s.a-we ja-s that-oBl poss-son.in.law-oBl car-oBl frequently-ADv loc-sit pe.t-m-ja,
when-coND-ADD
3'a.re ne-s-c'e prava qa-d-ja-x-a-qam
now DiR-reach-ms license DIR-Loc-3sG.ERG-carry-psт-NEG
'Her son-in law, though he often drives a car, has not got the driver's license | cn till now.' £ ^
• VO re
(21) Z9le-m w9-s9-psew-m-j9 w9-s9-m9-psew-m-j9
village-oBL 2sG.ABs-Loc-live-coND-ADD 2sG.ABs-Loc-NEG-live-coND-ADD ^
w-j9-ad9ge xabze z9-§9-b-Ke-Kw9pse
2sG.io-poss-Circassian law RFL.Io-Loc-2sG.ERG-cAus-be.forgotten
XW9-ne-q9m
happen-FuT-Neg
'Whether you live in a village or not, you should not let yourself forget the Circassian law.' (Ksenia Ershova's field notes)
There is, however, an important feature that differentiates UCCs from other concessive constructions (in addition to the obligatory presence of an interrogative pronoun): a UCC may contain an optional negative prefix, which does not carry negative semantics and therefore instantiates the so-called expletive negation, as in (22).1
(22) w9ne-dapse j9-m9-?e-m-j9, j9tan-j9 j-e-s9 house-how.many poss-NEG-be-coND-ADD then-Add 3 sG^RG-Dyn-make 'No matter how many houses he has, he continues to build them.'
As a result, the semantics of the subordinate clause cannot contain negation at al. In particular, any subordinate clause in a UCC is interpreted as having positive polarity even if it contains the negative marker.2 Cf.:
(23) a-b9 s9t-xwe.de s-j9-uc'enik j9-m9-c9xW9-m-j9 that-obL what-like 1sG.io-poss-pupil 3sG.ERG-NEG-knOW-COND-ADD pedarke-nc'-u qe-n-a
present-cAR-ADv DiR-remain-pst
'Whichever pupil of mine s/he knew, s/he was left without a present.' *'Whichever pupil of mine s/he did not know, s/he was left without a present.'
(24) #xet rusXan j9-m9-XaKW-a-m-j9, a-xe-r who Ruslan 3sG.ERG-NEG-See-PST-COND-ADD that-pL-Abs j9-ne?wase-(j9m
poss-acquaintance-NEG
'Whoever saw Ruslan, s/he is not an acquaintance of his.' *'Whoever did not see Ruslan, s/he is not an acquaintance of his.'
1 The expletive negation is observed in UCCs in many languages (Haspelmath, König, 1998). Some ideas concerning its motivation may be found in [Yoon, 2011] and [Makri, 2013], who both relate expletive negation to some kind of modality. An interesting feature of Circassian UCCs is that here negation is optional. Despite this, most elicited examples contain expletive negation, presumably because of the Russian stimulus, which obligatorily contains a negation marker.
2 In theory, double negation may be used for negative polarity but it is too hard to process.
In other concessives, the negative prefix has a clear semantic motivation:
(25) a-r ma-jana-m-ja, qwadame-m ja-sha.pe-m ne-s-u that-ABs NEG-big-coND-ADD branch-oBL poss-top-oBL DiR-reach-ADv s-t-a
Loc-stand-psT
'Athough he was not big, he could reach the top of the branch.'
4. The restricting function of the subordinate clause
In many, instances UCCs restrict the reference of some argument of the matrix clause. For instance, in (26) the subordinate clause restricts the reference of the abso-lutive argument of the verb 'you will eat (it)'.
(26) we st-xwe.de s'xa.n p-sa-m-ja, (a-r) you(sg) what-like food 2sG.ERG-make-coND-ADD that-ABs p-s'xa-z'a-ne
2sG.ERG-eat-RE-FUT
'You will eat whatever dish you will prepare.'
Not surprisingly, some descriptions of Kabardian treat this construction as a kind of relative clause construction, which restricts the reference of an argument of the matrix clause. For example, Colarusso [1992, p. 193] in his Kabardian grammar gives examples of UCCs as clauses containing an independent relative pronoun. In fact, already in 1957 a Kabardian reference grammar described sentences like (26) as correlatives [Abitov et al. (eds), 1957, p. 218].1
Typically, in correlatives the subordinate clause contains a pronoun which marks what is relativized, while the matrix clause contains a coreferent pronoun referring to the same entity. (27) is an example of the Hindi correlative construction, which serves as a canonical correlative in the literature. (28) presents a Turkish example of (what is considered) a correlative. Note that like in Kabardian UCCs, here the subordinate clause is marked with a conditional marker, which is actually a recurring phenomenon in correlatives [Liptak, 2009, p. 26-27].
(27) Hindi [Srivastav, 1991, p. 639]
jo laRkii khaRii hai vo lambii hai REL girl standing is DEM tall is
'The grl who is standing is tall.'
(28) Turkish [Kornflit, 1997, p. 60]
ben [Chomsky ne yaz-ar-sa] on-u oku-r-um
I Chomsky what write-AoR-coND that-Acc read-AoR-1sG 'I read that which Chomsky writes.'
