6 the Caucasian Languages

kort_006.pod 125 07-12-08 13:18:12 -mt- mt

6 The Caucasian Languages

Michael Daniel and Yury Lander

1. Introduction 2. Sociolinguistic situation 3. Phonetics and phonology 4. Morphological makeup 5. Grammatical categories 6. Syntax: some highlights 7. Perspectives

1. Introduction

Europe’s linguistically most exotic area is the Caucasus. In terms of linguistic den- sity it is the subcontinent’s New Guinea. Languages of Western, Central and East- ern Europe are less typologically diverse – and not much more numerous – than the languages spoken in its southeastern corner. Three “endemic” language families are spoken here, South Caucasian (Kartvelian), Northeast or simply East Cau- casian (Nakh-Daghestanian) and Northwest or simply West Caucasian (Abkhaz- Adyghe). The latter two are sometimes considered to form a deep-level North Cau- casian family (see Nikolaev and Starostin 1994), but this entity is disputed. An earlier hypothesis of genealogical relationships among these families (the assumed Ibero-Caucasian family) has now been largely abandoned (cf. Tuite 2008). The linguistic diversity of the area is further expanded by the presence of Turkic and Indo- European languages. The South Caucasian family includes four languages that are relatively close from the typological point of view. West Caucasian is somewhat less homogene- ous, counting five languages, which split into two branches. East Caucasian is the most diverse, with over 30 languages divided into eight branches.

Endemic language families: South Caucasian: Georgian (about 5 million speakers), Mingrelian (Megrelian) and Laz (sometimes merged into one Zan language, about half a million and several dozens of thousands speakers, respectively) and Svan (less than 20 thousand); West Caucasian: Kabardian (East Circassian), Adyghe (West Circassian), Abkhaz, Abaza and Ubykh, ordered according to the number of speakers from about one million in Kabardian (although the demographic data may not be quite reliable) to Abaza with several dozens of thousands speakers to the now extinct Ubykh; kort_006.pod 126 07-12-08 13:18:12 -mt- mt

126 Michael Daniel and Yury Lander

Map 1: Languages of the Caucasus

East Caucasian: the Nakh branch, including Chechen (about 1 million), Ingush (over 200 thousand) and Bats, alias Tsova-Tush (according to some accounts, a few hundred speakers); the Tsezic branch, including Tsez (alias Dido), Bezhta, Hunzib, Khwarshi, Hinuq (from under 10 thousand speakers for Tsez to less than one thousand for Khwarshi and Hinuq); Avar (over half a million); the Andic branch, including Andi (about 20 thousand), Tindi, Chamalal, Bot- likh, Karata, Akhvakh, Godoberi, Bagvalal (each less than ten thousand speakers and as few as two thousand for Bagvalal); Lak with over 100,000 speakers; the Dargwa branch, including Kubachi, Icari, Mehweb and other (Dargwa has long been considered to be a dialect cluster and does not have a generally accepted classification; speakers of Dargwa languages count between 300,000 and 400,000); Khinalug (around 1,500); the Lezgic branch, including Lezgian (over 400,000), Tabasaran (under 100,000), Rutul, Tsakhur and Agul (all about 20,000), Udi (about 7,000), Kryz (less than 4,000 each), Archi (about 1,500) and Budukh (500 at most). kort_006.pod 127 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 127

Non-endemic language families: Turkic: Azerbaijani, Kumyk, Nogai, Karachay-Balkar Indo-European: Armenian (Eastern) and the Iranian languages Ossetic and Tat.

Below, we will primarily deal with endemic families, sometimes also with Arme- nian (for the purposes of this survey, the difference between Eastern and Western Armenian may be neglected), to a lesser extent with Iranian languages, which are also spoken outside the Caucasus, and minimally with the Turkic languages, as Turkic is dealt with in a separate chapter of the present volume. There are only two languages with an uninterrupted written tradition that goes back to the early Middle Ages: Georgian and Armenian, both of which use their own writing system. The earlier stages of these languages, Old Georgian and Clas- sical Armenian (Grabar) were in many respects significantly different from the re- spective modern languages and will not be considered either. Until recently, only these two languages provided reliable data for studies of language-internal development (a small number of written records of Avar are available from the 14th century on). The situation has changed due to the discovery of a substantial manuscript in Caucasian Albanian, a language which had been previously known from a very limited number of inscriptions found on the territory of the modern northern Azer- baijan and which proved to be an early (around the middle of the first millennium) dialect of Udi (Gippert et al. 2009). Major East and West Caucasian languages used Arabic-based systems throughout the 19th century, but these were substituted by Cyrillic-based systems introduced in late thirties (after a brief period of using the Latin alphabet). Since around 2000, Azerbaijani only uses a Latin-based alphabet, close to the Turkish one. Some languages have introduced writing systems in the last two decades (e.g. Udi, Rutul, Tsakhur), some have recent writing systems that are not used in practice (e.g. Archi and Khinalug), and some remain unwritten (various Tsezic and Andic languages, Svan).

2. Sociolinguistic situation

An important factor fostering language diversity in the Caucasus is the mountain- ous landscape, but there are sociocultural factors as well; for instance, until recently many of the communities were endogamous. Many languages, including Armenian and Avar, show considerable internal variation, to the extent that different dialects are not easily mutually intelligible. Several East Caucasian languages are spoken in one village only, viz. Godoberi, Archi, Budukh, Khinalug, Bats. Mi- gration may lead to peculiar sociolinguistic situations of co-existence of and inter- ference between two distinct dialects within one village (the dialect of the original population and that of newcomers), observed in some regions of Armenia. kort_006.pod 128 07-12-08 13:18:12 -mt- mt

128 Michael Daniel and Yury Lander

The ecological situation varies greatly, and no complete picture can be drawn because of the lack of systematic sociolinguistic surveys. There seem to be very few moribund languages in terms of Krauss (1992), as most languages are still learned as first languages. Only one Caucasian language, Ubykh (West Caucasian), is known to have become extinct. In the second half of the 19th century, the Ubykhs were forced to leave Russia; those who survived settled in Turkey, and the last speaker of Ubykh, Tevfik Esenç, died on November 7, 1992. Highly endangered or, according to some accounts, even moribund are Budukh (Lezgic) and Bats (Nakh). But many other languages are at least potentially endangered because of politically and economically driven migrations, especially by the outflow from native villages to the towns, where Russian, Azerbaijani or Georgian is the dominant if not the only means of interethnic communication. In Tsezic or Andic language communities, a major language of the area, Avar, is taught in schools as a mother tongue. Similarly, only Lezgic or Azerbaijani are taught to speakers of some Lezgic languages and Khinalug and only Georgian is taught to the speakers of Mingrelian and Svan. The combination of all these factors makes the survival prospects of these languages unpredictable. Diasporic situations are very characteristic of the Caucasian languages in the 20th century. Just as Ubykhs, many Circassians (i.e. speakers of Adyghe and Ka- bardian) were expelled from Russia to settle in Turkey, Jordan, Israel and elsewhere in the Near East. Ethnic cleansings in Turkey in the early 20th century led to the almost complete emigration of the surviving Western Armenians and the growth of Armenian diasporas in Europe and the USA. Today, Russia has sizeable diasporas of Georgians, Armenians, Chechens, Dagestanians and Azerbaijanians, partly dating back to Soviet times but reinforced by recent economic and political events, such as the collapse of the Soviet Union and the Chechen wars. Multilingualism is very common in the Caucasus. In the last two centuries, Russian spread throughout the area as a language of urban interethnic communication, although now its role obviously diminishes in Armenia, Georgia and Azer- baijan. In some areas of Daghestan, Avar and Lezgian serve as linguae francae on par with Russian. As a result, speakers of some minor languages of Daghestan have a full command of at least three languages – the language of their village, a major language dominant in the area, and Russian. Sometimes, the same language is spoken on the two sides of a national border, which produces parallel patterns of bilingualism with the same mother tongue but different literacies (as is the case of Lezgian, Avar and Akhvakh spoken both in Russia and Azerbaijan; Ossetic spoken in Russia and Georgia, Udi spoken in Azerbaijan, Georgia and Russia etc.). Cau- casian languages provide a rich field for investigating code-switching. kort_006.pod 129 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 129

3. Phonetics and phonology

Caucasian languages are famous for their rich phonological inventories, extremely abundant in consonants. Georgian and Armenian both have a considerable array of consonantal elements (including a typologically rare opposition between a trilled r and a flapped ɾ in Armenian), but this richness reaches an extreme in West and East Caucasian. Some languages, such as Ubykh and Archi, count over 70 phonologi- cally distinct consonants. Ejectiveness involves not only stops but also fricatives; it is quite widespread in each of the three endemic families but is also present in Ossetic and, according to some accounts, starts to develop in Armenian and Kumyk dialects. Other phonological features of consonants include voicing, aspir- ation, tenseness or gemination and labialization, which is widespread as a phonological feature with relatively free combinability (as in Avar). Consonant palatalization is rare, but attested in e.g. Ubykh, Tsakhur, Khinalug. In the latter, a palatalized consonant preceding a back vowel or a non-palatalized consonant preceding a front vowel – so to speak, the reverse of palatalization – signal a morpheme boundary (e.g. nk-e mountain-GEN, where the subscript prime may be interpreted as ‘no expected palatalization occurs’) Typologically rare consonants include laterals [, , ’] (IPA k, k, k’), which are quite common in East Caucasian, and less common pharyngealized obstruents in e.g. Ubykh and Archi (see below). East Caucasian also features an extensive inventory of laryngeal and pharyngeal consonants, typically four [h, ʔ, ʕ, ] but adding the epiglottal stop [] in Budukh and Agul; these are fewer in West Caucasian (between one and three) and only one – [h] – in most South Caucasian languages. Table 2 demonstrates the consonantal system of Archi; pharyngealization [ʕ] combines with the uvulars; labialization [w] combines with virtually all obstruents except for the labials, laryngeals, and pharyngeals and thus nearly doubles the number of consonants. Bracketed are the obsolete lateral voiced fricative [], which is probably also secondary from the point of view of morphophonology, and [ʕ, ], which are almost exclusively used in loanwords. Consonant clusters are rare in East and West Caucasian, but widespread in South Caucasian and Armenian. It is not uncommon for Georgian and Armenian to have four or more consonants following each other in orthography: cf. a one-syl- lable Georgian word vbrdγvnis (‘he tears us into pieces’). However, their phonetic realization (and sometimes phonological representation) may involve schwas which are not spelled in writing: they are shown as (ə) in the following Armenian example: m(ə)sˇt(ə)nˇenakan ‘eternal’. Vowel inventories are poor in West Caucasian, standard in South Caucasian and vary from regular to rich in East Caucasian languages. West Caucasian languages distinguish two or three vowels only. South Caucasian has ‘triangular’ systems based on the standard five vowel inventory [i – e – a – o – u] (Svan is a more complicated case). The same inventory is present in many East Caucasian lan- kort_006.pod 130 07-12-08 13:18:12 -mt- mt

