Topological relations in Daghestanian languages.
One of the features that distinguishes Daghestanian languages of the Caucasus from many other languages is the richness of the nominal paradigm, which arises due to large locative subsystems, counting up to eighty forms. This paper presents a description of major distinctions made by Daghestanian in the topological domain. The paper reveals three basic semantic oppositions underlying systems of nominal locative markers and describes a number of minor points of variation. In the domain of location in Ground this is the distinction between location in container and location in substance. In the domain of location on Ground, the primary division in most languages of the family is between attachment and nonattachment configurations. In the domain of location near Ground, several languages use the formal distinction between localization marker and postposition to reflect the semantic contrast between location in a space associated with a given Ground and location near Ground. The detailed comparison shows that the same distinction can function quite differently even in related languages. The paper also makes a number of crosslinguistic observations related to patterns found in Daghestanian.
Semantic study of prepositions and, more widely, other means of expressing location and motion has by now developed into a quite well-established field with its own approaches, problems and results (most important in this respect are various cognitive approaches). Study has so far focused mostly on theoretical problems and applying them to particular locative markers in particular languages, especially European, such as English, French, German, Dutch, and Russian (cf., among others, Brugman 1988; Talmy 2000; Herskovits 1986; Cuyckens 1991; Vandeloise 1991; Mal'ar and Seliverstova 1998; Tyler and Evans 2003; not to mention numerous other articles). The set of theoretical and methodological problems pursued includes, for instance, the following: how to distinguish between meanings and contextual inferences, which of several meanings of a marker should be taken as the central]primary sense, how to formally represent (spatial) semantics (e.g., in the form of image schemas and polysemy networks) and, most crucially, what is the relationship between language, thought and the real world.
Somewhat less known is another approach that aims at building a crosslinguistic typology of the spatial domain (Choi and Bowerman 1991; Bowerman and Pederson 1992; Svorou 1994; Bowerman 1996; Levinson 2003; Levinson and Meira 2003). Adopting many of views expressed by the first approach, this line of research concentrates on another set of questions as to whether spatial categories such as ON and IN are universal, how languages draw borders between different locative expressions, whether we can observe crosslinguistically recurrent clustering of spatial configurations etc. An important step towards answering these questions is the recent collective work at the Max Planck Institute for Psycholinguistics in Nijmegen published as Levinson and Wilkins (2006). Based on the same methodology and using the same elicitation technique, the papers in this volume provide a solid basis for further typological research on spatial categories.
One of the most fundamental parts of spatial language and cognition, often considered to be universal, is the expression of topological relations. Contrary to this, as Levinson and Wilkins (2006: 526) conclude, "there are no simple, surface universals in this subdomain--for example, no universal coding of prototype ON or IN categories of spatial relationship" (see also Levinson and Meira  for a detailed demonstration). Given this general conclusion, it is particularly interesting how languages structure the universal semantic space and how various spatial topological configurations are distributed across locative markers. However, a necessary preliminary step to drawing any robust typological generalizations is to accumulate data from various languages collected in a compatible format.
Within this broader context, the goal of this paper is to contribute to the crosslinguistic investigation of the spatial domain by providing ah account of basic formal means expressing topological relations in Daghestanian languages. (1) In this paper, I do not try to discuss all details of the distribution of locative markers across various spatial configurations in these languages. Rather, my goal here is to provide a description of major distinctions made by Daghestanian languages in the topological domain.
One feature which distinguishes Daghestanian from the languages discussed in Levinson and Wilkins (2006) is the richness of the nominal paradigm, which arises due to large locative subsystems, counting up to eighty forms. In other words, many of the spatial configurations denoted by locative prepositions in the languages of Europe are expressed by bound affixes in Daghestanian. (2) Nevertheless, postpositions are also employed to mark topological relations calling for an interesting question as to which spatial configurations are expressed by (more grammaticalized) nominal locative markers and which are conveyed by means of (less grammaticalized) postpositions. As this paper shows, this is a matter of considerable variation.
The paper is organized as follows. In Section 2, 1 briefly describe typical morphological patterns of the nominal locative forms in Daghestanian. The following discussion is organized around three basic topological categories. Section 3 discusses locative markers related to the expression of location in Ground (IN category). In Section 4, I describe locative markers denoting location on surface of Ground (ON category). Section 5 addresses locative markers expressing various kinds of location near Ground (NEAR category). Section 6 briefly mentions conflation patterns found in Daghestanian. In the final section, the main points are summarized.
2. Morphological makeup of nominal locative forms in Daghestanian (3)
A typical nominal locative form in a Daghestanian language minimally includes markers of two different categories. The first category, traditionally called localization in East Caucasian studies, points to a particular spatial domain of a Ground where a Figure is located, e.g., 'behind Ground', 'under Ground', 'on surface of Ground'. (4)
The second category is orientation which indicates whether a Figure moves from, or moves towards, or rests at a spatial domain defined by the localization marker (some languages make finer distinctions in the orientation domain, e.g., 'move in direction of Ground' vs. 'move to reach Ground' etc.). The typical structure of the nominal locative form in East Caucasian is shown in Table 1.
Usually, one of the orientation values has a zero marker, so that a locative form appears to have only the localization marker. In most Daghestanian languages, it is location at a localization domain (essive), while in Dargwa dialects it is movement towards a localization domain (lative), as in Table 2.
In several Dargwa dialects, a third morphological slot is added to the localization and orientation slots, which contains deictic ('hither'/'thither') and gravity ('up'/'down') markers, as in Table 3.
Here, I do not go into detail concerning various possibilities of filling the second and the third slots found in Daghestanian and turn instead to the presentation of data on the category of "localization". (5)
3. Location inside a Ground
3.1. Hollow containers vs. substances
Unlike many other languages of the world, it is very typical of Daghestanian languages to distinguish between two different localization markers indicating location in Ground, as in the following examples from Archi:
(1) jamu bosor noL'-a-si uq[??]a-li. that.1 man house-IN-ALL 1.gO.PF-EVID (6) 'This man went home.' (Kibrik et al. 1977: 12) (2) ha[??]tarce-qla-k eku-li jamu-t lo. river.OBLPL-INTER-LAT 4.fall.PF-EVID that-4 child 'This boy fell into the river.' (Kibrik et al. 1977: 56)
The distribution of these localization markers is traditionally attributed to the distinction between two types of Grounds, viz. hollow containers and amorphous substances (Kibrik 1970; Kibrik 2003: 43). The IN localization marker expresses location in various kinds of containers, such as 'jug', 'pot', 'kettle', 'hole', 'pit', 'box', 'room', 'house', while the INTER localization denotes location in liquids, substances etc. Compare, for instance, nouns that require the INTER localization in Archi:
(a) substances that do not have their own natural shape: bi 'blood', harq 'sour cream', huq' 'smoke', l:an 'water', oc' 'fire', k'un 'wheat' etc.;
(b) water spaces: ha[??]t[??]ra 'river', lat 'sea';
(c) indiscrete sets of objects: ba[??]k'i 'heap, pile', Lon 'herd', x: [??]ak 'forest', x[??]or 'village' etc.;
(d) collective nouns denoting fruits and plants: arq[??]ut 'nuts', a[??]ns 'apples', mac 'nettle', pil 'onion' etc. (Kibrik 1977: 161-163)
Most Daghestanian languages have similar systems and it is often the container/ substance distinction that is responsible for the choice between two localization markers. However, some languages do not make such distinction and some others show that the choice is not always such straightforward.
Before I pass to the description of semantic variation, it should be noted that in all Daghestanian languages that draw the distinction between location in container and location in substance it is always built into the system of nominal locative markers, i.e., no language uses postpositions to express one or both of these configurations.
3.2. Variation in the expression of containment
With respect to the distinction introduced above, Daghestanian languages can be classified into three types. The first type makes no formal distinction and expresses location in Ground by means of the same marker irrespective of the structure of the former, like European languages. This type includes Bezhta, Hunzib, Lak, Tsakhur, Budukh, Kryz, Udi, Khinalug, and Urahi Dargwa. Compare examples from Udi:
(3) javas~javas ta-z-c-i k:oj-a. slowly go-1sG-ST-AOR house-IN 'I went home slowly.' (4) zu xe-n-a zIe-z bos-e. I water-OBL-IN stone-1sG throw-PFT 'I threw the stone into the water.'