1 It is worth noting that no other construction could be regarded as a correlative in Kabardian.
However, Kabardian UCCs cannot be described as correlatives, despite the apparent similarities to these patterns. In fact, even if the subordinate clause restricts the reference of some NP in the matrix clause, the interrogative pronoun in a UCC is not necessarily coindexed with this matrix NP. In (29), for instance, the interrogative refers to the possessor of the nominal predicate, but the generic subject of the matrix clause, whose reference is restricted, is coindexed with the subject of the subordinate clause rather than the possessor of the predicate. Hence the interrogative pronoun in such constructions cannot be considered a relative pronoun.
(29) xet ja-ma-qwe-m-ja, z-ja-pa-z'a-n xw.je
who PoSS-NEG-son-CoND-ADD RFL.ABS-3sG.ERG-bring.up-RE-MoD must 'Whoever's son (you) are, (you) must bring up yourself.'
Moreover, it is easy to construct an example of the UCC where the matrix clause does not have any argument whose reference could be restricted by the subordinate clause:
(30) xet d-ja-diijektora-m has-u qa-xwe-ma-kw-a-m-ja,
who 1PL.Io-PoSS-director-oBL guest-ADV DIR-BEN-NEG-go-PST-CoND-ADD de d-ja-skola-r d-o-Ke-qabze
we 1PL.Io-PoSS-school-ABS 1PL.ERG-DYN-CAUS-Clean
'Whoever comes to our director as a guest, we clean our school.'
(31) xet ja-z'e-m ma-baw-a-m-ja,
who PoSS-cow-oBL NEG-moo-PST-CoND-ADD s-ja-z'e-m z-ja-wasexw-t
1SG.Io-PoSS-cow-oBL RFL.ABS-3SG.ERG-hide-IPF
'Whoever's cow mooed, my cow kept silence.'
Clearly, UCCs have a broader function than correlatives. Informally, the subordinate clause in UCCs describes the conditions under which the situation described in the matrix clause is discussed, hence restricting the range of relevant contexts. If so, restricting the reference of an argument of the matrix clause is only a particular case of this broader function.
This suggests a link to conditionals, which are now believed to restrict the domain of implicitly or explicitly introduced quantifiers [Kratzer, 2012]. This may explain the frequent use of conditional markers with universal concessives, but leave open the question of the semantic representation of such restriction: what domain does the subordinate clause restrict? I will return to this question in Section 7.
5. The reference of the interrogative pronoun
The NP expressed by the interrogative pronoun does not behave as specific or definite. The following example demonstrates that it cannot be associated with a definite referent, here expressed with a proper name:
ID Z .
• VO re
E °
(32) *st-xwe.de hazret pjat-ja-x q-ja-h-a-m-ja,
what-like Hazret five(mark)-LNK-six DiR-3sG.ERG-carry-psT-coND-ADD gramote xwe-fase deed BeN-do.for
('Which Hazret got six ecellent marks is worthy of a diploma.')
(33) and (34) show that the interrogative pronoun cannot introduce a specific referent. In the first of these examples, the interpretation of the matrix clause implies that the referent introduced by the interrogative pronoun be unique ("Lone Rider" is the name of a well-known novel by the Circassian writer Tembot Kerashev), so the sentence appears to be infelicitous. In (34) the UCC is found acceptable, but it cannot be continued with a clause that implies the unique reference.
(33) *sat adage-xe-m ja-ha-X-a-we
what Circassian-pl-obl 3pL.ERG-carry-ALL-psт-ADv
s-je-3'-a-m-ja, ja-ce-te-r saw-zaqw
1sG.Aвs-DAт-read-psт-coND-ADD poss-name-ipf-Abs rider-lone ('What I read devoted (lit. brought) to Circassians, its name was "Lone Rider".')
(34) xet daKwase sa-p-je-Kwec'-a-m-ja,
who yesterday 1sG.Aвs-Loc-DAт-meet-psт-coND-ADD
tha.w.je.Ke.psew qa-z-z'-ja-?-a.
thank.you DIR-1SG.IO-PREF-3sG.ERG-Say-PST
#a-r rusXan jad-u qa-se-c'-a.
that-Abs Ruslan poss+father-ADv DiR-Loc-go.out-pst
'Whomever I met yesterday, s/he thanked me.
#He turned out to be Ruslan's father.'
Rather the translations suggest that the reference of the interrogative pronoun can be characterized as free choice (FC), the kind of reference which originally was associated with the "universal" reading of English any as in The most attractive part of any camel is the toe,1 but later was shown to be relevant for the description of indefinite pronouns cross-linguistically [Haspelmath, 1994; Giannakidou, 2001; Jayez, Tovena, 2007; Menendez-Benito, 2010]. Note that interrogatives in Circassian
1 Interestingly, Zeno Vendler, who first described this reading of any as "freedom of choice", provided a UCC paraphrase for the sentence I can beat any one of you: "no matter whom you select from among you, I can beat him" [Vendler, 1967, p. 80].
languages indeed may serve as the bases for FC indefinites, as in (35) (see also | cn [Kapitonov, 2010] for West Circassian): ^ ^
• VO re
(35) xet-ja q-je-3'e, se-r-c'e twa-r-ja za
who-ADD DIR-DAT-call(IMP) I-PRED-INS two-ABS-ADD One ^
'Call anyone, no matter whom (lit., for me both are one)!' (Evgeny Mozhaev's field notes)
Two basic approaches to FC are found in literature [Haspelmath, 1994, p. 90ff]. The first approach identifies it with universal quantification, while the second approach suggests that FC indefinites represent narrow scope existential quantification (we will consider both possibilities later). In either case, the subordinate clause in the UCC construction must contain a quantificational pronoun.