130 Michael Daniel and Yury Lander

Table 1: Consonantal inventory of Archi (Kibrik et al. 1977a, 4: 223–248; Kibrik and Kodzasov 1990: 335)

bd g(w) pt(w) c(w),cˇ(w)(w) k(w) q(w)(ʕ pt¯ ¯(w) k¯ (w) p’ t’(w) c’(w),cˇ’(w) ’(w) k’(w) q’(w)(ʕ) ʔ c’¯ (w),cˇ¯’(w) q’¯ (w)(ʕ) z(w),zˇ(w) [] ʁ(w)(ʕ) [ʕ] s(w),sˇ(w) (w) χ(w)(ʕ) [h¯ ]h s¯(w),sˇ¯(w) (w) χ¯ (w)(ʕ) m, w n r, l, j

guages (Avar, Andic) and is the basis of the vowel systems in many others, though some south Lezgic languages may be described as ‘rectangular systems’. Vari- ations on the triangular theme include adding a middle vowel (for instance, [ə] in Archi or [ ] in some Tsezic languages or distinguishing two “a”s in Lak), introducing a row of fronted back vowels etc. (cf. Kibrik and Kodzasov 1990). Vowel length (attested in various branches of East Caucasian and West Caucasian and dialects of Svan) and nasalization (attested in Nakh, Tsezic and Andic branches of East Caucasian) are on the whole rare and peripheral. An important phonetic feature is pharyngealization, reported for Ubykh and various branches of East Caucasian (Lezgic, Tsezic, Lak, Dargwa); see also Ni- chols (2000). It is essentially a prosodic feature, but sometimes it ‘focuses on’ or ‘gravitates to’ one or other phoneme and may thus be interpreted as a phonological property. Most often its focus falls on a vowel (which duplicates the number of vowel phonemes), but e.g. Ubykh and Archi have pharyngealized consonants. Vowel harmony is rare (outside Turkic). To various extents it is attested in Lez- gian, Budukh, Khinalug, Udi, Kryz (Kibrik and Kodzasov 1990), Tat,all of which are spoken in Azerbaijan or southern Daghestan, while some dialects of Kumyk, a Turkic language which is supposed to have the characteristic Turkic vowel harmony but is strongly influenced by East Caucasian, are reported to have unstable vowel harmony. The status of stress in the Caucasian languages is studied rather unevenly. In general, fixed stress seems to be rare (found in Armenian). Other languages have stress mobility across paradigms, e.g. in direct vs. oblique nominal stems (e.g. Avar). kort_006.pod 131 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 131

4. Morphological make-up

Caucasian morphology is close to the agglutinative prototype. This is not to say that there is no fusion at morpheme boundaries; internal sandhi may be quite complex, depending on the language. However, combining two distinct grammatical meanings in one marker is rare, while conveying the same grammatical meanings by different markers depending on the class of the stem (declensions and conjugations of early Indo-European or e.g. Slavic) is absent or quite infrequent (with a possible exception of Svan). Armenian, which used to have nominal declensions in its classical period, only preserved them peripherally, converging with the agglutinative model of other Caucasian languages; Ossetic underwent a similar development. Major inflectional categories are expressed by suffixation; but most Caucasian languages also use other kinds of affixation. Thus, various Caucasian languages have systems of verbal prefixes, including West and South Caucasian, Ossetic, Nakh languages, Lezgic languages, Dargwa and Khinalug. Often these preverbs have clear spatial functions or origins; in Lezgic most of them are formally close to nominal spatial suffixes. Pre-root position in the verbal complex is where most argument marking takes place in West Caucasian. Personal agreement markers in South Cau- casian and class agreement markers in East Caucasian are likewise primarily preverbal. In some East Caucasian languages, class agreement prefixes may appear on nouns (typically, kinship terms) and, in Archi, in some case forms of personal pronouns. Adjectives, participles, and sometimes case forms of nouns have either post- root (e.g. Lezgic and Andic) or pre-root (e.g. Nakh, Tsezic and Dargwa) agreement. Abundant verbal infixation is a salient morphological property of East Cau- casian. It primarily involves agreement class markers, but sometimes also other categories, such as imperfective in Archi or prohibitive in Budukh and Rutul. As class infixes are often identical to class prefixes, it is quite likely that these are former prefixes that were trapped between the verbal root and a former preverbal element. In Udi, personal markers, similarly trapped, are inserted, by analogy, into what may never have constituted a complex verb. Interestingly, the same markers occur not only inside the verbal root but also attach to different constituents word- finally. Harris (2002), who first introduced the notion of ‘trapped morphology’, considers them to be clitics that also occur word-internally and hence constitute a typologically rare category of endoclitics: (1) Udi endoclitics: same marker as infix and (apparent) clitic beʕ‹=ne=›ʁ-sa, iz bava=ne.

look1‹=3SG=›look2-PRS self.GEN father=3SG ‘He looks up – (that is) his father’. West Caucasian languages are often described as polysynthetic. There is no noun incorporation in the sense of Mohawk or Chukchi, but morphology can express much of what is conveyed by syntax in most languages of the world: kort_006.pod 132 07-12-08 13:18:12 -mt- mt

132 Michael Daniel and Yury Lander

(2) Kabardian: polysynthetic complex (Kumakhov 1989: 123) w-a-q’ə-də-Ø-d-Ø-je-z-ʁe-sˇə-zˇə-f-a-te-q’ə.m-jə 2SG.S/P-3PL.IO-DIRECTIVE-COMIT-3SG.IO-LOC-3SG.IO-OBL.PRVB-1SG.ACAUS- bring-REVERSIVE-ABLE-PST-THAT.TIME-NEG-AND ‘And that time then I was not able to make him bring you back out of there with them’.

Utterances like (2) may not be frequent because of their morphological complexity, but the important thing is that they are structurally possible. To put it differently, there is a kind of competition or overlap of potentials, between syntax and morphology, at least in some domains: some functions that are more often expressed morphologically can be covered by syntax, and some functions rendered by syntax can be expressed morphologically. For instance, there are both reciprocal prefixes and independent reciprocal pronouns, or both an inchoative verb and an inchoative suffix on the verb. Unsurprisingly, morphology of this kind manifests some properties of syntax. Certain slots in the predicative complex are optional rather than obligatory, and affix ordering is often more variable than it is in non-polysynthetic languages. The fact that reflexives, reciprocals, and relativizers belong to the same category (hold the same slots) as personal indexes also suggests a parallel to independent pronouns in syntax. Yet these similarities with syntax are only partial. Morphology may apply specific constraints to the ‘grammar of verbal slots’ which, in combination with the phonetic properties of the complex, preclude a multi-word analysis of these word forms. Derivational morphology varies across families. In South Caucasian and especially in Armenian it is similar to European patterns, with root compounding and various denominal and de-adjectival suffixes. West Caucasian is traditionally also believed to have abundant derivation, but it is difficult to distinguish it from ‘inter- nalized syntax’ discussed above. In East Caucasian, derivation plays a much more important role in verbal than in nominal morphology: the number of non-derived verbs is often quite small, especially in Lezgic (as few as a dozen in Khinalug), and other verbal meanings are derived by using primary verbs as conjugated heads in combination with nouns, adjectives and formally autonomous elements with un- clear origins. Such conjugated heads – light verbs or conjugators – are most often abstract roots such as ‘do’, ‘become’, ‘say’, ‘give’, etc. In terms of phonetics and morphosyntax, the degree of univerbation varies and the conjugators may eventually grammaticalize into semantically non-transparent thematic affixes. Thus Khinalug has ten odd conjugators, most of which can not be easily traced back to any autonomous verb. Another type of univerbation is grammaticalization of autonomous verbs into suffixes or grammatical clitics. One example is the so-called verificational mood suffixes in Archi and Agul; these are very likely to originate from the verb ‘see’. kort_006.pod 133 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 133

Interestingly, these clitics may preserve some predicative properties, for instance, introducing their own arguments. (3) Agul: verificational marker (Maisak and Merdanova 2004) sift¯a zun rusˇ-a-s gada k¯andej-cˇuk’-ase at.first I(ERG) girl-OBL-DAT boy love-VERIF-FUT ‘First I’ll find out whether the girl loves the boy’. The verb ‘say’ is strongly grammaticalized in reported speech constructions (half- way from clitic to suffix) in many East Caucasian languages. In Archi and Agul it is cliticized to the main predicate it but can introduce its own arguments, the person whose speech is being reported (eventually resulting in having two agentive er- gatives in the same clause) and the addressee for quotative; cf. also the ‘verificator’ participant in (3). In Archi, the quotative marker may also have its own auxiliary. Sentence (4) is an interesting example of mismatch between phonetic and morphosyntactic word boundaries: the quotative clitic depends on the auxiliary, which is an autonomous word. (4) Archi: mismatch between phonetic [] and morphosyntactic {} boundaries to-r “zari a‹b›cˇu [be¯c’u-q” -{er-sˇi] that-2 I.ERG ‹HPL›kill.PFV [be.able.PFV-POT -{QUOT-CVB] e‹r›di -li} ‹2›AUX.PFV-EVID} ‘She was saying that she will be able to kill them’.