The second type, of which Archi is a good example, has two localization markers to indicate location in Ground. The characteristic feature of this type is that the two localization markers are in complementary distribution. This possibility is found in Avar, Andi, Akhvakh, Bagwalal, Chamalal, Karata, Tindi, Tsez, Hinukh, Khwarshi, Archi, Mugi Dargwa, i.e., it is typical mainly of the western part of Daghestan. Following the terminology introduced by Plungian (2002), such systems can be called "classifying". As the Archi examples above show, the INTER localization marker in this type of languages is used with various kinds of substances, but cannot be used with nouns denoting containers. In contrast, the IN localization is used with the latter class of nouns, but does not combine with nouns denoting substances. These restrictions are so strong that the IN forms of the names of substances and the INTER forms of the names of containers do not even exist in such languages. Compare examples from Mugi Dargwa in Table 4.
However, not all languages of this type draw the borderline between the two localization markers in an entirely straightforward manner. Particularly interesting in this respect are Bagwalal and Chamalal. In Bagwalal, several nouns belonging to the class of containers require the INTER localization and do not have the 1N form. Daniel (2001: 228-229) points out several "exceptions": baL' 'intestine', hand 'it 'a 'ear', misa 'room', awal 'house', masina 'car', kruzka 'cup', hoha 'thimble'. In Chamalal, the distinction between the IN and INTER localizations is even weaker, since many containers have both forms, which are used without any visible semantic contrast (Ganenkov 2005a: 136-138). Perhaps, this may be taken as evidence of the loss of the IN marker and the diachronic drift of the INTER localization towards a generalized marker of location inside.
The distinction between two localization markers conveying the idea of location in Ground is also found in Lezgian, Rutul, Agul, Tabasaran, and Aqusha Dargwa, which fall into the third type. However, the division of labor between two markers is quite different here. In contrast to the languages of the "classifying" type, the use of the markers overlaps to some extent. Moreover, unlike in Bagwalal and Chamalal, the choice between the two markers in the overlapping domain is semantically motivated ("semantic strategy" according to Plungian 2002). Interestingly, almost every language of this type has its own pattern of distribution between the IN and INTER markers.
In Lezgian and Rutul, location in substance is expressed by means of both 1N and INTER forms, but only IN forms can be used for location in container. The following example demonstrates the use of the 1N localization with the noun wac' 'tiver' in Lezgian:
(5) i wac'-a balu[??]-ar awa. this river-IN fish-PL IN:be 'There are fish in this river.'
It should be noted that both localization markers can be used in the same situation in Lezgian and Rutul. It seems that the choice between these forms depends on the extent to which the structure of a Ground is in the speaker's focus. If the physical structure of the Ground is irrelevant in a given context, then the IN localization is used. However, if it is important to point out that the Ground is a substance, then the INTER localization should be used.
Agul also has 1N forms of substances and, like Lezgian and Rutul, uses them if the structure of the Ground is irrelevant in a given context. However, Agul makes a further step and, in addition to the possibility to use the IN marker with substances, allows INTER forms of containers. Such forms are metonymically interpreted as location in a substance that is in this container. Note that this metonymic interpretation is not possible in the languages of the "classifying" type discussed above. In the following Agul example the INTER form of the noun Hujeg 'pan' has the reading 'in the water boiling in the pan':
(6) mi [??]ix-a-j-e Hujeg-i-[??] pe[??]. this:ERG INTER:put-IPF-CONV-COP pan-OBL-INTER chicken 'She puts a chicken in the pan.'
The most interesting situation is found in Tabasaran, which also does not restrict the use of the localization markers to either of the classes of Grounds. Again, like Agul, Lezgian, and Rutul, Tabasaran uses IN forms of nouns denoting substances when the physical structure of a Ground is irrelevant. However, the interpretation of INTER forms of nouns denoting containers here is very different from that found in Agul. The use of the INTER form here conveys the idea of "close containment", which can come in three different guises: (i) Figure occupies the whole of the inner region of a container, so that there is no longer any free space inside, (ii) Figure hardly enters Ground, or (iii) Figure is fixed in Ground. What all these interpretations have in common is that the spatial position of the Figure is confined by the limits of the inner space of the Ground. Compare the following examples:
(7) uzu canta-ji-[??][??] ip'rub [??][??]iwun-za. I bag-OBL-INTER food INTER :put.PST-1SG:AG 'I put food in the bag (so that the bag is full of food).' (8) urq-a-[??][??] k'az [??][??]iwun-za. hole-OBL-INTER paper INTER:put.PST-1SG:AG 'I stuffed a piece of paper into the hole.' (9) uzu taxc:a-ji-[??][??] [??][??]eun-za. I wardrobe-OBL-INTER INTER:climb.PST-1SG:AG 'I squeezed into the wardrobe (e.g., in order to hide).' (10) uld-[??][??]-an suse [??][??]ada-b-[??][??]-ub window-INTER-ELAT glass INTER:EL AT-N-carry-NMLZ 'to pull out glass from the window'
Another pattern is attested in Aqusha Dargwa, where the IN forms of nouns denoting substances are also possible. However, here they express a specific meaning of location in Ground perceived as a property of the Ground. Compare the following examples:
(11) [??]um-li-zi-r GarGu-bi ler. sand-OBL-INTER-ESS stone-PL be 'There are stones in the sand.' (12) [??]um-la-r daqal GarGu-bi ler. sand-IN-ESS many stone-PL be 'There are a lot of stones in the sand.'
The difference between these examples is that in (11) a simple location of stones in sand is meant, while (12) describes properties of sand, i.e., the sand is such that it contains a lot of stones ('stony sand') and, for example, cannot be used for particular purposes. In Daghestanian, this meaning is usually conveyed by the INTER localization marker and, thus, is not distinguished from location in substance. The only exception are Agul and Tabasaran (both closely related East Lezgic languages) where this meaning is expressed by the attachment localization (see Section 4).
3.3. Other spatial configurations related to containment
The last meaning introduced in the previous section brings us to the discussion of several other spatial configurations typically expressed by IN or INTER forms in Daghestanian.
First of all, most Daghestanian languages use the 1NTER localization to denote location among Grounds, as the example from Archi in (13) demonstrates:
(13) qabaq-mulce-q[??] cele bo-xo. pumpkin-PL-OBLPL-INTER stone 3-find.PF 'They found a stone among pumpkins.' (Kibrik 1977: 174)
The only exceptions are Lezgian, Rutul and Udi (all Lezgic languages), where this meaning is expressed by a postposition. Consider the following Lezgian example (Talibov and Gadzhiev 1966: 48):
(14) ci arada axtindi awa-c. we:GEN among/between such IN:be-NEG 'There is no such person among us.'
Another similar meaning is location between two (or more) Grounds. In virtually all languages of the family, this spatial configuration is expressed by means of a postposition, as in Example (15) from Archi:
(15) os q[??]'[??]e-bu ga[??]nt'i-li-s q[??]'anak uq[??]a-li. one two-3 stack-OBL-DAT between 1.go.PF-EVID 'He came between two haystacks.' (Kibrik et al. 1977: 46)
The only languages where this meaning is expressed by means of a localization marker are Agul and Tabasaran, which both use the INTER form. Compare Example (16) from Agul:
(16) ce zul-ar-i-[??] [??]u xal [??]a-a. We(EXCL):GEN house-PL-OBL-INTER two house INTER:be-PRS 'There are two houses between our houses.'
Interestingly, in accordance with the general idea of close containment discussed above, Tabasaran uses the INTER localization only if the space between Grounds is very small, so that a Figure hardly fits in it. Otherwise, the postposition araji[??] 'between' is used. Compare:
(17) ic xul-ar-i-n araji[??] har-ar kudbc'[??]-ura. we(EXCL):GEN house-PL-OBL-GEN between tree-PL grow-PRS (18) ic xul-ar-i-[??][??] har-ar kudbc'[??]-ura. We(EXCL):GEN house-PL-OBL-INTER tree-PL grow-PRS 'A tree grows between our houses.'
Example (17) describes a situation where there are two houses located at a certain distance from one another and a tree grows in between them. In contrast, Example (18) only describes a situation where the two houses adjoin one another and a (shoot of a) tree grows from a small space between the adjoining walls of the houses.
Third, location of a Figure wrapped in Ground is always conveyed by means of a localization marker. Typically, Daghestanian languages use the INTER localization in this meaning, as in the following example from Karata:
(19) adjal-i-Li mak'e beg[??]az-a:la. blanket-OBL-INTER child wrap.oneself-CAUS:INF 'to wrap the child in a blanket' (Magomedova and Khalidova 2001: 41)
Nevertheless, at least some Dargwa dialects (e.g., Mugi Dargwa) use the IN localization in such contexts, while some other dialects of Dargwa allow both localizations. Consider the following example from Xuduc Dargwa:
(20) nis:e caq[??]a-le/ caq[??]a-l-c:i bik:ibark:-a. cheese fabric-IN fabric-OBL-INTER wrap-IMP 'Wrap the cheese in the fabric.'