6. Multiple quantificational pronouns in subordinate clauses
There may be several quantificational pronouns in a single UCC, as in (36)-(37):
(36) xet xet ja-ma-Xa»w-a-m-ja, z-e-xwen-t
who who 3SG.ERG-NEG-see-PST-coND-ADD rec . io -DAT-abuse-IPF 'No matter who saw whom, they were abusing each other.'
(37) xet sat ja-ma-qwat-a-m-ja, jetane j-e-sa-z'
who what 3SG.ERG-NEG-break-PST-CoND-ADD then 3SG.ERG-DYN-make-RE
'No matter who broke what, after that he repairs this.'
Somewhat unexpectedly, for some speakers, one of the quantificational NPs in this case may be represented by a definite pronoun:1
(38) xet a-ba.m ja-rja-ma-Ke-zes'-a-m-ja,
who that-oBL LoC-3SG.ERG-NEG-CAUS-bored-PST-CoND-ADD nexa-bere z-e-pseXe-z'-a-qam
CoMP-many+TIME REC.Io-DAT-talk-RE-PST -NEG
'No matter who bored whom, they did not ever talk to each other.'
This reading is only possible for the pronoun a-, which formally behaves like a demonstrative but in fact serves as a neutral anaphoric pronoun. The replacement of a- with demonstratives proper blocks the quantificational reading; cf. (39a). If a participant is not expressed with a separate NP, it cannot get the quantificational reading either (39b).
1 The data on this curious construction were collected in 2015 from four informants. However, my informants of 2016 did not accept this pattern. Hence I have to accept that its presence is at least subject to variation. It is worth noting that the use of a definite pronoun in quantificational reading is certainly not a default construction for any speakers.
^ (39) a. xet ma-ba.m/mwe-ba.m/3'a-ba.m
i who this-oвL/that-oвL/that-oвL
• v,
ja-ga-ma-Ke-zes -a-m-ja,
^ L0C-3sG.ERG-NEG-CAUS-b0red-PST-C0ND-ADD
_D
<S nexa-bere z-e-pseXe-z'-a-qam
COMP-many+TIME REC.IO-DAT-talk-RE-PST-NEG
b. xet ja-rja-ma-Ke-zes'-a-m-ja,
who L0C-3sG.ERG-NEG-CAUS-b0red-PST-C0ND-ADD nexa-bere z-e-pseXe-z'-a-qam
COMP-many+TIME REC.IO-DAT-talk-RE-PST-NEG
'Whomever s/he bored, they did not ever talk to each other.' *'No matter who bored whom, they did not ever talk to each other.'
However, it is not the case that a definite pronoun may always acquire a quantificational reading in UCCs with multiple quantificational NPs. In particular, if there are two quantificational pronouns, the one that is lower in the hierarchy erg > abs > io > possessor/postpositional object should be marked by the interrogative.1 Otherwise it receives a simple anaphoric interpretation. Some examples follow.
In the subordinate clause of (40a), the definite pronoun refers to the ergative argument and the interrogative pronoun refers to the absolutive argument. While the latter gets the FC interpretation, the former gets either a quantificational reading or a definite interpretation (in which case it functions anaphorically). In (40b), however, the definite pronoun refers to the absolutive argument and being lower on the hierarchy, it cannot receive a quantificational reading.
(40) ERG > ABs
a. xet a-ba q-ja-ma-XeKw-a-m-ja,
who that-oBL DIR-3 SG.ERG-NEG-See-PST-COND-ADD
zec'e-m-ja selam ze-r-a-x-xe
all-oBL-ADD greeting REc.io-DAT-3pL.ERG-carry-pL
'Whomever s/he sees, all greet each other.'
'No matter who sees whom, all greet each other.'
b. xet a-r q-ja-ma-XeKw-a-m-ja,
who that-ABs DIR-3 SG.ERG-NEG-See-PST-COND-ADD zec'e-m-ja selam ze-r-a-x-xe all-oBL-ADD greeting REc.io-DAT-3pL.ERG-carry-pL 'Whoever sees him/her, all greet each other.' *'No matter who sees whom, all greet each other.'
1 I have no data on what happens if there are more than two elements of this kind in the subordinate clause.
The following examples present similar facts concerning the combinations of the | cn absolutive argument of an intransitive verb and an indirect object (41), the absolutive ^ ^ argument of a transitive verb and a possessor (42), the absolutive argument of an intransitive verb and the object of a postposition (43). In all these cases, in order for ^ a definite pronoun to get a quantificational reading, it is the absolutive (which is ^ higher in the hierarchy) that must be expressed with this pronoun.