5. Grammatical categories

Case inventories. Nominal case is typical of the Caucasus. The exception is West Caucasian, in which cases are few to none. Abkhaz and Abaza have no core cases at all, while Circassian languages and Ubykh only have two, nominative (or absolutive, depending on the descriptive tradition) and oblique (traditionally called ergative). South Caucasian case inventories are typical for a medium-size ergative case system. The core cases are nominative for S/P (the only argument of an intransitive predicate or the patient of a transitive predicate), ergative for A (the agent of a transitive predicate) and dative, which also plays an important role in core argument marking (see Section 6 below). Non-core cases include genitive, instrumental, and a case form called “adverbial” or “transformative” (used in contexts like ‘become someone’, ‘turn into something’). Mingrelian and Laz additionally have two local cases, lative and elative, but, interestingly, no dedicated locative marker. To express general location, Mingrelian often uses dative (in Georgian this use is more kort_006.pod 134 07-12-08 13:18:12 -mt- mt

134 Michael Daniel and Yury Lander

limited and peripheral) or postpositions with genitive or dative. Less typologically common are a special case form for vocative in Georgian and benefactive in Min- grelian. Modern Eastern Armenian also has the usual medium-size case inventory including nominative, genitive, dative, ablative, locative, and instrumental. Dative and genitive are formally identical (except in a handful of nouns and personal pronouns); dative typically occurs with a definite article, while genitive cannot take it (a rare point where Eastern and Western Armenian diverge). There is no dedicated accusative marker for direct objects: Armenian uses dative for human patients and nominative for non-human ones (the treatment of non-human animates varies), which is a widely attested differential object marking pattern. In a way, Armenian has only two core cases: nominative or direct, covering subjects and inanimate patients, and genitive/dative or oblique, covering all other functions, including possession, some common types of essives (similarly to Mingrelian), indirect objects and animate direct objects. Interestingly, Udi, which has a very different case system but has been in a close contact with Armenian, shows an identical pattern, using dative for definite patients as opposed to nominative for indefinite ones. Such distribution is unique in East Caucasian, where it belongs, differential object marking being on the whole atypical of ergative languages. Differential object marking under a different guise of marked vs. unmarked accusative is also a common property of Turkic languages including Azerbaijani, another source of strong structural pressure on Udi. Similar patterns are found in Iranian languages, e.g. Ossetic. In Circassian, too, P may be marked or left unmarked, and this also correlates with its specificity. However, it is not obvious that we deal with differential object marking here, as the same is observed on S; and proper names and possessives rarely carry S/P or A marking, most probably because of their inherent specificity. Ossetic has a sizeable inventory of nine cases including four spatial forms – allative, ablative, locative and superessive. This is not typical of the modern Iranian languages, and only few case markers seem to go back to Indo-European. The present inventory is thus an innovation, which may be due to structural pressure from other Caucasian languages. It is the East Caucasian languages which are typologically unparalleled in this respect – the crowning glory of nominal declension in the Caucasus (Kibrik 2003b, and Daniel and Ganenkov 2009 for a brief overview). Here, nominal paradigms can count many dozens of forms. The paradigm splits into two sets of forms, syntactic (abstract) cases and the spatial subparadigm. The inventory of syntactic cases includes the usual labels – nominative (alias absolutive), ergative, genitive. East (and South) Caucasian are in general experiencer-prominent: unlike many European languages, they consistently mark most experiencers with dative. But some additionally introduce a special case: affective, marking experiencers with modal verbs, verbs of perception or emotion etc. This case is probably of spatial kort_006.pod 135 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 135

origin, attested first of all in Andic but also e.g. in Tsakhur. Sometimes the system is further extended by the comitative ‘with’ as well as less common cases such as forms meaning ‘about’ or ‘instead of’. But it is the spatial subparadigm which is a true cornucopia and the focus of typologists’ attention. Spatial forms commonly include at least two suffixes: a localization marker, which specifies a spatial domain with respect to a landmark, and an orientation marker, which indicates motion to, from or through this domain or absence of motion. Some Tsezic and Dargwa languages have a third slot, but this is less common; in Tsezic it is often described as approximate location, while in Dargwa the third slot is used for vertical or personal deixis. The inventory of orientations may include two to five and more items: essive, elative, lative, allative, sometimes also translative. The inventory of localizations varies from none (more precisely – no productive forms) in Udi or two in Khinalug to eight in Agul. Most common localizations include IN (often irregular), location inside a hollow object or in a building – ‘in the house’, ‘in the jug’; SUPER, location on the upper surface of an object – ‘on the table’; CONT, location on a surface with a strong fixation on the supporting plane – ‘picture on the wall’; SUB, location under the landmark – ‘under the table’; APUD, location in the vicinity of (and independently of the contact with) the landmark – ‘near the wall’. Cf. an example of localization x orientation combinations in Lak in Table 2.

Table 2: Some spatial forms from Lak from the total of about 30 (Zhirkov 1955: 36–37)

qat-lu-w¯ qat-lu-wu-n ¯ qat-lu-w-a ¯ house-OBL-IN house-OBL-IN-LAT house-OBL-IN-EL ‘in the house’ ‘into the house’ ‘from inside the house’ qat-lu-j¯ qat-lu-j-n ¯ qat-lu-j-a ¯ house-OBL-SUP house-OBL-SUPER-LAT house-OBL-SUPER-EL ‘on the top of the house’ ‘onto the house’ ‘from the top of the house’ qat-lu-¯ χ qat-lu-¯ χu-nqat-lu- ¯ χ-a house-OBL-POST house-OBL-POST-LAT house-OBL-POST-EL ‘behind the house’ ‘to the back side of the house’ ‘from behind the house’

It is not always possible to draw a clear-cut line between East Caucasian spatial forms and syntactic cases, as the former often develop non-spatial meanings and thus expand into the domain of abstract cases. Abstract meanings conveyed by spatial forms usually include such typologically rare categories as unintentional agent (do something inadvertently) or temporary recipient (lend to someone; typically one of the latives) or may be introduced by verbs such as ‘be afraid’ (of what; typically one of the elatives) or ‘say’ (to whom; often a lative or essive of APUD). However, even highly abstract usages such as temporary recipient may preserve elements of spatial semantics (‘hand something over to someone’ rather than ‘lend something to someone’) kort_006.pod 136 07-12-08 13:18:12 -mt- mt

136 Michael Daniel and Yury Lander

(5) Avar: temporary recipient expressed by a spatial form (Daniel et al. 2010) di-q-e ’e-cˇ’o-go, di-je sˇ¯aj du-ca I.OBL-APUD-LAT give.IPFV-NEG-SEQ I.OBL-LAT why you.sg.OBL-ERG ha-b t’ex ’-ola-r-e-b? this-N book give-FUT-NEG-PART-N ‘Instead of giving this book to me for a while, why don’t you offer it to me?’ Crosslinguistically, spatial adpositions may be quite close to suffixes, bound to the host so strongly that one may question their status as autonomous words. Thus, Vogt (1971) considers Georgian postpositions to be ‘secondary case markers’. Yet East Caucasian localization markers are different in that they leave no doubt about their morphological nature; they are bound to the stem in the same way as syntactic cases are, being added to the oblique stem. The use of oblique stems is another peculiar feature of the East Caucasian languages. One case is the unmarked nominative (absolutive). It is opposed to all other cases that are formed from the so-called oblique stem: these are oblique cases. Oblique stem formation patterns are diverse, but most commonly include an oblique stem suffix, sometimes in combination with further modifications of the stem (stress shift, vowel change, etc). The oblique stem is sometimes identical to one of the core case forms, genitive in some languages, ergative in the others. The plural, which is derived from the singular by suffixation, may behave in exactly the same way: the nominative is identical to the nominal stem plus plural suffix, and in oblique cases it is extended by a plural oblique suffix followed by case markers (identical in the singular and the plural). This pattern is illustrated in Table 3 for Lezgian. It is present in quite a few East Caucasian languages, and some other patterns of nominal paradigms may be considered as further variations (see Kibrik and Kodzasov 1990).

Table 3: Case formation in East Caucasian: a fragment of the paradigm of luw ‘wing’ in Lezgian

nominative Luw luw-aár Wing wing-PL oblique stem luw-aá- luw-aá-i- wing-OBL wing-PL-OBL.PL ergative luw-aá luw-aár-i wing-ERG wing-PL-ERG genitive luw-aá-n luw-aár-i-n wing-OBL-GEN wing-PL-OBL.PL-GEN super luw-aá-l luw-aár-i-l wing-OBL-SUPER wing-PL-OBL.PL-SUPER kort_006.pod 137 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 137

In East Caucasian the category OBLIQUE is in a way ‘internal’ to the word form and not related to the syntactic function of the form: it is different from the oblique case in West Caucasian. In some East Caucasian languages, however, obliqueness does come into morphosyntax by triggering a special type of attributive agreement (see Section 6 below on noun phrase structure). Morphological expression of possessive relations. Outside West Caucasian, notoriously poor in case marking, most languages have a dedicated genitive case. In East Caucasian, Tsakhur and Rutul (Lezgic) have a general suffix to mark at- tributives: the same marker is used not only on possessors but also on adjectives, participles, attributivized adverbs etc. Possessor indexing exists in Turkic, Arme- nian and West Caucasian, but is absent from East and South Caucasian (but see (Lacroix 2009: 54) for a discussion of possessive clitics in Laz, also present in (11) below). In Ossetic, dative-marked possessors (as opposed to genitive possessors) are cross-referenced by a cliticized possessive pronoun: (6) Ossetic: possessive proclitic (Erschler 2009: 426) azɐmɐt-ɐn jɐ=qazˇɐn-t-ɐj Azamat-DAT 3SG.POSS=toy-PL-ABL ‘with Azamat’s toys’ Among all Caucasian languages, the category of (in)alienability in adnominal possession is present in Adyghe, the Besleney dialect of Kabardian, Budukh and Khinalug. For West Caucasian alienable possessees, the personal possessive pre- fix is followed by a special marker of possession, while for inalienable ones it immediately precedes the nominal stem. Remarkably, in Budukh and Khinalug the contrast between alienable and inalienable possession is not conveyed by marking on the possessee but by marking on the possessor, which is typologically very rare. Number. South Caucasian and Armenian as well as Turkic have one marker for nominal plural number; the use of additional markers is exceptional. West Cau- casian languages are essentially similar, although e.g. Abkhaz distinguishes between human and non-human plurals. Many East Caucasian languages, on the other hand, have rich inventories of plural markers, some of which are limited to human nouns, while others follow morphophonological distributions. The Cauca- sus lacks additional number categories such as dual (except some north-eastern dialects of Georgian, see Boeder 1998, Tuite 1998) or paucal. On the other hand, associative plurals ‘X and associates’ are quite widespread, and may be expressed either by regular plural markers (when attached to first names or kin terms) or by dedicated suffixes. An interesting fact about Armenian and Georgian number morphology is that both have ‘old’ plurals preserved from the classical period. Arme- nian uses its old plural in personal pronouns, Georgian uses it in some specific constructions and for stylistic purposes, and both languages use the old plural in demonstrative pronouns and in associative plural formation. kort_006.pod 138 07-12-08 13:18:12 -mt- mt