Finally, it should be noted that an interesting variation can be observed in the use of the localization markers with the nouns 'forest' and 'village'. This is evidently conditioned by the ambivalent nature of the objects denoted by these nouns. On the one hand, they denote objects that consist of a set of similar elements and, thus, can combine with the INTER localization. On the other hand, the nouns denote large spaces which favors the use of the IN localization.
The spatial configuration 'location in forest' is usually expressed by means of the 1NTER localization, as in Agul:
(21) gada dar-a-[??] gul-u-ne. boy forest-OBL-INTER disappear-PF-PFT 'The boy got lost in the forest.'
Sometimes there is variation with regard to which marker is used. For example, in Archi both possibilities exist, so that the form x[??]ak:-e-q[??] [forest-OBL-INTER] represents the forest as a collection of trees, while the form x[??] ak:-a [forest-IN] highlights the idea that the forest is a limited space (Kibrik 1977: 162).
The expression of location in a village varies across the languages of the family. In Avar, Agul, and Lezgian, only the IN localization is possible, compare the Lezgian example below:
(22) pahliwan-ar isatda ci xur-e awa. artist-PL now we:GEN village-IN IN:be 'The tightrope walkers are in our village now.' (Haspelmath 1993: 102)
On the other hand, the only option in Archi and Andic languages is to use the INTER localization, as in Godoberi:
(23) den han-Li [??]idu w-a[??]a. I village-INTER Godoberi M-come.PST 'I came to the village of Godoberi.'
3.4. The crosslinguistic perspective
To summarize, the previous discussion reveals that several semantic contrasts in the spatial domain are relevant for the description of crosslinguistic variation in Daghestanian. First of all, it is the distinction between hollow containers and substances. From a typological point of view, this is not entirely unexpected, since the "classifying strategy" (Plungian 2002) based on the Ground geometry/ structure is also found in other languages, cf. the Tiriyo "aquatic" postposition hkao (Meira 2006), "dispositional" predicates in Tzeltal (Brown 1994, 2006) or verbal locative suffixes in Atsugewi (Talmy 2000). What distinguishes Daghestanian from, for example, Tzeltal is that the distinctions of the Ground geometry in the former are not so elaborated as in the latter. As far as I can judge from the existing literature, no such type of classifying system has been described elsewhere.
The distinction between containers and substances divides Daghestanian into three types. The first type simply does not make this distinction. The "classifying" type has two localization markers, which are in complementary distribution. The distribution of markers is slightly different in the languages of the "semantic" type. Here, minimally, one marker denotes location inside of both containers and substances, while the other marker is used to highlight the physical properties of a mass Ground.
From a genetic perspective, there is a clear tendency for closely related languages to behave similarly with respect to the distinction between containers and substances. Thus, all Andic languages are of the "classifying" type, while most Lezgic languages belong to the "semantic" type. However, as Dargwa dialects demonstrate, this feature is not necessarily so stable. For example, Urahi Dargwa uses the same marker for all kinds of Grounds, while Mugi Dargwa and Aqusha Dargwa employ two different localization markers, with the former belonging to the "classifying" type and the latter belonging to the "semantic" type.
Another interesting point is the distinction between "loose containment" and "close containment" found in Tabasaran. Again, it is not the case that Tabasaran is unique in having such a distinction, since similar contrasts were described, for instance, in Dutch and Korean (Lemmens 2002; Choi and Bowerman 1991). However, what makes Tabasaran really unique is that here this opposition is employed in the grammaticalized system of nominal locative markers, while Dutch builds it into the lexical system (cf. the verb zitten 'sit' in Dutch and the verb kkita in Korean).
The behavior of localization markers in Daghestanian also shows that a specific meaning 'location in Ground as property' should be distinguished crosslinguistically. While in most languages it is not formally (and conceptually) distinct from location in substance, Aqusha Dargwa classifies this meaning with location in container. Furthermore, Agul and Tabasaran distinguish it from both location in substance and location in container and use the attachment localization in this meaning.
Finally, Daghestanian languages are quite unusual in that the majority of them express the spatial meaning 'among' by the same marker as containment, while it is generally more typical to use two different formal means, as in English in and among. Even more unusual is that Agul and Tabasaran expand the scope of the INTER localization by including the meaning 'between'.
4. Location on the surface of Ground
4.1. Two basic ON-configurations
As in the case of location in Ground, many Daghestanian languages distinguish two localization markers, both of which (at least in some contexts) denote location on surface of Ground, as in the following Agul examples:
(24) ust:ul-i-l kitab ald-e-a. table-OBL-SUPER book SUPER-be-PRS 'There is a book on the table.' (25) cil-i-k isk:il k-e-a. wall-OBL-CONT picture CONT-be-PRS 'There is a painting (hanging) on the wall.'
The study of various kinds of location on surface shows that the distinction between two localization markers is organized around two basic spatial configurations (i) and (ii). (7) In other words, in all Daghestanian languages that distinguish between two markers the SUPER localization expresses at least (i), whereas the CONT localization expresses at least (ii):
(i) location of a Figure on the supporting (horizontal) surface of a Ground, e.g., 'book on the table',
(ii) location of a Figure attached to a (vertical) Ground, so that the latter prevents the former from falling down, e.g., 'picture on the wall'.
Again, as in the case of containment, a number of languages uses the same marker in all spatial configurations pertaining to location on surface, i.e., they do not distinguish between SUPER and CONT. This group includes Akhvakh, Avar, Lak, Archi, Budukh, Kryz, Udi, and Khinalug. Compare these examples describing two basic configurations (i) and (ii) in Lak:
(26) stol-dani-j bur lu. table-OBL-SUPER is book 'There is a book on the table.' (27) c'ira-j larqun dur surat. wall-SUPER hung is picture 'A photo hangs on the wall.'
It should be also noted that, as in the case of containment markers (see Section 3 above), the distinction is always rendered by means of morphological nominal markers. No language is attested where postpositions would be used for one or both relations.
4.2. Other spatial configurations
A number of other spatial configurations belonging to the ON category has been discussed in the literature (see especially "The topological relations picture series" in Levinson and Wilkins ). In Ganenkov (2009), I investigated the distribution of localization markers in the following contexts:
(28) (a) living Figure on nonsupporting Ground: 'fly on wall' (b) a part on the whole: 'ears on head', 'fingers on hand' etc. (8) (c) possession: 'X has fingers', 'X has no tail' etc. (d) 'lump (wart, mole, wound) on skin/body part' (e) natural extensions: 'apples/leaves on twig' (f) 'ring on finger' (g) 'glove on hand' (h) 'coat on hook' (i) 'clothes on line' (j) 'mark/image on surface': 'spots on table', 'man on portrait'
The spatial configurations listed in (28) do not behave in a uniform way in all Daghestanian languages, i.e., each of them patterns with (i) in some languages, but goes with (ii) in others. Compare Example (29) from Agul, where the CONT localization is used in the configuration 'spots on tablecloth', and Example (30) from Tsez, which allows only SUPER in this context: (9)
[FIGURE 1 OMITTED]
(29) sufra-ji-k t'ant 'a-jar k-e-a. tablecloth-CONT spot-PL CONT-be-PRS 'There are spots on the tablecloth.' (30) stol-jo-L' medo-bi jol. table-OBL-SUPER spot-PL is 'There are spots on the table.'
Ganenkov (2009) concludes that the distribution of Daghestanian localization markers in the ON-domain is captured by a semantic map. A modified version of this map is presented as Figure 1. (10)
The semantic map in Figure 1 is designed to constrain possible variation in semantics of localizations in Daghestanian. However, it does not itself provide an explanation that would account for why the semantic structure of a marker is organized in a given way. To understand motivations underlying the distribution of markers, it is crucial to note that configurations (i) and (ii) do not logically exclude one another. Thus, nonsupport does not automatically mean attachment (as in 'fly on wall'), while nonattachment does not necessarily imply support (as in 'man on portrait'). This corresponds to two basic ideas according to which spatial configurations clustered around the basic contexts (i) and (ii), viz. (a) whether Ground supports Figure or not, and (b) whether Figure is attached to Ground or not. Accordingly, Daghestanian languages further divide into two types.