(41) ABS > io
a. xet a-r je-ma-zew-a-m-ja,
who that-ABS DAT-NEG-beat-PST-CoND-ADD
zec'e-r-ja z-o-fa-z'-xe
all-ABS-ADD REC.Io-DYN-good-RE-PL
'Whomever s/he bit, all made it up with each other.' 'No matter who bit whom, all made it up with each other.'
b. xet a-ba je-ma-zew-a-m-ja,
who that-oBL DAT-NEG-beat-PST-CoND-ADD
zec'e-r-ja z-o-fa-z'-xe
all-ABS-ADD REC.Io-DYN-good-RE-PL
'Whoever bit him, all made it up with each other.' *'No matter who bit whom, all made it up with each other.'
(42) abs > Possessor
a. xet ja-pxwa-m a-r je-ma-pX-a-m-ja,
who PoSS-daughter-oBL that-ABS DAT-NEG-look-PST-CoND-ADD z-e-xwen-t
rec . io -DAT-abuse-IPF
'At whoever's daughter s/he looked, they were abusing each other.' 'No matter who looked at whose daughter, they were abusing each other.'
b. a-ba.m ja-pxwa-m xet je-ma-pX-a-m-ja,
that-oBL Poss-daughter-oBL who DAT-NEG-look-PST-coND-ADD z-e-xwen-t
REC. io -DAT-abuse-IPF
'Whoever looked at his/her daughter, they were abusing each other.' *'No matter who looked at whose daughter, they were abusing each other.'
(43) abs > Postpositional object
a. xet-dje a-r ma-kw-a-m-ja,
who-at that-ABS NEG-go-PST-CoND-ADD
pedarke qa-r-ja-t-a
present DIR-DAT-3SG.ERG-give-PST
'Whoever s/he came to, s/he gave him/her a present.' 'No matter who came to whom, s/he gave him/her a present.'
(43) b. a-ba-dje xet ma-kw-a-m-ja,
that-oBL-at who NEG-go-psT-coND-ADD
pedarke qa-r-ja-t-a
present DIR-DAT-3sG.ERG-give-PST
'Whoever came to him/her, s/he gave him/her a present.'
*'No matter who came to whom, s/he gave him/her a present.'
It is not that easy to analyze the combinations of the ergative argument and non-absolutive participants, since they are marked with the same (oblique) case. It is clear, however, that the ergative argument outranks at least possessors and postpositional objects:1
(44) ERG > Possessor
a. xet ja-qwe a-ba ja-ma-s'a-z'-a-m-ja ja.dje, who poss-son that-obl 3sG.ERG-NEG-lead-RE-psт-coND-ADD at.home jane-jade-xe-m tha.w.je.Ke.psew q'a-z'a-r-a-?e poss+mother-poss+father-PL-oвL thank.you DIR-PREF-DAт-3pL.ERG-say 'Whoever'son s/he led home, his parents thank him/her.'
'No matter who led whose son home, his parents thank him/her.'
b. a-ba ja-qwe xet ja-ma-s'a-z'-a-m-ja ja.dje, that-obl poss-son who 3sG.ERG-NEG-lead-RE-psт-coND-ADD at. home jane-jade-xe-m tha.w.je.Ke.psew q'a-z'a-r-a-?e poss+mother-poss+father-pL-oBL thank.you DIR-pREF-DAT-3pL.ERG-say 'Whoever led his/her home son, his parents thank him/her.'
*'No matter who led whose son home, his parents thank him/her.'
If there are two participants that have the same rank in the hierarchy, neither can have a quantificational reading if expressed with a definite pronoun:
(45) Postpositional object = Possessor
a. xet she.c'e a-ba.m ja-s ma caxwa-m
who for that-oBL poss-horse this person-oBL
ja-ma-daKw-a-m-ja, a-xe-r ze-ra-caxw-a-qam
3sg.erg -neg -steal-pst -cond-add that-pl-abs REc.Io-RCP-know-psт-NEG 'For whomever this person stole a horse of his, they did not get to know each other.'
*'No matter this person stole whose horse for whom, they did not get to know each other.'
1 Unfortunately, I have no reliable examples which allow to compare indirect objects and possessors.
fC CN S o, CO z
VD
(45) b. a-ba.m she.c'e xet ja-s ma caxwa-m
that-oBL for who Poss-horse this person-oBL
ja-ma-daKw-a-m-ja, a-xe-r ze-ra-caxw-a-qam | 5
3sG.ERG-NEG-steal-PST-CoND-ADD that-PL-ABS REC.io-RCP-know-PST-NEG
'Whoever's horse this person stole for him/her, they did not get to know each other.'
*'No matter this person stole whose horse for whom, they did not get to know each other.'
Curiously, the same holds for the absolutive and the agentive indirect object of inverse verbs. Recall that these verbs are formally intransitive but have an indirect object which is more agentive than the absolutive. (46) and (47) demonstrate that the quantificational reading is inaccessible for a definite pronoun in either the absolutive position or the indirect object position for the inverse verb 'to be forgotten by' and for a potential inverse verb:
(46) a. xet a-ba.m sa-gwapse-z'-m-ja,
who that-oBL Loc-be.forgotten-RE-coND-ADD ze-xwe-ze-z'-xe-qam
REC. io -BEN-meet-RE-PL -NEG
'Whoever s/he forgot, they don't meet anymore.' *'No matter who forgot whom, they don't meet anymore.' b. a-r xet sa-gwapse-z'-m-ja,
that-ABs who Loc-be.forgotten-RE-coND-ADD ze-xwe-ze-z'-xe-qam
REC. io -BEN-meet-RE-PL -NEG
'Whoever forgot him/her, they don't meet anymore.' *'No matter who forgot whom, they don't meet anymore.'