138 Michael Daniel and Yury Lander

A peculiar use of plural suffixes is attested in Armenian. As in many other languages, Armenian postpositions may take possessive suffixes instead of a dependent nominal. But Armenian has first and second person possessive suffixes only in the singular, and to express meanings such as ‘on us’ the singular possessive suffix is combined with the nominal plural marker (e.g. vra-ner-s literally “on-PL-MY”). This is distantly parallel to the usage some Turkic languages make of the plural marker on third person possessed nouns (where noun-PL-POSS3 is ambiguous between plural possessum and plural possessor or both). In all Caucasian languages, the use of numerals (and sometimes other quanti- fiers as well) requires the singular form of the noun. However, Armenian, just like Turkic languages, also allows plurality under certain semantic or pragmatic condi- tions, usually associated with specificity and related factors. The Digor dialect of Ossetic has a special form of nouns in oblique noun phrases containing numerals: orɒt-i (town-LOC) ‘in the town’, but cuppar gorɒt-e-mi (four city-COUNT-LOC) ‘in the four towns’, cuppar gorɒt-e-j (four city- COUNT-GEN) ‘of the four towns’, while in nominative noun phrases the genitive of the noun is used, as in cuppar gorɒt-i (four city-GEN) ‘the four towns’; (Isaev 1966: 51). Agreement. Caucasian Indo-European, Turkic, West and South Caucasian all have personal indexing on the verb (although in some West Caucasian languages these indexes are functionally closer to arguments than to genuine agreement) but no nominal classes. Abkhaz and Abaza are (possible) exceptions, because they have personal indexes distinguishing gender and animacy. East Caucasian languages are different. Not many of them possess personal marking on the verb (Lak, Dargwa, Udi, Tabasaran, Bats) and, when they do, this seems to be an innovation; other languages have various indirect ways of express- ing agreement-like categories on the verb (e.g. Archi, Akhvakh). Instead, their whole structure is permeated with nominal classes. With the exception of a few Lezgic languages (Agul, Lezgian, Udi) that lost their class agreement, East Cau- casian have three to five classes in the singular, including classes for male and fe- male on the one hand, and non-human classes with some semantic clusters but without consistent motivation on the other. The common pattern in the plural is to distinguish between humans and non-humans only. The most complex class systems are attested in Nakh languages, with sophisticated mapping between singular and plural classes. The striking feature of class agreement in East Caucasian is its pervasiveness. In some languages (e.g., Andic languages), certain types of possessors agree in class with their heads. Under clausal agreement, the nominative (S/P) may control agreement on verbs, adverbs, some case forms of nouns and pronouns, particles, as in (7). In several Andic languages, a converb may agree with an NP which is not an argument of the converb but an argument of the main predicate (8). Finally, an important syntactic phenomenon in East Caucasian is so-called “long-distance agree- kort_006.pod 139 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 139

ment”, where the controller and the target belong to different clauses (Polinsky and Potsdam 2001); cf. (9). (7) Archi: pervasive agreement (Kibrik et al. 1977b: 84) zari o‹r›ka-qi ja-r, bo-li, d-ez d-oq’i I.ERG ‹2›take.away.PFV-POT this-2(NOM) say.PFV-EVID 2-I.DAT 2-give.IMP ‘I will marry her, he said, give (her) to me’. (8) Akhvakh: external agreement of the converb (Creissels, in press) mol¯a rasadi w-u’-˚ı sˇwela-¯a m-a¯ne Molla Rasadi M-die-HPL[CVB] graveyard-IN.LAT HPL-go.PROG b-ak’-˚ı goli. HPL-be-HPL(CVB) AUX.HPL ‘Molla Rasadi died, and they were going to the graveyard’. (9) Ingush: long-distance agreement (Nichols 2008: 61) ma¯sa cˇo b-a¯q-a¯ b-ieza a¯z how.many hair(6) 6-cut-INF 6-should I.ERG ‘How many hairs should I cut?’ Note also that agreement patterns are formally quite complex and may involve suf- fixing, characteristic of attributive agreement, and prefixing and infixing, typical of verbal agreement. One word form may have several agreement slots. At the same time, in Khinalug all agreement markers except feminine, while present, are real- ized as zeros in most morphophonological contexts. Tense-Aspect-Mood. Aspectual morphology is rather diverse. East Caucasian often has verbal paradigms based on several primary stems; the two most important stems are perfective and imperfective, and the imperfective stem is sometimes derived from the perfective one. Specific tense, aspect and modal meanings choose one of the two stems, according to their functional semantic preferences. This two- stem architecture of the verbal paradigm is absent from e.g. Avar and is completely unknown in other endemic families (there are, however, quite a few verbs that distinguish perfective and imperfective stem in Armenian, cf. Plungian 2006). South Caucasian and Ossetic use verbal prefixes for perfectivization, in a way partly similar to Slavic. Armenian uses analytical constructions with ‘be’ and perfective vs. imperfective converbs; the verbal paradigm is largely analytic, except for aorist, imperative, subjunctive and positive conditional. In the western Caucasus, Abkhaz and Abaza have rich but on the whole common inventories of tense, aspect and mood markers bound to specific slots. In Adyghe and Kabardian, however, these markers may combine in sequences of up to three or more markers to form complex TAM configurations: kort_006.pod 140 07-12-08 13:18:12 -mt- mt

140 Michael Daniel and Yury Lander

(10) Adyghe: stacking of tense-aspect markers wəne-m he-r tjə-ha-ʁʁʁe-ʁe-n faje house-OBL head-NOM loc-put-PST-PST-POT must ‘Presumably, the house had been covered by a roof’. (e.g. there are traces of a roof on a now-roofless house)

An important distinction found in various Caucasian languages, partly correlated with aspect, is the opposition between stative and dynamic verbs, attested in e.g. West Caucasian and Lezgic; experiential statives also play an important role in the morphosyntactic classification of the predicates in South Caucasian. Thus Lezgic statives (‘want’, ‘know’, ‘be ill’) are deficient compared to regular verbs; they may lack the two-stem distinction, infinitives etc. The Caucasus is fairly rich in volitional moods. Some East Caucasian languages have mild imperatives as a morphological category and Nakh also has fu- ture imperatives (Johanna Nichols, p.c.). Imperatives may show special morphological behavior: in Archi, imperatives of intransitive verbs optionally express the number of addressees; in Akhvakh, the optative shows a special pattern involving unusual agreement (Denis Creissels, p.c.); and in some Andic languages there is a morphological difference between imperatives of transitive and intransitive verbs, not expressed anywhere elsewhere in morphology (Kibrik 2003a). East Caucasian has dedicated prohibitive (negative imperative) morphology. West Caucasian and Turkic also express negative imperative morphologically, but they use a marker of negation that is used in other morphosyntactic contexts, while Armenian uses a dedicated marker, but it is a particle; both patterns are cross-linguistically much more widespread than the one attested in East Caucasian. Several languages use dedicated marking for apprehensives (as in ‘do not play with the dog lest it bites you’). West and East Caucasian are known to have morphological performative optatives (forms used in curses and blessings), and at least Abkhaz and Nakh additionally introduce desiderative optatives (forms used to express speaker’s wishes and dreams). Both categories are cross-linguistically rare – and yet both are also attested in Caucasian Turkic languages Karachay-Balkar and Kumyk. Notably, optatives are on the whole atypical of Turkic languages – the category traditionally labeled optative in Turkic linguistics is closer to the imperative. For a detailed discussion of optatives in the Caucasus see (Dobrushina 2009). Inferential evidentials (‘apparently, a situation P took place’) are common in the Caucasus (cf. Chirikba 2008) but are rarely if ever conveyed by dedicated markers. The inferential meaning of a form is determined by the context. It may be semantic context, as with Armenian or Georgian or some East Caucasian perfects, or even a special syntactic pattern, as the evidential reading of the past converb when used as independent predicate in Archi or some Tsezic languages. Hearsay evidentials (‘they say that P’) often grammaticalize from speech verbs contracted into clitics in East Caucasian; other evidentials seem to be peripheral and weakly kort_006.pod 141 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 141

grammaticalized or not present at all. For details, see Johanson and Utas (2000), Tatevosov (2001), Chirikba (2003). Valency increase and decrease. East Caucasian languages are poor in valency changing processes. The only valency-related process regularly present in the family is causative formation, which formally varies from morphological marking in e.g. Tsezic and Andic to transparent ‘do’ (or, rarely, ‘give’) periphrasis in some Lezgic languages and show many cases that are intermediate between biclausal and monoclausal constructions. South Caucasian, West Caucasian and Armenian all have robust morphological causatives. Valency decrease is almost absent from East and West Caucasian, but prospers in Armenian which has a mediopassive (middle) suffix covering passive, reflexive, reciprocal and other functions, similar to reflexive verbs in Slavic. South Caucasian, on the other hand, has a most sophisticated system of valency derivation – the so-called versions, expressed by vowels immediately preceding the verbal root. With a typical Georgian version, a secondary semantic role – experiencer, recipient and beneficiary in (11), external possessor or a spatial function in (15) below – receives indirect object marking and becomes eligible to control the agreement on the verb (which is otherwise impossible). In a way, versions are promoting non-core arguments into core argument position and are thus typologically close to applicatives (Lacroix 2009: 484ff.); Lacroix further argues that pro- moted arguments are ambiguous between indirect (dative marking) and direct (agreement control) objects. (11) Laz “objective version” – recipient promotion (Lacroix 2009: 496) ajdaa m-i-il-a-t-na bozo-cˇkimi me-k-cˇ-ae-a dragon 1-APPL-kill-OPT-1/2.PL-COND girl-1PL prvb-2-give-1/2.SG-QUOT ‘If you kill a dragon for our sake, I will give you my daughter’. A version marker may also convey various mediopassive meanings or, more generally, indicate that the subject will pop up elsewhere in the predication – in the following example, the agent is also the possessor of the waistcoat: (12) Laz: “subjective version” (Lacroix 2009: 496) t’abak’a do mendil jeleʁi-sˇ ceb-epe-s snuff.box and handkerchief waistcoat-GEN pocket-PL-DAT dol-i-du-mer-nan prvb-med-put-TH-PRS3PL ‘They are putting their snuff-box and their handkerchief in their (own) pockets’. (lit. ‘they are putting snuff-box and handkerchief into waistcoat pockets to themselves’) Verbs often carry a lexical preroot vowel, which is substituted by other vowel markers in derived versions. All this makes it difficult to account for South Cau- casian versions in conventional terms of valency derivation. kort_006.pod 142 07-12-08 13:18:12 -mt- mt