4.3. The support type
In general, the support type is much less numerous than the attachment type. In fact, only Tsez and, possibly, two other closely related West Tsezic languages Hinukh and Khwarshi use support as the property determining the choice between two localization markers; i.e., all contexts where a Ground does not support a Figure are expressed by CONT, while SUPER is used only in support configurations. This leaves the SUPER localization in Tsez with only two spatial configurations 'book on table' and 'marks on surface', while the CONT localization covers the rest of the map. Apart from the basic scene 'book on table', only the second scene 'marks on surface' does not fall under the rubric of 'nonsupport'. Note that the notion of support is merely not applicable to this scene, which happens to be a sufficient condition for the use of SUPER. The following examples from Tsez demonstrate the use of the localization markers:
(31) stol-jo-L' t'ek jot. table-OBL-SUPER book is 'There is a book on the table.' (32) stol-jo-L' medo-bi jol. table-OBL-SUPER spot-PL is 'There are spots on the table.' (33) karta qido-q jok'a map wall-CONT hang(tr.) 'to hang the map on the wall' (34) qido-q t'ut' jol. wall-CONT fly is 'There is a fly on the wall.'
4.4. The attachment type
Other Daghestanian languages employing two localizations for ON scenes belong to the second type and distribute spatial configurations between two markers according to the attachment parameter. A good example of this type is Bezhta, where the CONT localization is used in various attachment configurations, but cannot express nonattachment relations. Compare the following Bezhta examples:
(35) joso-l gababek'e bozon-na gej. wall-CONT carpet hang-RES COP 'The carpet hangs on the wall.' (Khalilov 1995: 48) (36) XiL.a-l bi[??]jk'i micuc'-na. trousers-CONT agrimony attach-RES 'An agrimony (pod) stuck to trousers.' (Khalilov 1995: 57) (37) *ist'oli-ja-l t'ek gej. table-OBL-CONT book COP 'There is a book on the table.'
While it is quite obvious what is meant by the notion of support, it is not always so with the notion of attachment. In particular, many Daghestanian languages express at least some attachment configurations by means of the SUPER localization. For instance, Bezhta allows both markers in certain attachment contexts:
(38) susa.li-L'a / susa.li-l Libo gej. bottle-SUPER bottle-CONT paper COP 'There is a label on the bottle.'
Attachment configurations like (38), where both SUPER and CONT can be used, constitute the category of nonsalient attachment, i.e., attachment configurations where the presence of a sticky substance (or some other means to attach the Figure to the Ground) is not salient, e.g., 'label on bottle', 'stripes on sleeve' etc. Usually, in such configurations the Figure is a thin flat object attached to the flat surface of the Ground, so that there is no special visual evidence of the attachment relation between them. Rather, these configurations are viewed as simple location of a Figure on a Ground with no focus on the attachment relation.
Languages differ in the extent to which they employ the SUPER localization marker in attachment configurations. In Bezhta, Godoberi, and Chamalal, the SUPER localization can replace the CONT marker only in nonsalient attachment configurations with no observable semantic difference, as in Example (38) above.
In Agul, Lezgian, Tabasaran, and Rutul, only the SUPER localization is possible in nonsalient attachment configurations. In contrast to Bezhta, the use of the CONT localization in contexts favoring the nonsalient attachment reading gives the interpretation that the Figure is conceived as being in a very tight attachment relation, so that it is very difficult to separate it from the Ground. In the configuration 'label on bottle', for instance, the use of CONT is interpreted by speakers of these languages as indicating that there remain little pieces of the label on the bottle that are very difficult to remove from the surface.
Finally, in Andi and Xuduc Dargwa the SUPER localization can be used in all kinds of location on a surface, whereas the use of the CONT localization is restricted only to configurations where the attachment relation is present. This means that in these languages the CONT localization can be replaced by the SUPER localization even in configuration (ii), as in:
(39) lac-li-c[??]i-b / lac-le-b sakemq-un-ni c'eb surat. wall-OBL-CONT-ESS wall-SUPER-ESS hang-AOR-CONV is picture 'A picture hangs on the wall.'
As is shown in Section 4.2, in Tsez the scene 'man on portrait' to which the notion of support cannot be directly applied is clustered with the support scene rather than with the attachment scene. With regard to the configurations to which the notion of "attachment" is not applicable, both possible clusterings, i.e., with (i) and (ii), are observed. For example, the scene 'ring on finger' is expressed by SUPER in Agul, Tabasaran, Lezgian, Tsakhur, Rutul, and Andi, whereas Bezhta and Godoberi use CONT here. Several examples are given in (40)-(43).
(40) Xuduc Dargwa iti-la naq[??]-le-d kam-li cad k'[??]el t'up. he-GEN hand-SUPER-ESS lack-CONV is two finger 'He lacks two fingers on his hand.' (41) Bezhta hogco-la miqa[??]-l habo-da gej. he-GEN back-CONT wing-PL is 'He has wings on his back.' (42) Rutul hemi bedri.je-k t'ul ki-dis. this bucket-CONT handle CONT.be-PRS:NEG 'This bucket has no handle.' (43) Agul gi-n qusi-k p[??]ara t'ant'a-jar k-e-a. that-GEN face-CONT many spot-PL CONT-be-PRS 'There is a lot of pimples on his face.'
Several languages (particularly, Agul and Rutul) use both the SUPER and the CONT marking for 'mark/image on surface' and 'lump on skin'. Interestingly, according to my informants, the two localizations have different interpretation in these contexts and distinguish location of mark/lump on surface (the SUPER localization) from an indication that a mark/lump is a distinctive feature of the Ground (the CONT localization).
4.5. Location all over surface
A very tare localization marker denoting location on surface is attested in some languages of the Andic branch, particularly in Chamalal, Tindi, and Godoberi. Here, the DIF localization indicates that a (mass) Figure is spread all over the surface of a Ground of a plural Figure is located in various parts of the surface of a Ground. The same localization marker is used in contexts describing path. Compare examples from Chamalal:
(44) amu-q boL'uq az ida. roof-DIF many snow COP 'There is a lot of snow over the roof.' (45) q'unna-q t'unt'-e ida. wall-DIF fly-PL COP 'There are flies over the wall.' (46) amu-q-o ida di wunaLaq. roof-DIF-ELAT COP I going 'I am going on the roof.'
4.6. Summary and discussion
To summarize, many Daghestanian languages make use of two different localization markers to express location on surface. It is shown that the basis for this distinction is the notions of support and attachment. Correspondingly, some languages base the formal distinction between markers on the opposition "support vs. nonsupport", whereas other languages discriminate between the markers on the basis of the "attachment vs. nonattachment" opposition. It is also shown that the notions of support and attachment can be metaphorically extended to include spatial configurations to which they cannot be directly applied. The distribution of markers is constrained by the semantic map on Figure 1.
Crosslinguistically, this distinction cannot be called exotic. Among European languages, the distinction between support and attachment is made by several West Germanic languages including German and Dutch, which have the prepositions auf/an and op/aan, respectively. (11) It is also found in various languages outside Europe, such as Yeli Dnye (Levinson 2006) and Tiriyo (Meira 2006). Moreover, as Bowerman and Pederson (1992) (see also Bowerman and Choi 2001) show, the attachment category can also be seen from comparison of distribution of IN and ON markers in different languages. Levinson and Meira (2003) also recognize the crosslinguistic importance of the attachment category.
However, not much linguistic data on this topic has been collected so far. Daghestanian languages provide valuable information about the overall semantic organization of such markers. As is discussed above, support and attachment do not exclude one another so that some languages interpret the formal distinction between markers as the "support vs. nonsupport" opposition, while others take the opposite opportunity and use the formal distinction to contrast attachment and nonattachment. Interestingly, oppositions can be diachronically related to one another. In Andi, for example, the CONT localization expresses attachment configurations. Taking into account that the latter marker is a cognate to the "nonsupport" marker in Tsez, this may give an idea of how diachronic shifts from one type to another can happen. In a similar way, a diachronic change to the system with only one ON marker can occur, as in Chamalal where the CONT localization became the generalized ON-marker.
Another point that Daghestanian languages demonstrate is that the difference between attachment and nonattachment is not as clear-cut as it may appear. In particular, there is a class of spatial configurations that have the attachment relation but, nevertheless, tend to cluster with nonattachment scenes. This is clearly seen in Bezhta, which indiscriminately uses both the support and attachment markers in such contexts. Furthermore, Agul clusters this scene exclusively with nonattachment. One can conclude from these observations that it is not the formal presence of the attachment relation that influences the choice between markers, but rather the conceptualization of the Figure as attached or nonattached.