(47) a. a-ba.m xet xwe-wac'-te-m-ja,
that-oBL who ben-beat-ipf-cond-add jen-u j-e-wac'
whole-ADV 3sG.ERG-DYN-beat 'Whomever s/he could beat, s/he is beating all the time.' *'No matter who could beat whom, s/he is beating him/her all the time.' b. xet a-r xwe-wac'-te-m-ja,
who that-ABs BEN-beat-iPF-coND-ADD
jen-u j-e-wac'
whole-ADV 3sG.ERG-DYN-beat
'Whoever could beat him/her, s/he is beating all the time.' *'No matter who beats whom, s/he is beating him/her all the time.'
-60 7.
Accounting for the restricting function and FC semantics
Izvorski [2000] noted that "intuitively, concessives that involve free adjunct free relatives are interpreted as an exhaustive conjunction of conditionals", so a sentence like Whatever John cooks, he will win the cooking contest is interpreted as 'If John cooks x1, he will win the cooking contest, & If John cooks x2, he will win the cooking contest, & ... If John cooks xn, he will win the cooking contest'. If conditionals were taken to denote implication, Izvorski's representation of the meaning of UCCs could be represented as quantification over the unknown part of the conditions, as in (48). This would correspond to the interpretation of FC as universal quantification.
(48) Vx [||John cooks||(x) ^ ||John will win the cooking contest||]
However, as was said above, following the mainstream view (see, for instance, [Kratzer, 2012]) originating from David Lewis' [1975] work on adverbs of quantification, I assume that conditionals have another function, namely the restriction of the domain of quantification of some operator in a tripartite structure like (49), where x is a variable quantified over.
(49) Opx [restrictorJ [nuclear scopeJ
In (49), the bound variable should be present in both parts of the sentence. Yet as we saw above, the variable corresponding to the interrogative need not be present in the matrix clause. Hence I assume that the universal quantifier should bind some other variable, which has scope over the indefinite pronoun, as in (50). This corresponds to the interpretation of FC as narrow scope existential quantification.
(50) Vx [||John cooks something||(x)] [John will win the cooking contest||(x)]
To solve this problem, I will use the approach called 'Davidsonian semantics' and represent the semantics of UCCs as a kind of quantification over events. In the Davidsonian semantics, every event is associated with an event variable which can serve as an argument not only of verbs but also, for example, of adverbials.1 According to this approach, the English sentence (51a) states the existence of an event as is shown in (51b).
(51) a. John quickly cooked something
b. 3e 3x [cooked(John, x, e) & quick(e)]
1 Kratzer [1995] argued that the event variable is only introduced by stage-level predicates. Yet for simplicity's sake, I assume that event variables are introduced by individual-level predicates too, since the subordinate clause in the Kabardian UCC construction may contain an individual-level predicate (as in (29)) and my proposal should hold for such examples as well. To retain Kratzer's view, the external quantifier can be represented as quantifying over propositions in the style proposed by Kratzer, Shimoya-ma, [2002] for the description of indefinites in Hamblin semantics. See also [Rawlins, 2008, 2013] for an attempt to apply Hamblin semantics to UCCs.
I assume the variable which is quantified over in the UCC construction to be an ro rN
event variable which corresponds to the complex event described in a sentence and ^ ^
includes (sub)events described in the subordinate clause and in the matrix clause. | ^
Then we get the following interpretation for Izvorski's example, where e} is the event jy ^
described in the sentence, which includes the events e2 and e3 corresponding to the ^ subordinate clause and the matrix clause respectively, as its parts:
(52) Ve1 3e2[||John cooks somethingH(e2) & part-of(e2, et)]] 3e3[||John will win the cooking contest||(e3) & part-of(e3, et)]]
Note that here I assume that other variables except the "external event" variable should be bound at the clause level. This is motivated by a mereological hierarchy of events which is reflected in syntax to some extent.
While this has an expected result in the case of Izvorski's sentence, I would like to modify the idea a little further and suggest that the UCC construction involves unselective quantification, i.e. quantification that binds every unbound variable in its scope [Lewis, 1975] (below, such variables are displayed as indices that follow the unselective quantifier). For Izvorski's example this will give the following representation:
(53) Ve1 3e2[||John cooks something||(e2) & part-of(e2, e1)] 3e3[||John will win the cooking contest||(e3) & part-of(e3, e1)]
A similar translation can be proposed for a Kabardian sentence like (54a); cf. (54b) (for simplicity's sake, I represent definite descriptions as constants):
(54) a. sat a-ba ja-ma-wapsef-a-m-ja,
what that-oBL 3sG.ERG-NEG-cook-PST-coND-ADD ja-Xa-r qa-sa-txwa-t
Poss-man-ABs DiR-Loc-compliment-iPF 'Whatever she cooked, her husband complimented her.' b. Ve1 3e23x[cooked(she, x, e2) & part-of(e2, e1)] 3e3[complimented(her-husband, she, e3) & part-of(e3, e1)]
Abstracting away from specific examples, I propose that the semantics of Kabardian UCCs may look as in (55):
(55) Ve1 3e23x[P(x, e2) & part-of(e2, e1)] 3e3[Q(e3) & part-of(e3, e1)]
Here there is an unselective universal quantifier which binds the "external event" variable and there are two subevents corresponding to the restrictor and to the nuclear scope. Note that the restrictor should contain an existentially quantified variable (expressed with the interrogative pronoun) which naturally has narrow scope with respect to the universal quantifier. This narrow scope is a source of the FC reading of the interrogative pronoun.