142 Michael Daniel and Yury Lander

West Caucasian languages are also described as languages rich in applicatives, as they readily introduce benefactive, malefactive, comitative, locative and other slots into the verbal complex: cf. (2). However, in polysynthetic languages like West Caucasian, these slots are not – or not straightforwardly – agreement slots (see Section 2 for discussion of West Caucasian morphology). Thus the same slot may occur several times in the word form: (13) Adyghe: two benefactive slots in one word form (Lander and Letuchiy 2010) s-a-fə-Ø-f-e-txe 1SG.S/P-3PL.IO-BEN-3SG.IO-BEN-DYN-write ‘I write to them for his benefit’. To argue that these two benefactive slots are agreement slots we need to posit two different benefactive agreement positions. More natural seems a parallel with two dative NPs which, in many languages, may be present in the same clause to convey similarly different meanings. In other words, at the morphological level the two benefactives behave more like adjuncts at the level of syntax: they are optional and may be repeated. The issue of choosing between template and polysynthetic analysis also applies to valency decrease patterns, e.g., reflexive forms, which may be analyzed as a verbal complex containing a pronominal-like reflexive element rather than verbal forms containing a morphological marker of reflexivization. Pronouns. In East, West and South Caucasian, all languages with a dominant ergative construction, personal pronouns often do not distinguish ergative and nominative forms. This does not necessarily involve all persons and numbers, but in e.g. Georgian first and second person pronouns are indeclinable, and when case marking is required, possessive pronouns are added to the declined noun ‘head’: (14) Georgian: substitute declension of personal pronouns (Boeder 2005: 54) sabralo cˇven-o tav-o, sac’q’al-o sakartvelo-v pitiful our-VOC head-VOC, pitiable-VOC Georgia-VOC ‘O pitiful us, poor Georgia!’ Inclusives are attested in various East Caucasian languages (although not present in all of them – they were lost in e.g. Tsezic, Lak and some Lezgic and Dargwa languages). Archi has a peculiar inclusive pattern: this language probably first lost the distinction, integrating formerly inclusive and exclusive forms into one ‘neu- tral’ first person plural paradigm, but then reintroduced the distinction by adding a reflexive / emphatic particle to specify inclusive reference. The only other inclusive reported in the Caucasus is attested in Svan (personal agreement and possessive pronoun); inclusive in Abkhaz (Hewitt 1979: 157) is contested. Caucasian languages tend to lack third person pronouns: in Armenian, Ossetic and South Caucasian (and probably also in Nakh) third person pronouns are historically derived from – though not identical to – demonstratives. Many West and East Caucasian languages have demonstratives that function like personal pro- kort_006.pod 143 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 143

nouns. The system of deictic reference may be rather complex: from Armenian and Georgian and West Caucasian, which have a tripartite deixis (‘this one here’ ~ ‘that one there’ ~ ‘that one yonder’), to East Caucasian languages, which are obsessed with space and introduce further spatial distinctions such as (‘this below’ ~ ‘this above’) and other spatial nuances (see Schulze 2003). Most East Caucasian languages also have a special pronoun with a very wide range of usages covering logophoric, reflexive, resumptive, emphatic and other functions. With this variety of functions, this pronoun seems to be a special linguistic means of reinforced reference tracking, whose similarity to reflexive pronouns in European languages or logophoric pronouns in African languages is only partial. For a discussion see Tes- telets and Toldova (1998) and Lyutikova (2000); cf. also Donabedian (2007) for a similar phenomenon in Armenian.

6. Syntax: some highlights

In a brief survey of such a diverse linguistic area as the Caucasus, giving a full account of syntactic variation is even less feasible than a full account for morphology or phonology. This section is a balance between central issues, on the one hand, and salient features, on the other, including noun phrase structure, alignment, finiteness, focus marking, relativization, clause combining and reported speech constructions. Noun phrase syntax. Although word order in Caucasian languages is quite flexible, the basic word order is invariably SOV (cf. e.g. Testelec 1998a and 1998b for East and South Caucasian). Consistently with SOV typology, NPs are left- branching, except for marked patterns, e.g. constructions involving heavy depend- ents. Typical for SOV is also the use of postpositions rather than prepositions: Ar- menian has few prepositions (preserved from Classical Armenian), and so does Ossetic, but West and East Caucasian do not have any. In noun phrases, West Cau- casian and Turkic languages use head marking, while East and South Caucasian lack possessive pronouns and use genitive marking on the dependent (Armenian uses genitive pronoun or possessive suffix for the first and second person singular possessor and only genitive marking in other cases). Within the noun phrase, agreement between the head and the dependent varies from poor to none. In South Caucasian, there is no NP-internal agreement except partial agreement in case on adjectives. East Caucasian languages typically have class agreement, which also includes number (class / number agreement is syn- cretic), although some have special plural agreement suffixes for adjectives (e.g. Dargwa). Nominal possessors may agree with their heads in class e.g. in Andic languages. Less common patterns include cases of double case marking in South Cau- casian (15), where a genitive dependent in a non-canonical position copies the case suffix of the head noun (Suffixaufnahme), and agreement in obliqueness in Tsezic kort_006.pod 144 07-12-08 13:18:12 -mt- mt

144 Michael Daniel and Yury Lander

languages, where there are two genitive markers, one used when the head noun is in the nominative and the other when it is in any of the oblique cases. Similar agreement in obliqueness is also attested in e.g. Khinalug and Nakh, but here it applies only to demonstratives or adjectives. (15) Mingrelian: Suffixaufnahme (Boeder 1995: 194) gi-Ø-a-ntχu¯-d-esəək’ata-sə te χenc’əpe-sˇi PRVB-3O-LOC-fall-IPF-3S.PL.PST people-DAT this king-GEN γəmala-sˇi-sə dominion-GEN-DAT ‘They attacked the people of this king’s dominion’. Another rare pattern is incorporation in West Caucasian NPs: (16) Adyghe: noun – adjective complex (Arkadiev et al. 2009: 50) c’əf-gwəsˇxwe-m man-brave-OBL ‘a brave man’ (oblique form) When used headlessly, all attributes take nominal inflections, which directly follow the nominal stem. With genitive attributes, this leads to having two case markers in one word form (cf. ex. 14) and, in some languages, to clearly group- marking patterns, where the nominalized attribute attaches its own dependent ‘prior’ to nominalization; (in (17) the brackets show NP boundaries): (17) Bagvalal: group marking (Kibrik et al. 2001: 212) a-b hob [in-¯i-¯-da wasˇa]-sˇ¯u-b-¯i-ba this-N tomb(NOM) self-OBL-GEN-PTCL son-OBL.M-N.GEN-OBL-AFF b-a¯i-li-b-o, o-w-la raq’wa-la¯ w-a¯-w-o N-similar-VBLZ-N-CVB that-1(NOM)-and heart-SUP.LAT 1-come-1-CVB ‘As this tomb was similar to the one of my own son, I remembered him …’ This type of interaction between syntax and morphology is very different from the competition described above in Section 3 for the West Caucasian; rather, it is a mismatch between phrasal constituents and morphological word boundaries. Of all Caucasian languages only Armenian and Abkhaz possess grammaticalized definite articles (the Old Georgian definite article was lost in the modern language). The East Armenian definite article, which most probably comes from the third person possessive marker, is used in a wider range of contexts than European articles – it is obligatory on inherently specific nouns such as proper names. There is also a peculiar interaction between the definite article and the category of case: it cannot be used on non-core cases or the genitive, but is almost obligatory on the dative. It is thus only in the nominative that the marker has the value of a genuine definite article. Some East Caucasian languages use demonstratives more or less article-wise, but these means are much less obligatory and less bound to the head kort_006.pod 145 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 145

noun than in Armenian or Abkhaz. In many languages the numeral ‘one’ is used as an introductory indefinite article, but it does not seem to grammaticalize to the extent of European indefinite articles to include other domains of non-specificity. Ergativity. Notably, all three endemic families are consistently ergative (though see below on South Caucasian), but Indo-European and Turkic languages spoken in the Caucasus show no traces of ergative alignment at all. In East Cau- casian, case marking follows the S/P vs. A distinction (but see Section 5 on neu- tralizing the distinction in pronouns and on differential object marking in Udi), and all agreement slots are controlled by nominative arguments. West Caucasian languages also use cases – when they have any – distinguishing between direct for S/P and oblique for all other functions including A; but in all West Caucasian languages there are different slots for indexing S/P and A. Now, the real challenge of ergative morphosyntax comes from sophisticated systems of case marking in South Caucasian, where the same marker is used to indicate different roles depending on the morphosemantic category of the predicate and its tense. Cf. Table 3, which represents the distribution in a rather simplified way (without taking into account classes of predicates):

Table 4: Tense and case function in Georgian

Present series Aorist series Perfect series

P-s DAT -i / Ø NOM S-i / Ø NOM A-ma / -m / -man ERG -s DAT

Other South Caucasian languages show variations on this theme. In some Laz var- ieties, the ergative extended to A in the present, and in some others case marking of the core arguments was lost completely and the dative is used exclusively as a locative marker. The Mingrelian “ergative” has extended to S in the aorist. This change affected the whole system: the aorist switched from the ergative to accusative construction, and -i (Ø) assumed the accusative function, opposite to what it conveys in the present. For a discussion see, for example, Harris (1985). Syntactic ergativity and the distribution of subject properties between S and O is another rather complicated issue and subject to variation. In Modern Georgian, conventional tests such as coordinate deletion and reflexivization demonstrate accusative syntactic behavior. In West Caucasian, too, reflexivization most often follows the accusative pattern; but other tests are more problematic. And especially problematic is the notion of subject in East Caucasian. Some studies suggest the complete absence of a syntactic subject or, if there is one, the presence of a weak subject (Kibrik 1997). Compare the following reflexive constructions in Lak, where, according to Kibrik (2003a), possessive reflexivization is controlled by the linear order rather than the role of the NP: kort_006.pod 146 07-12-08 13:18:12 -mt- mt