Finally, the meaning 'spread all over the surface' is grammaticalized as a separate localization marker in a few languages. Crosslinguistically, it is also not often attested, though there are some indications in the literature (e.g., see Levinson  for Yeli Dnye).
5. Location around Ground
The semantic domain of location around Ground comprises two different sub-domains. On the one hand, unspecified location somewhere near a Ground belongs here, which is the main point in Sections 5.1-5.3. On the other hand, location around Ground includes various kinds of more specific topological relations like 'under', 'in front of', 'behind', and 'above' which are discussed in Sections 5.4-5.5.
Before I pass to the description, it should be noted that the NEAR domain in Daghestanian is investigated in less detail than the IN and ON domains, so that the limits of semantic variation within this domain still remain unclear to a considerable extent. For example, most Andic and Tsezic languages possess two different localization markers describing location near Ground. However, it is not clear from existing descriptions of the languages what is the difference between these markers and whether there is any semantic contrast at all, as is the case of the markers -[??]a and -dor in Bezhta (Kibrik and Testelec 2004). Moreover, it is sometimes noted that one of the markers is used mostly in nonlocative meanings, while locative uses are very rare, cf. Magomedova (1979: 64-65) on Andic languages. Compare, for instance, examples from Akhvakh where both localization markers -xari and -qe denote location near a Ground, but the second of the markers -qe is reported to occur very rarely in this meaning:
(47) imixi beL:'o-xari bik'-ari. donkey barn-APUD1 be-PRS 'A donkey is standing near the barn.' (48) igo-qe q'inu-ruLa window-APUD2 stand-INF 'to stand by the window' (Magomedova and Abdulaeva 2007: 668)
Similarly, Lak grammars also describe two localization markers denoting location near a Ground -c'a and -x (Zhirkov 1955). However, my informant did not confirm the existence of a form like t:urc 'a-x 'near a (telephone) pole' given in Khajdakov (1999: 353) in the paradigm of the noun t:urc 'a 'pole', so that it is not clear whether the form actually exists and whether it has a locative meaning.
Below, I describe two patterns that exist in Daghestanian languages and the discussion is confined mostly to languages of the East Lezgic and Dargwa groups for which I have my own field data.
5.1. Location in the region of the Ground vs. location near the Ground
The first pattern attested in Daghestanian is the formal contrast between 'location in the region associated with a given Ground' and 'location near a Ground'. This pattern was first found in the East Lezgic languages Agul, Lezgian, and Tabasaran as shown in Examples (49) and (50) from Agul:
(49) diwan dak'ar-i-q qix-i-ne zun. sofa window-OBL-POST POST:put-PF-PFT I 'I put the sofa near the window.' (50) ha-te dar-ala-n q[??]alaq qame-a. EMPH-that tree-OBL-GEN near POST:remain-PRS 'He stayed near that tree.'
Note that, unlike two binary distinctions discussed above, the second meaning here is always expressed by a postposition, as in the Agul examples (49) and (50), where the localization marker -q and the postposition q:alaq are used. (12)
The meaning 'location in the region associated with a given Ground' can be described as follows. When using the localization marker -q speaker has in mind not only a particular Figure-Ground pair, but also the whole space to which both belong. This whole space is distributed between several prominent Grounds, so that every fragment of the space is associated with one of these Grounds. Most often, immobile objects which have a fixed position within the space and organize it are selected as prominent Grounds. For example, the whole of a yard is divided into 'space near the house', 'space near the fence', 'space near the gate', 'space near the barn' and the whole of a room is divided into 'space near the window', 'space near the door' and, perhaps, some others. Besides, a fragment of space 'the middle, the center' in such a division is not associated with any of the Grounds. The function of the localization marker is to indicate a fragment of the space where the Figure is located. Often, such forms get a conventionalized interpretation if a Ground has an established space that is considered to be associated with it (e.g., space in front of the mayor's office in a village).
Importantly, the limits of such fragments are not strictly defined, i.e., it is not a matter of distance between a Ground and a Figure, but rather a matter of conceptualization by speakers. This means that one cannot define exactly whether a particular Figure at a certain distance from the Ground still belongs to the region of this Ground of if it is already outside it. Of course, the Figure in such cases is indeed close to the Ground, but the main idea expressed by the localization marker is not to indicate the distance, but to identify the position of the Figure within the whole space.
The description above implies that localization markers cannot be used to denote location near a Ground if several Grounds of the same kind are present; for example, one cannot use the POST marker to express location near a tree in a forest. Another restriction is that the localization marker cannot be used in case of small or movable Grounds, such as 'spoon', 'book', or human beings.
The postposition in such systems expresses simple location of a Figure near a Ground. In contrast to the first marker, it is a close-up description ('the Figure is not far from the Ground') without the reference to the overall makeup of the whole space.
A typical feature of such systems is that the localization markers are used comparatively rarely in this function, and the main means of expressing location near a Ground is the postposition. The second feature following from the semantic contrast between two markers is that the localization marker is very unlikely to be elicited from native speakers by means of translation of a sentence from another language. Rather, it either occurs spontaneously or can be elicited if a linguist constructs the form of a noun and asks a speaker to provide an example of the form in a locative context. This can be explained by the specific semantic content of the localization marker. As is described above, when using the localization marker, the speaker should have in focus not only a particular Figure, but also the whole space it belongs to. This is easily achieved in the naturally occurring discourse flow, when the whole situation is activated in a speaker's mind, but is hardly possible in an isolated sentence.
Although I have no reliable evidence of such a system in any other language of the family, it seems flora existing descriptions that Tsakhur (the localization marker -s:/-sana and the postposition janalj), Archi (the localization marker -r and the postposition l:ak), and Khinalug (the localization marker -x and the postposition koli) also follow this pattern. (13)
5.2. Generalized 'near'
The second pattern of the NEAR marking is found in various Dargwa dialects. It is characteristic of this pattern not to distinguish between two kinds of location near a Ground as described in the previous section. Moreover, the two kinds are not even distinguished conceptually. An important consequence is that the localization marker is normally used in all contexts describing location near a Ground, while postpositions play only a marginal role. In Dargwa, the same marker -s:u/-s:a is used in both meanings. Compare Example (51) from Qunqi Dargwa, which has both interpretations: (i) the speaker (mother) and the hearer (daughter) are on the beach and the speaker tells the hearer not to approach the water, and (ii) both the speaker and the hearer are at home in the village, and the speaker tells the hearer not to go to the (region of the) river.
(51) erk'[??]-li-s:a ma-q'at:! river-OBL-APUD PROH-go 'Don't go to the river!'
Judging from examples found in descriptions of Rutul and Avar, these languages also possess this type of the NEAR marking (the localization markers -x and -q, respectively).
5.3. Other NEAR configurations
A very rare localization marker that is attested only in Lak is the Foot localization that denotes location 'at the foot' of a Ground. (14)
(52) na murx-ira-c' ut:uiwxura. I tree-OBL-FOOT is.lying 'I am lying at the foot of a tree.'
It is common to East Caucasian languages that one of the localization markers expresses location in a space associated with a person. Most often, such a marker refers to a person's house, but it can also extend to a wider space, such as village, ethnic group, etc. Crosslinguistically, this meaning is typically expressed either by a NEAR marker, as in Dargwa, or by an attachment localization, as in many Andic languages. However, Bagwalal possesses a distinct localization that conveys the idea of location in someone's personal domain (including house and body), as in the following example: (15)
(53) di-la lel-u-li s'ana b-eli. I-PERSLOC hand-OBL-INTER thorn M-went 'I've got a splinter in my hand.' (lit. 'At me in hand a thorn came') (Daniel 2001)
Interestingly, at least one language of the family, Bezhta, has a localization marker dedicated to the expression of location at somebody's place and cannot describe other kinds of locations associated with a human Ground:
(54) do zuq'oro [??]ali-la?. I was Ali-HOME 'I was at Ali's place.'
5.4. Horizontal axis
A few Daghestanian languages have a distinct marker denoting location behind Ground. This group includes Agul, Lezgian, Tabasaran, Rutul, Lak, and several Dargwa dialects. Compare the following example from Icari Dargwa:
(55) dehni uskul-le-he-b b-ub[??]-a[??]-t[??]i-di. children school-OBL-POST-ESS HPL-play-PRG-CONV-PST 'The children were playing behind the school.' (Sumbatova and Mutalov 2003)
In other languages of the family, such a distinct localization marker is absent and a postposition is used to express this spatial configuration.