But what is the source of universal quantification in (55)? I suggest that it is implied from the semantics of FC, which presume existential quantification occurring in the scope of universal quantification [Menendez-Benito, 2010]. In Kabardian, then, universal quantification is accommodated due to the appearance of an interrogative phrase.1
This semantics, of course, does not capture some peculiar meaning components noticed in the literature on UCCs such as mutual exclusivity of the alternatives described within the concessive [Rawlins, 2008, 2013] and the possible unexpectedness of the co-existence of some of these alternatives and the event described by the matrix clause. However, in Kabardian these components turn out not to be necessary, as shown by (56):
(56) st-xwe.de-rje txaX adage-xe-m ja-ha-X-a-we what-like-ADj book Circassian-pL-oBL 3pL.ERG-carry-ALL-psT-ADv s-je-3'-a-m-ja, saw-zaqw s-ja-gw 1sG.Aвs-DAт-read-psт-coND-ADD rider-lone 1sG.io-poss-heart r-ja-h-a
LOC-3sG.ERG-Carry-PST
Lit., 'Whichever book devoted (lit. brought) to Circassians I read, I liked "Lone Rider".'
Importantly, the semantic representation (55) may explain the asymmetry observed in constructions with quantificational definite pronouns. I assume that anaphoric pronouns introduce free variables. If an expression contains an unselective universal quantifier, it may bind the variable introduced by the anaphoric pronoun just as it binds the external event variable.
Consider (40a) repeated here as (57a) and its semantic representation (57b), where the variable x corresponds to the agent of the predicate 'see' and the variable y corresponds to its undergoer:
(57) a. xet a-ba q-ja-ma-XeKw-a-m-ja,
who that-obL DIR-3 SG.ERG-NEG-See-PST-COND-ADD zec'e-m-ja selam ze-r-a-x-xe all-obL-add greeting REc.Io-DAт-3pL.ERG-carry-PL 'No matter who sees whom, all greet each other.' b. Ve1x 3e23y[see(x,y,e2) & part-of(e2, et)] 3e3[all-greet-each-other(e3)]
Here the unselective universal quantifier binds the first event variable and the variable introduced by the definite pronoun. A free variable can be introduced by
1 Alternatively, it can be assumed that the universal quantification is something as a default for conditionals and conditional-like structures in general.
63
an anaphoric pronoun but not by a deictic pronoun, which explains why deictic pronouns are not available in this construction. On the other hand, the variable introduced by the interrogative phrase is bound by an existential operator, according to the FC semantics of this phrase. As in other cases, the presence of an interrogative pronoun is necessary, because it forces the appearance of the universal quantifier.
Now, note that there is scopal asymmetry between the two arguments of the verb 'see' in this example: the ergative argument has scope over the absolutive argument. This may be a clue to the asymmetry discussed earlier in this section. It seems that the argument bound by the universal quantifier must take scope over the argument expressed by the interrogative pronoun.
On the one hand, one can suggest that the scope effects observed here are directly motivated by the syntactic structure. This approach meets several problems. First, there is not much evidence that Circassian languages are syntactically accusative and that the ergative argument should be structurally higher than the absolutive argument (but see [Lander, Testelets, to appear]), on the contrary, there is nontrivial evidence that these languages are syntactically ergative [Lander, 2010, 2015]. Second, a purely syntactic approach cannot explain the effects (or, to be more precise, the absence of effects) observed for "inverse" verbs. On the other hand, the scopal asymmetry may be explained by topicality: more topical elements should have wider scope than less topical elements. The problem is that we cannot compare topicality of non-referential arguments. However, the hierarchy, or at least a part of it (e.g., erg > abs) may be related to some default topicality dimension, which could undergo grammaticalization. Still, it is most likely that several factors come into play in this case, so both the syntactic structure and the default topicality are relevant.
A further problem is related to the syntax-semantics interface. How can the argument expressed with the definite pronoun within the subordinate clause be interpreted at a higher level? There is, however, an additional and quite exotic piece of evidence that suggests that this is possible in Kabardian. In particular, an NP that contains a universal distributive quantifier may intrude into the subordinate clause and bind an argument of the matrix clause, as in (58):
(58) xet Xa-pebz' q-ja-s'-a-m-ja,
who man-every DiR-3sG.ERG-lead-PST-coND-ADD
a-ba.m ja-wane ja-Ke-psa-n xw.je
that-oBL Poss-house 3sg.ERG-cAus-shine-MoD must 'No matter who marries whom, he must decorate his house.'
ID Z .