146 Michael Daniel and Yury Lander

(18) Lak: role-independent possessive reflexivization (Kibrik 2003a: 477) a. gwani-l ci-l-b-asu ¯ b-awt¯un-di she-ERG self.F-GEN-3-PTCL sister(NOM) 3-beat-3SG ‘She beat up her own sister’. b.su ¯ ci-l-Ø-a but¯a-l b-awt¯un-di sister(NOM) self.F-GEN-1-PTCL father-ERG 3-beat-3SG ‘The sister was beaten by her father’. (lit. ‘her own = sister’s own father beat the sister’) The diversity of accounts of syntactic alignment comes from the fact that the situation varies from language to language, from test to test (Kibrik 2003a) and pos- sibly from speaker to speaker. Finiteness. In South Caucasian finite forms bear personal agreement markers, while e.g. participles and action nominals do not; this is reminiscent of European patterns. In West Caucasian, finiteness is especially salient in Abkhaz and Abaza, where verbal forms clearly split into a finite and a non-finite sub-paradigm, while in Kabar- dian and especially in Adyghe, the distinction is less pronounced. Formal differences are less obvious in East Caucasian: participles, action nominals, infinitives and converbs include the same class markers as finite forms do, although participles can also take class suffixes (typical of adjectives), and action nominals may attach case markers. In some East Caucasian languages participles (sometimes also other nonfi- nite forms) may in some contexts appear as predicates in independent clauses (see e.g. Kalinina and Sumbatova 2007). In Khinalug, the two main finite forms are most often formally identical to participles, which may be seen as the endpoint of the participles’ expansion into the finite domain. In some languages, general converbs can be used finitely (e.g. for unwitnessed past in Tsezic and Archi). This use is likely to originate from the auxiliary construction by omitting the finite auxiliary predicate. Focus constructions. By default, the focused element is placed in the preverbal position. South Caucasian also makes a wide use of clefts (Boeder 2005). But many other languages of the Caucasus feature peculiar types of focus marking (Testelec 1998c). In some languages (Lak, Udi, Dargwa), the focused constituent may be followed by a personal agreement clitic (cf. Kazenin 2002, Harris 2002). An important feature of these constructions is that the finite predicate in such clauses changes to a participle (Kalinina and Sumbatova 2007). In some other East Caucasian languages, e.g. Bagvalal and Agul, and in Armenian (see Comrie 1984), the focus is marked not by a personal marker but by the position of the copula, which must immediately follow the focused item. (19) Icari Dargwa: focus marking by personal clitic (Sumbatova and Mutalov 2003: 141) ih ʁibij c’il-di-q’al w-aʕr-un-ci u that Gibi later-2-but M-go.PFV-PRET-PTCP you.sg(NOM) ‘But it was later on that you went to that Gibi!’ kort_006.pod 147 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 147

Where East Caucasian employs special material, West Caucasian may rely on syntax only. To focus an argument, a syntactic reorganization is required, which transforms the original structure into a clause structurally similar to equative predi- cations. (20) Focus in Adyghe a. cˇəγə-r wəne-m ə-pahe jə-t tree-NOM house-OBL 3sg-front LOC-stand ‘The tree is in front of the house’. b. cˇəγə-r zə-pahe jə-tə-r wəne-r a-rə tree-NOM rel-front LOC-stand-NOM house-NOM that-PRED ‘It is in front of the house that the tree is’. (lit. ‘in front of what the tree stands is the house’) Relativization. Georgian and Armenian have European-style relative clauses headed by interrogative-based relative pronouns; while Ossetic follows cor- relative, and Tat participial pattern of relativization. In Laz relative clauses, the predicate immediately precedes the head noun, as in the participial strategy common in the Caucasus, but is finite, which makes it similar to European relative clauses. However, there is no relative pronoun, only a subordinating clitic hosted by one of the constituents inside the relative clause. Most East Caucasian languages lack dedicated relative pronouns (Udi and Bats are rare exceptions, probably contact-induced) and use a participial strategy. The most salient feature of relative clauses in East Caucasian is the extreme flexibility of relativization, which may be based on virtually any argument and adjunct as well as their possessors (see also Comrie and Polinsky 1999) and, in some rare cases, may not even be traceable to any syntactic position at all. In the following example, the noun ‘meat’ heads a relative clause it never belonged to, denoting an item which is possessed by a participant in the relative clause (‘cattle’): (21) Archi: ‘untraceable’ relativization (Kibrik et al. 1977b: 107)    [¯hawan bu- ’u-t¯u-t a ’]NP o-t’u cattle(NOM,3) 3-slaughter.PFV-ATR-4 meat(NOM,4) 4.give.PFV-NEG ‘They wouldn’t give us the meat of the slaughtered sheep’. Unlike the relative pronoun strategy, the participial strategy does not indicate the role of the target of relativization within the relative clause. Some East Caucasian languages may use reflexive pronouns in resumptive function. These are mostly employed when the role of the target cannot be easily established: (22) Lezgian: reflexive pronoun used as a resumptive (Haspelmath 1993: 342) cˇun wicˇi-kaj raxa-zwa-j kas we(NOM) self-SUB.EL talk-IPFV-PTCP man ‘the man we’re talking about’ kort_006.pod 148 07-12-08 13:18:12 -mt- mt

148 Michael Daniel and Yury Lander

West Caucasian languages can be described as having relative pronouns, but these are represented by prefixes. They also feature what is called internally-headed relative clauses (23), a phenomenon widespread in the world but quite rare in Eu- rope. Typologically unique is multiple relativization, with several coreferent relative prefixes within the same clause (see Lander 2009 on Adyghe); cf. (24). (23) Abkhaz: internally-headed relative clause (Andrej Kibrik 1992: 148) [s-ab a-ʕwnə jə-jə-rgəla-z] 1SG.PR-father DEF-house REL.S/P-3SG.M.A-build-NFIN.PST Ø-aajgwara s-gəlo-w-p’ 3SG.N.OBL-near 1SG.S/P-stand-PRES-FIN ‘I stand near the house built by my father’. (24) Abkhaz: multiple relativization (Hewitt 1979: 37) z-an jə-lə-cə-m-nəq’wa(-z) a-cˇ’k’wən REL.PR-mother REL.S/P-3SG.F.IO-COM-NEG-go-NFIN.PST def-boy ‘the boy who did not go for a walk with his (lit. whose) mother’ Clause combining. In European languages complex clauses are either coordina- teot subordinate. South Caucasian and Armenian are roughly compliant with this model. In East and West Caucasian, however, the situation is different. Conjunc- tions, both coordinating and subordinating, are few. For mere sequencing of events, chains of general converbs are built, which combine properties of subordination with properties of coordination (Kazenin and Testelets 2004). (25) Akhvakh: general converb in a coordinative construction (Denis Creissels, p.c.) molla-sw-e taχi-gunu ˜ıgora b-eχ-e uhora-ge˜ ge-i Molla-OBL.M-ERG pocket-EL bread N-take-CVB.N lake-ESS inside-ESS tuk-ini dip-EVID ‘Molla took some bread from his pocket and dipped it into the lake’. To specify the adverbial relation linking the dependent predicate to the main clause (‘because of the main event’, ‘despite the main event’, ‘simultaneously with the main event’, etc.) special converbs are used. The inventory of these special converbs may count up to some dozen forms; sometimes, they contain nominal case markers also used in adverbs. Other dependent forms corresponding to European complementation include case forms of action nominals (e.g. for purposive and causal relations), infinitives (used under e.g. phasal and modal matrix verbs), sometimes also participles and general converbs. Semidirect speech. In general, reported speech in East Caucasian is very different from what we find in (the prescriptive grammars of) European languages. In Standard Average European, reporting is either consistently direct (intended to re- kort_006.pod 149 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 149

produce the actual speech act) or consistently indirect (using special marking of reporting, such as the use of reflexive pronouns in logophoric function, shifting the personal, temporal and spatial deictic perspective to the actual speaker’s viewpoint). In East Caucasian, the same clause may combine direct and indirect elements. In the following Archi example, reported speech combines the use of indirect personal deixis (agreement) with direct illocution (prohibitive) within the same word. Direct reporting meets indirect reporting within the same word form: (26) Archi: ‘semidirect word form’ (from the texts collected in Dirr 1908) os iq-n-a jemib ummu one day-OBL-IN this.PL father.OBL(ERG) b-‹r›qʕ-ir-gi bo-t¯u-t raq-m-a-sˇi o‹b›qʕa-li. HPL-‹IPFV›go-IPFV-PROH say-ATR-4 side-OBL-IN-ALL ‹HPL›go.PFV-EVID ‘One day they went in the direction that the father said not to follow’. (lit. ‘to the father “do not go”-said side’) The actual prohibitive form being reported is orqʕirgi ‘you (plural) do not go!’, with a special 1st/2nd person plural agreement. When reported, the agreement ʕ switches to regular HPL borq irgi ‘they do not go’, because the actual addressee (the listener) is no longer identical with the addressee of the form. However, the narrator uses the form of the prohibitive mood, which may only be addressed to the second person. Patterns of semidirect speech reporting are an important and understudied phenomenon in East Caucasian. In some languages of the family, however, it may play a less important role, and seems to be closer to European in West and South Cau- casian (but see Hewitt and Crisp 1986) or Armenian.