The ANTE localization denoting location in front of the Ground is very rare in Daghestanian. It is found only in a number of Dargwa dialects and in Agul. Compare an example from Icari Dargwa:
(56) tukan-ni-sa-d q:arpuz-i d-irc-a-ca-d. shop-OBL-ANTE-ESS watermelon-PL NPL-sell:IPF-PRG-PRS-NPL 'They sell watermelons in front of the shop.' (Sumbatova and Mutalov 2003)
5.5. Vertical axis
In Daghestanian, the most common way to convey location above the Ground is to use a postposition. The only exception is Tabasaran, where the SUPER localization marker is possible in such contexts:
(57) tepe-ji-[??]in luq'[??] t'ibxu-ra. hill-OBL-SUPER eagle fly-PRS 'An eagle is hovering over the hill.' (Alekseev and Shikhalieva 2003: 41)
As for location under the Ground, most languages of the family, except Tsakhur, Kryz, Budukh, Khinalug, and Udi, express it by means of a distinct nominal suffix, as in Karata:
(58) mile ba[??]a-c'e bik'[??]a-la ros-b-ada-L'i-r je[??]i. sun receive-NEG be-INF tree-PL-OBL-SUB-LAT gO:IMP 'Go under trees in order to hide from sun.' (Magomedova and Khalidova 2001: 452)
6. Conflation patterns
In the previous discussion, I showed that not all languages make formal distinctions within each of the three basic categories, i.e., IN, ON, and NEAR. Thus, for example, Bezhta does not distinguish between containers and substances within the IN category, Avar does not have distinct markers for the support and attachment configurations within the ON category, while Lak lacks both oppositions. In this section, I briefly describe conflation patterns occurring between these main categories.
The first pattern is found in Rutul, Lezgian, Tindi, and Xuduc Dargwa, which conflate location in substances and attachment configurations in the same localization marker, as in the following Rutul examples:
(59) xinix xid-i-k ki-r-xu-ri. boy water-OBL-CONT CONT-M-fall-PST 'A boy fell down into the water.' (60) mas-ali-k sikil ki. wall-OBL-CONT picture CONT:be 'There is a picture on the wall.' (Makhmudova 2001 : 81)
From a diachronic point of view, as comparison with other Andic languages and Dargwa dialects shows, the marker -Li in Tindi and the marker -c[??]i in Xuduc Dargwa originally meant location in substance and only later expanded to attachment configurations. The same historical change happened in Bezhta and Hunzib, where the attachment marker -l is cognate to markers denoting location in substance in other Tsezic languages.
In Lezgian, another meaning 'location under Ground' is also expressed by the same marker, which thus covers three different domains: location in substance, attachment configurations, and location under Ground. Crosslinguistically, a similar conflation of IN and UNDER domains was reported for some Australian languages (Wilkins and Evans 1995). I am not familiar enough with the details of the history of locative markers in Australian to judge whether this conflation pattern is a result of semantic development or a mere coincidence. However, the reconstruction of localization markers in Proto-Lezgic shows that the conflation of 'under' with 'location in substance/attachment' is a result of the merger of two different localization markers due to historical phonological changes (Alekseev 2003).
The third conflation pattern is attested in Xuduc Dargwa where the marker -le denotes both location in containers and support scenes:
(61) ust'ul-le-d Hinc-bi ted. table-SUPER/IN-ESS apple-PL are 'There are apples on the table.' (62) jisik'-le-d Hinc-bi led. box-SUPER/IN-ESS apple-PL are 'There are apples in the box.'
The information on conflation patters in Daghestanian is summarized in Table 6. Note that the combination of two conflation patterns in Xuduc Dargwa yields a minimal system where two basic oppositions 'location in containers' vs. 'location in substances' and 'support' vs. 'attachment' are expressed by only two markers, in contrast to maximally possible four markers (e.g., in Agul).
In this paper, I have given an overview of expression of topological relations in Daghestanian languages. I have revealed three basic semantic oppositions underlying systems of nominal locative markers and described a number of minor points of variation. In the domain of location in Ground this is the distinction between location in container and location in substance, though some languages deviate from this strict division to some extent. Tabasaran is the most remarkable case, since the original distinction between two classes of Grounds gave rise here to the distinction between 'loose containment' and 'close containment'. In the domain of location on Ground, the primary division in most languages of the family is between location attachment and nonattachment configurations. The three West Tsezic languages employ the other strategy and distinguish between support and nonsupport scenes. Finally, in the domain of location near Ground, several languages use the formal distinction between localization marker and postposition to reflect the semantic contrast between location in a space associated with a given Ground and location near Ground.
As was mentioned at the outset of this paper, the striking difference between Daghestanian languages and many other languages of the world is that topological relations are expressed by means of nominal suffixes, rather than by adpositions or verbal roots/suffixes. However, from a semantic point of view, Daghestanian systems of nominal locative markers are close to systems of prepositions in European languages. Although there are several specific semantic contrasts which are not found in all or most European languages, the logic of the organization of the universal semantic space is basically the same.
Nevertheless, Daghestanian systems of localization markers are fairly interesting from a typological point of view. The most important is that they provide rich data for further investigation of already known categories, such as attachment, support, and containment. The detailed comparison shows that the same distinction can function quite differently even in related languages. Thus, for example, the same underlying semantic contrast between support and attachment gives different distribution of markers in Tsez and Agul. Another thing worth mentioning here is that Daghestanian nominal locative systems include several spatial configurations, such as 'all over surface', 'at the foot of Ground', 'at someone's place (chez)', which are not otherwise frequently expressed by a highly grammaticalized dedicated marker.
Institute of Linguistics, Moscow
1SG first singular; ADD additive particle; AG agentive agreement; ALL allative; ANTE localization 'in front of Ground'; AOR aorist; APUD localization 'near Ground'; CAUS causative; CONT localization 'in attachment with Ground'; CONV converb; COP copula; OAT dative; DIF localization 'over the Ground'; DOWN motion down; ELAT motion from Ground; ERG ergative; ESS location at', EVID evidential; EXCL exclusive 1st pronoun; FOOT localization 'at the foot'; GEN genitive; HITHER motion towards speaker; HOME localization 'at human Ground's place'; HPL plural human class agreement; IMP imperative; IN localization 'inside hollow Ground'; INF infinitive; INTER localization 'inside mass Ground'; IPF imperfective stem; LAX motion to; M masculine class agreement; N neuter class agreement marker; NEG negative; NMLZ nominalization; NPL plural nonhuman class agreement; OBL nominal oblique stem; OBLPL nominal plural oblique stem; PERSLOC localization 'in space associated with human Ground'; PF perfective stem; PFT perfect; PL plural; POST localization 'behind Ground'; PRG progressive; PROH prohibitive; PRS present; PST past; RES resultative; SG singular; ST detached part of verbal stem in Udi; sub localization 'under Ground'; SUPER localization 'on supporting surface of Ground'; THITHER motion from speaker; UP motion up.
Received 4 February 2009
Revised version received
15 January 2010
Alekseev, Mikhail E. 2003. Sravnitel'no-istorideskaja morfologija naxsko-dagestanskix jazykov. Kategorii imeni [Comparative-historical morphology of Nakh-Daghestanian languages: Nominal categories]. Moscow: Academia.
Alekseev, Mikhail E. & Sabrina Kh. Shikhalieva. 2003. Tabasaranskij jazyk [The Tabasaran language]. Moscow: Academia.
Belien, Maaike. 2002. Force dynamics in static prepositions: Dutch aan, op, and tegen. In Hubert Cuyckens & Gunther Radden (eds.), Perspectives on prepositions, 195-210. Tubingen: Max Niemeyer.
Bowerman, Melissa. 1996. Learning how to structure space for language: A crosslinguistic perspective. In Paul Bloom, Mary A. Peterson, Lynn Nadel & Merrill F. Garrett (eds.), Language and space, 385-436. Cambridge, MA: MIT Press.
Bowerman, Melissa & Soonja Choi. 2001. Shaping meanings for language: Universal and language-specific in the acquisition of spatial semantic categories. In Melissa Bowerman & Stephen C. Levinson (eds.), Language acquisition and conceptual development, 475-511. Cambridge: Cambridge University Press.
Bowerman, Melissa & Eric Pederson. 1992. Crosslinguistic studies of spatial semantic organization. Annual report of the Max Planck Institute for Psycholinguistics 1992, 53-56. Nijmegen: Max Planck Institute for Psycholinguistics.
Brown, Penelope. 1994. The INs and ONs of Tzeltal locative expressions: The semantics of static descriptions of location. Linguistics 32. 743-790.
Brown, Penelope. 2006. A sketch of the grammar of space in Tzeltal. In Stephen C. Levinson & David P. Wilkins (eds.), Grammars of space: Explorations in cognitive diversity, 230-272. Cambridge: Cambridge University Press.