• VO re
E °
(58) does not mean 'Whomever every man married', presumably for pragmatic reasons. Yet its resulting interpretation suggests that the nominal Aa-pebz' has scope
^ over the whole sentence, as it serves as an antecedent of the ergative pronoun of the 1 matrix clause. It is worth noting that in other constructions binding of an element Fo of the matrix clause by a quantifier in the subordinate clause has been observed in | Circassian before [Testelets, 2009; Lander, Testelets, to appear]. Yet, in our case it <S harmonizes with the interpretation proposed for the whole UCC construction. The agent variable in (58) is universally quantified explicitly and hence need not be bound by unselective quantification:
(59) Ve1 Vx 3e2 3y [see(x,y,e2) & part-of(e2, et)] 3e3 [all-greet-each-other(e3)]
Recall that [Lander 2010, to appear] presented evidence that (some) NPs in Circassian languages need not be included in the "core" of the clause but may be interpreted somewhere outside. The findings presented here support this idea for Kuban Kabardian.
8. Conclusion
In this paper, I described the UCC construction of Kuban Kabardian and proposed a semantic analysis based on an assumption that one of its principal constituents, the interrogative phrase has free choice semantics described as existential quantification within the scope of a universal operator. I showed that this universal operator should or may be unselective, which explains such an unusual pattern as the quantificational use of a definite pronoun (bound by the unselective universal quantifier).
It is important that the possibility of the patterns found for UCCs in Kabardian depends on the polysynthetic nature of the language. In particular, I assumed that the binding of definite pronouns and the unexpected use of an NP containing the distributive quantifier 'every' shown in (58) are both related to the possibility of (some) NPs to be interpreted somewhere outside the "core" of the clause. However, I have neither specified the status of this core, nor outlined the basic architecture of the Kabardian sentence, which is needed for the full compositional treatment of this construction. The problem is that for the time being we are not in a position where we can describe this syntactic structure in sufficient details, so I have to postpone any elaboration on a complete theory of Kabardian UCCs.
In fact, achieving this goal is even more difficult, since there are elements which are commonly used for detecting the details of the syntactic structure but which in Circassian languages show very idiosyncratic properties, such as the distributive universal quantifier which may even form predicative NPs [Arkadiev, Lander, 2013]. It may be that it is these idiosyncratic properties that allow NPs with the distributive quantifier to appear in constructions like (58), but combining these details of the puzzle requires giving up many standard assumptions. This step could be supported or precluded by cross-linguistic data, but there is still much to do with the semantics of polysynthetic languages.
L
References S ™
ct. Z
Abitov et al. (eds)., 1957 - Grammatika kabardino-cerkesskogo literaturnogo jazyka. Abi- ^ ^ tov M.L. et al. (eds). Moscow: Izd. AN SSSR, 1957. £ ^
Arkadiev, Lander, 2013 - Arkadiev P., Lander Yu. Non-quantificational distributive quantifiers in Besleney Kabardian. Snippets. 2013. Issue 27. Pp. 5-7.
Bagov P.M. et al. (eds)., 1970 - Grammatika kabardino-cerkesskogo literaturnogo jazyka. Bagov P.M. et al. (eds). Moscow: Nauka, 1970.
Baker, 1996 - Baker M.C. The polysynthesis parameter. New York: Oxford University Press, 1996.
Colarusso, 1989 - Colarusso J. East Circassian (Kabardian dialect). In: The indigenous languages of the Caucasus. Vol. 2. The North West Caucasian languages. Hewitt B.G. (ed.) New York: Caravan Books, 1989. Pp. 261-355.
Colarusso, 1992 - Colarusso J. A grammar of the Kabardian language. Calgary: University of Calgary Press, 1992.
Colarusso, 2006 - Colarusso J. Kabardian (East Circassian). München: LINCOM Europa, 2006.
Gawron, 2001 - Gawron J. M. Universal concessive conditionals and alternative NPs in English. In: Logical perspectives on language and information. Condoravdi C., de Lavalette G.R. (eds) Stanford: CSLI Publications, 2001. Pp. 73-105.
Giannakidou, 2001 - Giannakidou A. The meaning of free choice. Linguistics and philosophy. 2001. Vol. 24. Pp. 659-735.
Haspelmath, 1997 - Haspelmath M. Indefinite pronouns. Oxford: Clarendon, 1997.
Haspelmath, König, 1998 - Haspelmath M., König E. Concessive conditionals in the languages of Europe. In: Adverbial constructions in the languages of Europe. van der Auwera J. (ed.). Berlin: Mouton de Gruyter, 1998. Pp. 563-640.
Izvorski, 2000 - Izvorski R. Free adjunct free relatives. In: WCCFL 19: Proceedings of the 19th West Coast conference on formal linguistics. Billerey R., Lillehaugen B.D. (eds). Somerville: Cascadilla Press, 2000. Pp. 232-245.
Jakovlev, 1948 - Jakovlev N.F. Grammatika literaturnogo kabardino-cerkesskogo jazyka. Moscow: Izd. AN SSSR, 1948.
Jayez, Tovena, 2007 - Jayez J., Tovena L. Free Choiceness and Non Individuation. Linguistics and philosophy. 2007. Vol. 28. Pp. 1-71.
Jelinek, Demers, 1994 - Jelinek E., Demers R.A. Predicates and pronominal arguments in Straits Salish. Language. 1994. Vol. 70. Pp. 697-736.