7. Perspectives

While the tradition of grammatical description of literary Armenian and Georgian goes back to the early Middle Ages, the systematic linguistic study of East and West Caucasian languages started only in late 19th (Petr Uslar) and early 20th century (Adolf Dirr) and has been continued mainly by the Soviet scholars from Makhachkala, Tbilisi, Nalchik, Maikop and Baku. At that time, not many linguists from outside the Caucasus were interested in the area; an important player was the Georgian school. From the late 60s on, the Moscow school with its team fieldwork method started to play an important role in language documentation, producing a series of detailed up-to-date morphological and syntactic descriptions of Cau- casian (primarily East Caucasian) languages. Since the 70s and especially during the 90s, the interest in minority languages of the Caucasus has grown significantly in the West, and triggered what might be called a frontal attack on linguistic mi- norities of the Caucasus has started. Today, Caucasian studies form an impressive kort_006.pod 150 07-12-08 13:18:12 -mt- mt

150 Michael Daniel and Yury Lander

network of groups and individuals in the Caucasus, Moscow, Europe and the USA. New approaches with their shift from grammatical description to language documentation are coming to the Caucasus. Linguistic corpora are being compiled, both for major languages such as Eastern Armenian (www.eanc.net) and smaller languages including Archi, Khinalug, Agul, Udi, Dargwa, Bezhta, Ossetic. These corpora sometimes include digital archives of audio and even video recordings which are accessible via internet (cf. www.ossetic-studies.org and www.philol.msu.ru/ ~languedoc/eng). By now, no language in the Caucasus is left undescribed. This, and the absence of an immediate threat of language death, makes this area look quite well-off in terms of description, as compared to Central Africa or Papua, where many languages lack even primary accounts, or North America and Siberia, where languages are on the verge of extinction. However, descriptions of Caucasian languages, especially those produced in the early to mid-20th century, need revision from the perspective of the typological knowledge accumulated over the last few decades. The problem is especially acute in syntax, where older descriptions often followed the model of Russian grammar, which is drastically different from the syntactic patterns attested in the Caucasus. Syntactic research is thus a typological priority for Caucasian languages. Another important direction of research is the dialectal variation of major languages. While modern research has focused on minor and previously undescribed languages, the dialect documentation of Armenian, Azerbaijani, Georgian has moved off-focus (partly for economic reasons) before the new typology-oriented approaches to language description have emerged. The documentation of Avar or Dargwa dialects never did well because of the sociopolitical pressure of written standards. Corpus building is a necessary prerequisite for both dialect documentation (because it shows differences from the literary language in a natural, spon- taneous environment) and syntactic research in minority languages. Even more than descriptive and comparative studies, the Caucasian languages lack approaches that traditionally develop after the first descriptive effort has been completed. Thus the field of psycholinguistics (color terms, language acquisition, etc.) is almost vacant. Given the multilingualism of some areas, the Caucasus is looking forward towards fundamental sociolinguistic studies. The issue of language contact in the Caucasus clearly deserves more attention, even more so because Caucasian languages constitute one of the first Sprachbunds posited in the history of linguistics (Nikolai Trubetzkoy’s hypothesis, see Chirikba (2008) for a recent discussion); there are several non-endemic languages like Eastern Arme- nian, Tat and Ossetic (Indo-European), Azerbaijani and Karachay-Balkar (Turkic) that have been in contact with endemic languages for many centuries. A short overview chapter cannot represent the modern bibliography of Cau- casian linguistics – thousands of volumes and articles – in anything but a partial way. Here we provide the reader with references only to those general surveys kort_006.pod 151 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 151

where further indications to books and papers on specific languages, language families and linguistic topics related to the Caucasus can be found. Recent publi- cations include first of all Hewitt (2004) and van den Berg (2005); earlier introduc- tions to the Caucasian field can be represented by Klimov ([1986] 1994). For an overview of East Caucasian grammars see Schulze (2005). Detailed reference descriptions of nonendemic languages in English include Abaev (1964) for Ossetic and Dum Tragut (2009) for Eastern Armenian. Linguistic diversity in the Caucasus is discussed in Catford (1977) and later in Comrie (2008); main sources on the Caucasian ethnolinguistic geography are Koryakov (2002) and Tsutsiev (2006). Bokarev and Lomtatidze (1965), Greppin (1991–1994); and Alekseev et al. (1999) are collections of sketches of Caucasian languages.

Acknowledgments

We would like to express our deepest gratitude to Timofey Arkhangelskiy, Gilles Authier, Oleg Belyaev, Denis Creissels, David Erschler, Victoria Khurshudian, Timur Maisak, Arseny Vydrin, Alexander Rostovtsev-Popiel, Yakov Testelets and Ilya Yakubovich for their comments, suggestions and above all corrections, as well as for readiness to share their knowledge and linguistic examples. We are grateful to the editors of this volume for their helpfulness and patience. Obviously, it is the authors of the present paper who should be blamed for all mistakes and misinter- pretations. This work was supported in part by the RFFI grant No. 05-06-80351, RGNF grants No. 06-04-00194a, 07-04-00266à, 09-04-00168a, 10-04-00228a and the NSF grant No. 0553546.

References

Abaev, Vasilij I. 1964 A Grammatical Sketch of Ossetic. (Publication of the Indiana University Re- search Center in Anthropology, Folklore, and Linguistics 35.) Blomington: In- diana University, The Hague: Mouton. The English translation of: Kratkij grammaticˇeskij ocˇerk osetinskogo jazyka. In: B.B. Bigulaev and A.M. Kasaev, Osetinsko-russkij slovar’, 443–540. Moscow: GINS [1952]. Alekseev, Mikhail E., Georgij A. Klimov, Sergej A. Starostin and Yakov G. Testelets (eds.) 1999 Jazyki mira: Kavkazskie jazyki. [Languages of the World: Caucasian Lan- guages.] Moscow: Academia. Arkadiev, Peter M., Yury A. Lander, Alexander B. Letuchiy, Nina R. Sumbatova, Yakov G. Testelets 2009 Vvedenie. Osnovnye svedenija ob adygejskom jazyke. [Introduction: Basic in- formation on Adyghe.] In: Yakov G. Testelets et al. (eds.), Aspekty polisinte- tizma: Ocˇerki po grammatike adygejskogo jazyka, 17–120. Moscow: RGGU. kort_006.pod 152 07-12-08 13:18:12 -mt- mt

152 Michael Daniel and Yury Lander

Boeder, Winfried 1995 Suffixaufnahme in Kartvelian. In: Frans Plank (ed.), Double Case: Agreement by Suffixaufnahme, 151–215. Oxford: Oxford University Press. Boeder, Winfried 1998 Syntax und Semantik von Einheit und Vielfalt beim georgischen Nomen. In: Winfried Boeder, Christoph Schroeder, Karl Heinz Wagner and Wolfgang Wildgen (eds.), Sprache in Raum und Zeit: In memoriam Johannes Bechert, Band 2, 251–268. Tübingen: Gunter Narr Verlag. Boeder, Winfried 2005 The South Caucasian languages. In: Helma van den Berg (ed.), Caucasian, Lingua 115: 5–89. Bokarev, Evgenij A. and Ketevan V. Lomtatidze (eds.-in-chief) 1967 Jazyki narodov SSSR. Volume 4: Iberijsko-kavkazskie jazyki. [Languages of the Peoples of the USSR 4: Ibero-Caucasian Languages.] Moscow: Nauka. Catford, J. C. 1977 Mountain of tongues: the languages of the Caucasus. Annual Review of An- thropology 6: 283–314. Chirikba, Viacheslav 2003 Evidential category and evidential strategy in Abkhaz. In: Alexandra Y. Aik- henvald and R. M. W. Dixon (eds.), Studies in Evidentiality, 243–272. (Typological Studies in Language 54.) Amsterdam/Philadelphia: John Benja- mins Chirikba, Viacheslav 2008 The problem of the Caucasian Sprachbund. In: Pieter Muysken (ed.), From Linguistic Areas to Areal Linguistics, 25–93. (Studies in Language Companion Series 90.) Amsterdam/Philadelphia: John Benjamins. Comrie, Bernard 1984 Some formal properties of focus in Modem Eastern Armenian. Annual of Ar- menian Linguistics 5: 1–21. Comrie, Bernard 2008 Linguistic diversity in the Caucasus. Annual Review of Anthropology 37: 131–143. Comrie, Bernard and Maria Polinsky 1999 Form and function in syntax: relative clauses in Tsez. In: Michael Darnell, Edith A. Moravcsik, Frederick J. Newmeyer, Michael Noonan, and Kathleen M. Wheatley (eds.), Functionalism and Formalism in Linguistics, Volume II: Case Studies, 77–92. (Studies in Language Companion Series 42.) Amster- dam/Philadelphia: John Benjamins. Creissels, Denis In press External agreement in the converbal construction of Northern Akhvakh. In: Holger Diessel and Volker Gast (eds.), Clause Combining in Cross-linguistic Perspective. Berlin/New York: Mouton de Gruyter. Daniel, Michael and Dmitry Ganenkov 2009 Case marking in Daghestanian: limits of elaboration. In: Andrej Malchukov and Andrew Spencer (eds.), The Oxford Handbook of Case, 668–685. Oxford: Oxford University Press. kort_006.pod 153 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 153

Daniel, Michael, Zaira Khalilova, Zarina Molochieva 2010 Ditransitive constructions in East Caucasian: a family overview. In: Andrej L. Malchukov, Martin Haspelmath, Bernard Comrie (eds.), Studies in Ditransi- tive Constructions: A Comparative Handbook. Berlin/New York: Mouton de Gruyter. Dirr, Adolf M. 1908 Arcˇinskij jazyk. [The Archi Language.] (Sbornik materialov dlja opisanija mestnostej i pis’men Kavkaza 39.) Tiflis. Donabédian, Anaïd 2007 A la recherche de la logophoricité en arménien. In: M. M. Jocelyne Fernandez- Vest (ed.), Combat pour les langues du monde / Fighting for the World’s Lan- guages. Hommage à Claude Hagège, 165–176. (Grammaire & Cognition 4&5.) Paris: L’Harmattan. Dobrushina, Nina R. 2009 Semanticˇeskaja zona optativa v naxsko-dagestanskix jazykax. [The optative semantic domain in Nakh-Daghestanian languages.] Voprosy jazykoznanija 2009–5: 48–75. Dum-Tragut, Jasmine 2009 Armenian: Modern Eastern Armenian. (London Oriental and African Lan- guage Library 14.) Amsterdam/Philadelphia: John Benjamins. Erschler, David 2009 Possession marking in Ossetic: Arguing for Caucasian influences. Linguistic Typology 13: 417–450. Gippert, Jost, Wolfgang Schulze, Zaza Aleksidze, Jean-Pierre Mahé (eds.) 2009 The Caucasian Albanian Palimpsests of Mount Sinai. 2 volumes. (Monumenta Palaeographica Medii Aevi: Series Ibero-Caucasica 2.) Turnhout: Brepols. Greppin, John A.C. (ed.) 1991–1994 The Indigenous Languages of the Caucasus. 4 volumes. Delmar, N.Y.: Caravan Books. Harris, Alice C. 1985 Diachronic Syntax: The Kartvelian Case. (Syntax and Semantics 18.) New York: Academic Press. Harris, Alice C. 2002 Endoclitics and the Origins of Udi Morphosyntax. Oxford: Oxford University Press. Haspelmath, Martin 1993 A Grammar of Lezgian. (Mouton Grammar Library 9.) Berlin/New York: Mouton de Gruyter. Hewitt, B. George 1979 Abkhaz. (Lingua Descriptive Studies 2.) Amsterdam: North-Holland. Hewitt, B. George 2004 Introduction to the Study of the Languages of the Caucasus. (LINCOM Hand- book in Linguistics H19.) München: LINCOM Europa. Hewitt, B. George and Simon R. Crisp 1986 Speech reporting in the Caucasus. In: Florian Coulmas (ed.), Direct and Indi- rect Speech, 121–144. (Trends in Linguistics. Studies and Monographs 31.) Berlin/New York/Amsterdam: Mouton de Gruyter. kort_006.pod 154 07-12-08 13:18:12 -mt- mt