Brugman, Claudia. 1988. The story of over: Polysemy, semantics and the structure of the lexicon. New York: Garland Press.
Choi, Soonja & Melissa Bowerman. 1991. Learning to express motion events in English and Korean: The influence of language-specific lexicalization patterns. Cognition 41. 83-121.
Comrie, Bernard. 1999. Spatial cases in Daghestanian languages. Sprachtypologie und Universalienforschung 52(2). 108-117.
Comrie, Bernard & Maria Polinsky. 1998. The great Daghestanian case hoax. In Anna Siewerska & Jae Jung Song (eds.), Case, typology and grammar, 97-113. Amsterdam & Philadelphia: John Benjamins.
Cuyckens, Hubert. 1991. The semantics of spatial prepositions in Dutch. A cognitive-linguistic exercise. Antwerp: University of Antwerp dissertation.
Cuyckens, Hubert. 1994. Family resemblance in the Dutch spatial preposition op. In Monika Schwarz (ed.), Ergebnisse, Probleme, Perspektiven, 179-195. Tubingen: Narr.
Daniel, Mikhail A. 2001. Padez i lokalizacija [Case and localization]. In Aleksandr E. Kibrik (ed.), Bagvalinskij jazyk: Grammatika. Teksty. Slovari [Bagwalal. Grammar. Texts. Dictionaries], 203-231. Moscow: Nasledie.
Daniel, Mikhail & Dmitry Ganenkov. 2008. Case marking in Daghestanian: Limits of elaboration. In Andrej Malchukov & Andrew Spencer (eds.), The handbook of case, 668-685. Oxford: Oxford University Press.
Ganenkov, Dmitry. 2005a. Kontaktnye lokalizacii v naxsko-dagestanskix jazykax [Contact localizations in Nakh-Daghestanian languages]. Moscow: Moscow State University dissertation.
Ganenkov, Dmitry. 2005b. Prostranstvennye upotreblenija predloga u v slav'anskix jazykax [Spatial uses of the preposition u in Slavic languages]. Die Welt der Slaven L(1). 157-174.
Ganenkov, Dmitry. 2009. Towards a typology of 'attachment' markers: Evidence from East Caucasian languages. In Patience Epps & Alexandre Arkhipov (eds.), New challenges in typology: Transcending the borders and refining the distinctions, 127-148. Berlin & New York: Mouton de Gruyter.
Ganenkov, Dmitry & Yury Lander. Forthcoming. Lokativnye formy kak istocnik nelokativnyx padezej: darginskie dannye [Locative forms as a source for nonlocative cases: evidence from Dargwa]. In Acta Linguistica Petropolitana.
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and cross-Linguistic comparison. In Michael Tomasello (ed.), The new psychology of language, volume 2, 211-242. Mahwah, NJ: Lawrence Erlbaum.
Herskovits, Annette. 1986. Language and spatial cognition: An interdisciplinary study of the prepositions in English. Cambridge: Cambridge University Press.
Khajdakov, Said M. 1999. Lakskij jazyk [The Lak language]. In Mikhail E. Alekseev (ed.), Jazyki mira: Kavkazskie jazyki [Languages of the world: Caucasian languages], 347-357. Moscow: Academia.
Khalilov, Madzhid Sh. 1995. Beztinsko-russkij slovar' [Bezhta-Russian dictionary]. Maxackala: IIJaL.
Kibrik, Aleksandr E. 1970. K tipologii prostranstvennyx znacenij [Towards a typology of locative meanings]. In Jazyk i celovek. Moscow: MGU.
Kibrik, Aleksandr E. 1977. Opyt strukturnogo opisanija arcinskogo jazyka, vol. 2. [Structural description of the Archi language]. Moscow: MGU.
Kibrik, Aleksandr E. 1992. Principy organizacii imennoj paradigmy v dagestanskix jazykax (sopostavitel'no-tipologiceskie nabl'udenija) [Principles of organization of the nominal paradigm in Daghestanian languages (comparative-typological observations)]. In Aleksandr E. Kibrik, Ocerki po obscim i prikladnym voprosam jazvkoznanija [Essays on general and applied issues of linguistics], 80-101. Moscow: MGU.
Kibrik, Aleksandr E. 2003. 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. Berlin & New York: Mouton de Gruyter.
Kibrik, Aleksandr E., Sandro V. Kodzasov, Irina P. Olovjannikova I. P. & Dzhalil S. Samedov. 1977. Arcinskij jazyk. Teksty i slovari [Archi. Texts and dictionaries]. Moscow: MGU.
Kibrik, Aleksandr E. & Yakov G. Testelec. 2004. Bezhta. In Michael Job (ed.), The indigenous languages of the Caucasus, volume 3: The North East Caucasian languages, part 1, 217-298. Ann Arbor: Caravan Books.
Lemmens, Maarten. 2002. The semantic network of Dutch posture verbs. In John Nemwan (ed.), The linguistics of sitting, standing and lying, 103-140. Amsterdam & Philadelphia: John Benjamins.
Levinson, Stephen C. 2003. Space in language and cognition: Explorations in cognitive diversity. Cambridge: Cambridge University Press.
Levinson, Stephen C. 2006. The language of space in Yeli Dnye. In Stephen C. Levinson & David P. Wilkins (eds.), Grammars of space: Explorations in cognitive diversity, 157-205. Cambridge: Cambridge University Press.
Levinson, Stephen C. & Sergio Meira. 2003. 'Natural concepts' in the spatial typological domain --Adpositional meanings in cross-linguistic perspective: An exercise in semantic typology. Language 79(3). 485-516.
Levinson, Stephen C. & David P. Wilkins. 2006. Patterns in the data: towards a semantic typology of spatial description. In Stephen C. Levinson & David P. Wilkins (eds.), Grammars of space: Explorations in cognitive diversity, 512-552. Cambridge: Cambridge University Press.
Levinson, Stephen C. & David P. Wilkins (eds.). 2006. Grammars of space: Explorations in cognitive diversity. Cambridge: Cambridge University Press.
Magomedova, Patimat T. 1979. Osnovnye voprosy sklonenija v andijskix jazykax v sravnitel'nom aspekte [Basic issues of declension in Andic languages in comparative perspective]. In Unejzat A. Mejlanova (ed.), Imennoe sklonenije v dagestankix jazykax [Nominal declension in Daghestanian languages], 41-69. Maxackala.
Magomedova, Patimat T. & Rashidat Sh. Khalidova. 2001. Karatinsko-russkij slovar' [Karata-Russian dictionary]. Maxackala: IIJaL.
Magomedova, Patimat T. & Indira A. Abdulaeva. 2007. Axvaxsko-russkij slovar' [Akhvakh-Russian dictionary]. Maxackala: IIJaL.
Makhmudova, Svetlana M. 2001. Morfologija rutul'skogo jazyka [Morphology of the Rutual language]. Moscow: Institut jazykoznanija.
Mal'ar, Tatjana N. & Olga N. Seliverstova. 1998. Prostranstvenno-distancionnye predlogi i narecija v russkom i anglijskom jazykax [Space-distance prepositions and adverbs in Russian and English]. Munich: Otto Sagner.
Meira, Sergio. 2006. Approaching space in Tiriyo grammar. In Stephen C. Levinson & David P. Wilkins (eds.), Grammars of space: Explorations in cognitive diversity, 311-358. Cambridge: Cambridge University Press.
Nuse, Ralf. 1999. General meanings for German an, auf, in and unter towards a (neo)classical semantics of topological prepositions. Berlin: Humboldt University dissertation.
Plungian, Vladimir A. 2002. O specifike vyrazenija imennyx prostranstvennyx xarakteristik v glagole: kategorija glagol'noj orientacii [On specifics of expression of nominal spatial features in verb: the category of verbal orientation]. In Vladimir A. Plungian (ed.), Grammatikalizacija prostranstvennyx znacenij [Grammaticalization of spatial meanings], 57-98. Moscow: Russkie slovari.
van Staden, Miriam, Melissa Bowerman & Mariet Verhelst. 2006. Some properties of spatial description in Dutch. In Stephen C. Levinson & David P. Wilkins (eds.), Grammars of space: Explorations in cognitive diversity, 475-511. Cambridge: Cambridge University Press.
Sumbatova, Nina R. & Rasul O. Mutalov. 2003. A grammar of Icari Dargwa. Munich & Newcastle: Lincom.
Svorou, Soteria. 1994. The grammar of space. Amsterdam: John Benjamins.