Kapitonov, 2010 - Kapitonov I. Adyghe indefinites: Formal aspects. Paper presented at DGfS 32. Jahres-tagung, 26 February 2010.
Kornflit, 1997 - Kornflit J. Turkish. London: Routledge, 1997.
Kratzer, 1995 - Kratzer A. Stage-level and individual-level predicates. In: The generic book. Carlson G.N., Pelletier F.J. (eds). Chicago: The University of Chicago Press, 1995. Pp. 125-175.
Kratzer, 2012 - Kratzer A. Modals and conditionals. Oxford: Oxford University Press, 2012.
Kratzer, Shimoyama, 2002 - Kratzer A., Shimoyama J. Indeterminate pronouns: The view from Japanese. In: The proceedings of the Third Tokyo conference on psycholinguistics (TCP 2002). Otsu Y. (ed.). Tokyo: Hituzi Syobo, 2002. Pp. 1-25.
Kumakhov, 1969 - Kumakhov M.A. Kubanskij dialekt. In: Ocherki kabardino-cerkesskoj dialektologii. Kumakhov M.A. (ed.). Nalchik, 1969. Pp. 190-248.
Kumakhov (ed.), 2006 - Kabardino-cerkesskij jazyk. 2 vols. Kumakhov M.A. (ed.). Nalchik: El-Fa, 2006.
? Lander, to appear - Lander Yu. Circassian relative clause constructions. In: Relative clauses
o in West Asia. Donabedian A., Samvelian P. (eds). To appear.
2 Lander, 2005 - Lander Yu. A. "Pronominal'nye argumenty" i "adjunktnye imennye grup-
py" v adygejskom jazyke. In: Vtoraja konferencija po tipologii i grammatike dlja molodyx issledovatelej. Vydrin A.P. et al. (eds). St. Petersburg: Nauka, 2005. Pp. 90-95.
Lander, 2010 - Lander Yu. Subject properties of the Adyghe absolutive: evidence from relatives. Unpublished ms., 2010.
Lander, 2014 - Lander Yu. Philippine ergativity in a cross-linguistic perspective. Paper presented at the 12th Philippine Linguistics Congress, University of the Philippines - Diliman, Quezon-City, 26 November 2014.
Lander, Testelets, to appear - Lander Yu., Testelets Y. Adyghe. In: The Oxford handbook of polysynthesis. Fortescue M. et al. (eds). Oxford: Oxford University Press. To appear.
Letuchiy, 2012 - Letuchiy A. Ergativity in the Adyghe system of valency-changing derivations. In: Ergativity, valency and voice. Authier G., Haude K. (eds). Berlin: Mouton de Gruy-ter, 2012. Pp. 323-354.
Letuchiy, 2013 - Letuchiy A. Verb classes in Adyghe: derivational vs. nonderivational criteria. Linguistics. 2013. Vol. 51. No. 4. Pp. 729-766.
Lewis, 1975 - Lewis D. Adverbs of quantification. In: Formal semantics of natural language. Keenan E.L. (ed.). Cambridge: Cambridge University Press, 1975. Pp. 1-15.
Liptak, 2009 - Liptak A. The landscape of correlatives. An empirical and analytical survey. In: Correlatives cross-linguistically. Liptak A. (ed.). Amsterdam: John Benjamins, 2009. Pp. 1-46.
Makri, 2012 - Makri M.-M. Expletive negation beyond Romance: clausal complementation and epistemic modality. M.A thesis, University of York, 2013.
Menendez-Benito, 2010 - Menendez-Benito M. On universal Free Choice items. Natural language semantics. 2010. Vol. 18. Pp. 33-64.
Rawlins, 2008 - Rawlins K. (Un)conditionals: An Investigation in the Syntax and Semantics of Conditional Structures. Ph.D. diss., UCSC, 2008.
Rawlins, 2013 - Rawlins K. (Un)conditionals. Natural language semantics. 2013. Vol. 40. Pp. 111-178.
Srivastav, 1991 - Srivastav V. The syntax and semantics of correlatives. Natural language and linguistic theory. 1991. Vol. 9. Pp. 637-686.
Testelec, 2009 - Testelec Ja. Nevyrazennye aktanty v polipredikativnoj konstrukcii. In: Aspekty polisintetizma: ocerki po grammatike adygejskogo jazyka. Testelec Ja. et al. (eds). Moscow: RGGU, 2009. Pp. 654-711.
Turchaninov, Tsagov, 1940 - Turchaninov G., Tsagov M. Grammatika kabardinskogo jazyka. I. Moscow, Leningrad: Izd. AN SSSR, 1940.
van Riemsdijk, 2006 - van Riemsdijk H. Free relatives. In: The Blackwell companion to syntax. Vol. II. Everaert M., van Riemsdijk H. (eds). Oxford: Blackwell, 2006. Pp. 338-382.
Vendler, 1967 - Vendler Z. Linguistics in philosophy. Ithaca: Cornell University Press, 1967.
Xrakovskij (ed.), 2012 - Typology of concessive constructions. Xrakovskij V.S. (ed.). München: LINCOM, 2012.
Yoon, 2011 - Yoon S. 'Not' in the mood: the syntax, semantics and pragmatics of evaluative negation. Ph.D. diss., the University of Chicago, 2011.