154 Michael Daniel and Yury Lander

Isaev, M.M. 1966 Digorskij dialect osetinskogo jazyka. Fonetika. Morfologija. [The Digor Dia- lect of Ossetic. Phonetics. Morphology.] Moscow: Nauka. Johanson, Lars and Bo Utas (eds.) 2000 Evidentials: Turkic, Iranian and Neighbouring Languages. (Empirical Ap- proaches to Language Typology 24.) Berlin/New York: Mouton de Gruyter. Kalinina, Elena and Nina Sumbatova 2007 Clause structure and verbal forms in Nakh-Daghestanian languages. In: Irina Nikolaeva (ed.), Finiteness. Theoretical and Empirical Foundations, 184–249. Oxford: Oxford University Press. Kazenin, Konstantin I. 2002 Focus in Daghestanian and word order typology. Linguistic Typology 6: 289–316. Kazenin, Konstantin I. and Yakov G. Testelets 2004 Where coordination meets subordination: Converb constructions in Tsakhur (Daghestanian). In: Martin Haspelmath (ed.), Coordinating Constructions, 227–239. (Typological Studies in Language 58.) Amsterdam/Philadelphia: John Benjamins. Kibrik, Aleksandr E. 1997 Beyond subject and object: Toward a comprehensive clause patterning typology. Linguistic Typology 1: 279–346. Kibrik, Aleksandr E. 2003a Konstanty i peremennye jazyka. [Constants and Variables of Language.] St. Petersburg: Aletheia. Kibrik, Aleksandr E. 2003b Nominal inflection galore: Daghestanian, with side glances at Europe and the world. In: Frans Plank (ed.), Noun Phrase Structure in the Languages of Europe, 37–112. (Empirical Approaches to Language Typology, EUROTYP 20–7.) Berlin/New York: Mouton de Gruyter. Kibrik, Aleksandr E. and Sandro V. Kodzasov 1990 Sopostavitel’noe izucˇenie dagestanskix jazykov. Imja. Fonetika. [Comparative Studies in Daghestanian Languages. Noun. Phonetics.] Moscow: Izdatel’stvo MGU. Kibrik, Aleksandr E., Sandro V. Kodzasov, Irina P. Olovjannikova, Dzhalil S. Samedov 1977a Opyt strukturnogo opisanija arcˇinskogo jazyka. [A Structural Description of Archi.] 3 volumes. Moscow: Izdatel’stvo MGU. Kibrik, Aleksandr E., Sandro V. Kodzasov, Irina P. Olovjannikova, Dzhalil S. Samedov 1977b Arcˇinskij jazyk. Teksty i slovari. [The Archi Language. Texts and Dictionaries.] Moscow: Izdatel’stvo MGU. Kibrik, Aleksandr E., Konstantin I. Kazenin, Ekaterina A. Lyutikova, Sergei G. Tatevosov (eds.) 2001 Bagvalinskij jazyk. Grammatika. Teksty. Slovari. [Bagvalal. Grammar. Texts. Dictionaries.] Moscow: Nasledie. Kibrik, Andrej A. 1992 Relativization in polysynthetic languages. International Journal of American Linguistics 58: 135–157. kort_006.pod 155 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 155

Klimov, Georgij A. 1994 Einführung in die kaukasische Sprachwissenschaft. Hamburg: Buske. The German translation of: Vvedenije v kavkazskoe jazykoznanie. Moscow: Nauka [1986]. Krauss, Michael 1992 The world’s languages in crisis. Language 68:4–10. Koryakov, Yury B. 2002 Atlas of the Caucasian Languages. Moscow: Institute of Linguistics. Russian edition: Atlas kavkazskix jazykov. Moscow: Piligrim [2006]. Kumakhov, Mukhadin A. 1989 Sravnitel’no-istoricˇeskaja grammatika adygskix (cˇerkesskix) jazykov. [A comparative-historical grammar of the Circassian languages.] Moscow: Nauka. Lacroix, René 2009 Description du dialecte laze d’Arhavi (caucasique du sud, Turquie) – Gram- maire et textes. Thèse de doctorat, Université Lumière Lyon II. Lander, Yury 2009 Mnozˇestvennaja reljativizacija: podlinnaja i mnimaja. [The real and sham multiple relativization.] In: Yakov G. Testelets et al. (eds.), Aspekty polisinte- tizma: Ocˇerki po grammatike adygejskogo jazyka, 612–653. Moscow: RGGU. Lander, Yury and Alexander Letuchiy 2010 Kinds of recursion in Adyghe morphology. In: Harry van der Hulst (ed.), Recursion and Human Language. (Studies in Generative Grammar 104), 263–284. Berlin/New York: Mouton de Gruyter. Lyutikova, Ekaterina A. 2000 Reflexives and emphasis in Tsaxur (Nakh-Dagestanian). In: Zygmunt Frajzyn- gier and Traci S. Curl (eds.), Reflexives: Forms and Functions, 227–255. (Typological Studies in Language 40.) Amsterdam/Philadelphia: John Benja- mins. Maisak, Timur A. and Solmaz R. Merdanova 2004 “Proverjatel’naja forma” v agul’skom jazyke: struktura, semantika i gipoteza o proisxozˇdenii [“Verificational form” in Agul: its structure, semantics and a hypothesis about its origin]. In: Yury A. Lander, Vladimir A. Plungian and Anna Yu. Urmanchieva (eds.), Issledovanija po teorii grammatiki 3: Irrealis i irreal’nost’, 430–455. Moscow: Gnosis. Nichols, Johanna 2000 The historical geography of pharyngeals and laterals in the Caucasus. In: Jeff Good and Alan C.L. Yu (eds.), Proceedings of the 25th Annual Meeting of the Berkeley Linguistic Society, Special Session on Caucasian, Dravidian, and Turkic Linguistics, 1–13. Berkeley: Berkeley Linguistic Society. Nichols, Johanna 2008 Case in Ingush syntax. In: Greville G. Corbett and Michael Noonan (eds.), Case and Grammatical Relations. Studies in Honor of Bernard Comrie, 57–74. (Typological Studies in Language 81.) Amsterdam/Philadelphia: John Benja- mins. Greville G. Corbett and Michael Noonan Nikolaev, Sergej L. and Sergej A. Starostin 1994 A North Caucasian Etymological Dictionary. Moscow: Asterisk. kort_006.pod 156 07-12-08 13:18:12 -mt- mt

156 Michael Daniel and Yury Lander

Plungian, Vladimir 2006 K opisaniju armjanskoj glagolnoj paradigmy: “temporalnaja podvizˇnost’” i perfektiv. [Towards a description of Armenian verbal paradigm: temporal mobility and perfective.] Armjanskij gumanitarnyj vestnik 1: 7–20. Polinsky, Maria and Eric Potsdam. 2001 Long-distance agreement and topic in Tsez. Natural Language and Linguistic Theory 19: 583–646. Schulze, Wolfgang 2003 The diachrony of demonstrative pronouns in East Caucasian. In: Dee Ann Holisky and Kevin Tuite (eds.), Current trends in Caucasian, East European and Inner Asian linguistics: Papers in Honour of Howard I. Aronson, 291–347. (Current Issues in Linguistic Theory 246.) Amsterdam/Philadelphia: John Benjamins. Schulze, Wolfgang 2005 Grammars for East Caucasian. Anthropological Linguistics 47: 321–352. Sumbatova, Nina and Rasul Mutalov 2003 A Grammar of Icari Dargwa. (Languages of the World / Materials 92.) München: LINCOM Europa. Tatevosov, Sergei 2001 From resultatives to evidentials: Multiple uses of the perfect in Nakh-Daghest- anian languages. Journal of Pragmatics 33: 443–464. Testelec [Testelets], Yakov G. 1998a Word order in Kartvelian languages. In: Anna Siewierska (ed.), Constituent Order in the Languages of Europe, 235–256. (Empirical Approaches to Lan- guage Typology, EUROTYP 20–1.) Berlin/New York: Mouton de Gruyter. Testelec [Testelets], Yakov G. 1998b Word order in Daghestanian languages. In: Anna Siewierska (ed.), Constituent Order in the Languages of Europe, 257–280. (Empirical Approaches to Lan- guage Typology, EUROTYP 20–1.) Berlin/New York: Mouton de Gruyter. Testelec [Testelets], Yakov G. 1998c Word order variation in some SOV languages of Europe. In: Anna Siewierska (ed.), Constituent Order in the Languages of Europe, 649–679. (Empirical Ap- proaches to Language Typology, EUROTYP 20–1.) Berlin/New York: Mouton de Gruyter. Testelets, Yakov G. and Svetlana Yu. Toldova 1998 Refleksivnye mestoimenija v dagestanskix jazykax i tipologija refleksiva. [Re- flexive pronouns in Daghestanian languages and the typology of reflexives.] Voprosy jazykoznanija 1998–4: 35–57. Tsutsiev, Artur. 2006 Atlas etnopoliticheskoj istorii Kavkaza (1774–2004). [An Atlas of ethnopoliti- cal history of the Caucasus.] Moscow: Europa. Tuite, Kevin 1998 Kartvelian Morphosyntax. Number agreement and morphosyntactic orientation in the South Caucasian languages. (Studies in Caucasian Linguistics, 12). München: LINCOM Europa. Tuite, Kevin 2008 The rise and fall and revival of the Ibero-Caucasian hypothesis. Historiograp- hia Linguistica 35: 23–82. kort_006.pod 157 07-12-08 13:18:12 -mt- mt

The Caucasian Languages 157

van den Berg, Helma (ed.) 2005 Caucasian, Lingua 115. Vogt, Hans 1971 Grammaire de la langue géorgienne. (Instituttet for sammenlignende kultur- forskning B–57). Oslo: Universitetsforlaget. Zhirkov, Lev I. 1955 Lakskij jazyk: Fonetika i morfologija. [The Lak Language: Phonetics and Mor- phology.] Moscow: AN SSSR. kort_006.pod 158 07-12-08 13:18:12 -mt- mt

158 Michael Daniel and Yury Lander