Talibov, Bukar B. & Magomed Gadzhiev. 1966. Lezginsko-russkij slovar' [Lezgian-Russian dictionary]. Moscow: Sovetskaja Enciklopedija.
Talmy, Leonard. 2000. How language structures space. In Leonard Talmy. Toward a cognitive semantics, volume I: Concept structuring systems, 177-254. Cambridge, MA: MIT Press.
Tyler, Andrea & Vyvyan Evans. 2003. The semantics of English prepositions: Spatial scenes, embodied meaning and cognition. Cambridge: Cambridge University Press.
Vandeloise, Claude. 1991. Spatial prepositions: A case study in French. Chicago: University of Chicago Press.
Wilkins, David P. & Nicholas Evans. 1995. 'Inside' and 'down' in Australian languages. In Henriette Hendriks & James McQueen (eds.), Max Planck Institute for Psycholinguistics Annual Report 16, 99-102. Nijmegen: Max Planck Institute for Psycholinguistics.
Zhirkov, Lev I. 1955. Lakskij jazyk. Fonetika i morfologija [The Lak language: Phonetics and morphology]. Moscow: AN SSSR.
* I am very grateful to Sander Lestrade and two anonymous reviewers of Linguistics for very valuable comments and suggestions, which helped to improve this paper. All remaining deficiencies and shortcomings naturally remain my own. This work is supported by the Russian Fund for Humanities (grant #09-04-00332a). Correspondence address: Department of Caucasian Languages, Institute of Linguistics, Bolshoy Kislovsky 1, 125009 Moscow, Russia. E-mail: firstname.lastname@example.org.
(1.) Daghestanian languages include four (of five) branches of the East Caucasian family and are spoken in Daghestan and Azerbaijan. The Avar-Andic-Tsezic branch occupies the whole of the western part of Daghestan and is divided into Avar with a number of dialects, the Andic group, to which Akhvakh, Andi, Botlikh, Chamalal, Godoberi, Karata, Bagwalal, and Tindi belong, and the Tsezic group, including Tsez, Hinukh, Khwarshi, Bezhta, and Hunzib. Lako-Dargian is located in the center and is divided into Lak and Dargwa, with the latter further dividing into a number of very divergent dialects, some of which can he considered as separate languages. The languages of the Lezgic branch are mostly spoken in the southern part of Daghestan (Lezgian, Agul, Tabasaran, Rutul, Tsakhur) and in the northern part of Azerbaijan (Udi, Kryz, Budukh, as well as important Lezgian, Rutul, and Tsakhur communities), with the exception of Archi, which is spoken in Central Daghestan. Another language spoken in Azerbaijan is Khinalug, which constitutes a separate branch within the family.
(2.) Language data used in this paper were mostly collected from native speakers on the basis of a questionnaire that contained Russian sentences describing various spatial ON-configurations as well as from published texts, grammars, and dictionaries of the East Caucasian languages. The questionnaire was first compiled using spatial configurations presented in Bowerman and Pederson (1992); Bowerman (1996); Bowerman and Choi (2001); and Levinson and Meira (2003) and then modified and supplemented with additional configurations in the course of the fieldwork. If an example from a published source is given, an explicit reference is made in the text. The absence of a reference indicates that the example originates from my own fieldwork.
(3.) Detailed descriptions of Daghestanian case systems can be found in Kibrik (1992); Comrie and Polinsky (1998); Comrie (1999); Kibrik (2003), and Daniel and Ganenkov (2008).
(4.) In the late 1960s, A. E. Kibrik coined Latin-based labels for localizations, which have been widely used in Daghestanian studies ever since. The standard list of localizations includes: APUD 'near Ground', SUPER 'on Ground', IN 'in Ground', INTER 'between two Grounds / in a mass', CONT 'attached to Ground', SUB 'under Ground', ANTE 'in front of Ground', POST 'behind Ground'.
(5.) To avoid confusion, I use the term location to refer to a spatial disposition of Figure, whereas the term localization is reserved for the language-specific nominal category. It should be noted that some Daghestanian languages also include nonspatial relations in the system of localization markers. For example, the Bezhta system includes the possessive localization, which is regularly combined with the essive ('X has'), lative ('give to X') and elative ('take from X') orientation values. Obviously, this is a result of diachronic evolution of the localization marker, which lost its spatial meaning but retained nonspatial uses. A detailed discussion of such cases is beyond the scope of this paper. Some relevant observations on the evolution of locative markers in Dargwa dialects can be found in Ganenkov and Lander (forthcoming).
(6.) Arabic numbers in interlinear glosses of Archi examples indicate noun classes (gender). See also Examples (2), (13), and (15).
(7.) A more detailed account of the marking of location on a surface in Daghestanian languages can be found in (Ganenkov 2005a, 2009).
(8.) Uses in (b) and (c) typically occur with (body) parts that can be informally characterized as "hanging down" from the whole, such as finger, ear, nose, tail, handle etc.
(9.) Recall that the term SUPER is applied to localization markers expressing at least (i), while the term CONT refers to markers denoting at least (ii).
(10.) See Haspelmath (2003) on the semantic maps as a method for drawing typological generalizations from crosslinguistic comparison.
(11.) Detailed accounts of the semantics of these spatial prepositions can be found in Cuyckens (1991, 1994); Belien (2002); Nuse ( 1999); van Staden et al. (2006).
(12.) Note that in all the three East Lezgic languages the primary meaning of the localization marker -q is location behind Ground (see Section 5.4).
(13.) Crosslinguistically, this contrast is not confined to Daghestanian languages. Although it is not found in most European languages, several Slavic languages make the distinction, cf. the prepositions u and okolo in Russian. For more detailed discussion see Mal'ar and Seliverstova (1998) and Ganenkov (2005b).
(14.) An anonymous reviewer points out that the function of the marker -c' is 'very close to, exactly at', rather than 'at the foot of'. However, my data show that this marker is not possible in sentences like 'I am sitting by the window' which suggests that the characterization proposed by the reviewer cannot be applied to my data. This divergence may be due to dialectal or idiolectal variation and requires further investigation.
(15.) Another possibility attested in Daghestanian languages is to use a construction with the structure like 'at X's side', 'at X's there', 'at X's'.
Table 1. Examples of locative forms in Agul and Xuduc Dargwa (oblique) localization orientation nominal stem Agul dar-a- 1- as tree-OBL- SUPER- ELAT '(to fall down) from the tree' Xuduc Dargwa k:ark:a-l- s:u- rka tree-OBL- APUD- ELAT '(to move away) from the tree' Table 2. Zero-marked localizations in Agul and Xuduc Dargwa Agul Xuduc Dargwa 'in front of dar-a-h-[empty set] k:ark:a-l-sa-[empty set] tree-OBL-ANTE-ESS tree-OBL-ANTE-LAT '(to stand) in front '(to move to) in front of the tree' of the tree' 'under' dar-a-k:-[empty set] k:ark:a-l-gu-[empty set] tree-OBL-SUB-ESS tree-OBL-SUB-LAT '(to find smth.) '(to put smth.) under the tree' under the tree' Table 3. Examples of locative forms in Qungi Dargwa (oblique) nominal stem localization orientation deixis/ gravity k.ark:a-l- s:a- r- ca tree-OBL- APUD- ELAT- HITHER 'away from the tree towards the speaker' k:ark:a-l- s:a- r- de tree-OBL- APUD- ELAT- THITHER 'away from the tree from the speaker' k:ark:a-l- s:a- r- ha tree-OBL- APUD- ELAT- UP 'up from the tree' k:ark:a-l- s:a- r- ka tree-OBL- APUD- ELAT- DOWN 'down from the tree' Table 4. The IN and INTER forms of 'sin 'water' and sumka 'bag' in Mugi Datgwa IN INTER sin 'water' * sin-ne-ce [.sup.OK]sin-ni-zi water-OBL-IN water-OBL-INTER sumka 'bag' [sup.OK]Ksumka-le-ce * sumka-li-zi bag-OBL-IN bag-OBL-INTER Table 6. Three conflation patterns in Daghestanian SUPER IN CONT INTER SUB Rutul -a/i -a/-e/-i -k -xde Lezgian -l -da -k Xuduc Dargwa -le -c:i -gu
|Printer friendly Cite/link Email Feedback|
|Publication:||Linguistics: an interdisciplinary journal of the language sciences|
|Date:||Sep 1, 2010|
|Previous Article:||A hierarchy of locations: evidence from the encoding of direction in adpositions and cases.|
|Next Article:||The syntactic structure of locations, goals, and sources.|