Grammaticalization of new cases in Beserman Udmurt.
The emergence of new case markers in the grammatical system of a language is the outcome of a diachronic process whereby lexical items gradually lose their independence, eventually transforming into grammatical formatives. According to Lehmann (2002: vii), such a process, called grammaticalization, is governed by the same cognitive mechanisms in different languages, and thus there are grammaticalization paths which are common to many languages. One of such common paths involves first transforming relational nouns into adpositions that can be inflected for local cases only and then transforming such adpositions into cases (Creissels 2008).
The Permic languages are known to have elaborate spatial systems with active grammaticalization processes going on ([TEXT NOT REPRODUCIBLE IN ASCII] 1979: 106-110), including the one mentioned above. In the second stage of this process, new cases derive from the locative forms of adpositions (postpositions in the case of Permic languages), i.e. [N Postp-Cx[Loc.sub.1]] [right arrow] [N-Cx[Loc.sub.2]]; see Korhonen (1996: 233) for typical scenarios of the formation of Uralic cases. The inventory of new cases thus acquired can be considerable. According to Batalova, in several Komi Permyak dialects the local case forms of the postpositions vil- 'surface' and din- 'vicinity' have grammaticalized into about ten cases ([TEXT NOT REPRODUCIBLE IN ASCII] 1975: 136-141). Baker (1985), however, considers Batalova's claim to be not quite correct, as there are only two din-cases which occur very rarely in Southern Permyak speech and are absent in Northern Permyak (Baker 1985: 196). As for the vil-cases, Baker s analysis of several Permyak corpora, including Uotila's (Baker 1985: 175-181), revealed that they also do not exist in Northern Permyak, and in the Inva variety of Southern Permyak they frequently co-occur with full and reduced postpositional forms. Hence, the new vil-cases have just begun to replace the postpositional constructions they have grammaticalized from, and this process is active only in certain Southern Permyak dialects (and even there it will be possibly blocked by the standard language), while the new din-cases only appeared recently and are largely peripheral. Nevertheless, as far as we know, no one doubts the fact that in the recent 300 years Southern Permyak dialects have acquired several cases which originate from spatial postpositions.
The aim of our paper is to investigate a particular implementation of the grammaticalization path that transforms postpositional constructions into case forms by analyzing the properties of some new cases that have developed in the Beserman dialect of Udmurt. The cases in question are the recessive and a series of so-called approximate-local cases, which do not exist in the literary Udmurt language.
The starting point of the process we are investigating are relational (relator) nouns, which are widespread in Permic languages (often called inflected postpositions in grammars). Relational nouns are those which function as heads in spatial NPs with nominal dependents (Starosta 1985; DeLancey 1997). They denote the first term of relation and have a valency which must be filled at the semantic level ([TEXT NOT REPRODUCIBLE IN ASCII] 1998: 170-171). As Kahr (1975) notes, in expressions with spatial meanings two types of relational nouns are used:
1. Names of body parts, which at first begin to designate parts of physical objects and then become localization grammemes;
2. Names of geometrical parts of physical objects (like 'side', 'part', 'edge') grammaticalizing into designations of the basic localizations.
In (i), (1) the relational name urdes 'flank' illustrates the subtype 1, pal 'side', the subtype 2:
(i)c So-len urdesk-a-z, bur pal-a-z sdl-e petuk
that-GEN flank.OBL-LOC-P.3(SG) right side-LOC-P.3(SG) stand-PRS.3SG rooster
'On his side, on the right a rooster stands'
The data presented here was collected during our fieldwork in the Beserman village of Samardan (Jukamenskoje district, Udmurtia, Russia) in 2010-2016. Beserman is one of the Udmurt dialects spoken by a relatively small eponymous ethnic group occupying the basin of the Cepca river. According to the 2010 population census, there are 2201 people identifying themselves as Beserman. The morphological systems of Udmurt dialects differ from each other quite considerably (see, e.g., [TEXT NOT REPRODUCIBLE IN ASCII] 1969; [TEXT NOT REPRODUCIBLE IN ASCII] 1996; [TEXT NOT REPRODUCIBLE IN ASCII] 1998b), while many of the differences concern the structure and interactions of case systems and the systems of postpositions. These interactions are quite complex, because the spatial systems of many dialects are undergoing or have recently undergone serious changes. In some dialects, these changes result in a displacement of cases in favour of postpositional constructions, while in others they result in the appearance of new cases through grammaticalization. The first group is represented by Southern Udmurt dialects: in many of them the approximative and the terminative cases have been or are being replaced by PPs ([TEXT NOT REPRODUCIBLE IN ASCII] 1998a: 117-118), apparently due to their low frequency. This is also true for the approximative in standard Udmurt (see 3.3). The Beserman dialect, on the contrary, has developed a series of approximate-local cases, which are used with personal names and nouns denoting persons (pi-jezn-3S son-P.3SG-DMS-EL 'from (the house of) his son' ([TEXT NOT REPRODUCIBLE IN ASCII] 2010: 6)) and a new recessive case denoting 'from X's side'. The presence of the latter in Beserman has never been mentioned in the literature although forms with a recessive suffix have been attested by researchers in Beserman as well as in several other Udmurt dialects (Jelena Popova, p. c.). The former has been attested in some Udmurt dialects at least since Jemeljanov's grammar ([TEXT NOT REPRODUCIBLE IN ASCII] 1927: 127), where an approximate-illative case in -ne is listed as a variant of the illative for the Glazov district Udmurt variety.
The new cases have developed from postpositional constructions (or from constructions formed by relational nouns) and still have some postposition-like (on even noun-like) properties together with the case-like ones. We show that the cases in question, although they definitely cannot be analyzed as postpositions anymore, still have not reached a full degree of grammaticalization. They differ from more grammaticalized cases in the values of various cross-linguistically relevant grammaticalization parameters. Some of those differences have been shown to be relevant for other Uralic languages (Bielecki 2012, namely lexical associativity and linear separability).
2. Ongoing grammaticalization of the new cases
2.1. Approximate-local cases
The so-called approximate-local cases denote location in one's domicile, movement into one's domicile, from it, in it or to it, as in examples (1) below. Thus, from the point of view of semantics they compositionally combine the meaning 'domicile' with those of the local cases. These cases have also been attested in other Udmurt dialects ([TEXT NOT REPRODUCIBLE IN ASCII] 1981; [TEXT NOT REPRODUCIBLE IN ASCII] 2005: 60- 63), though not in the literary language. In Hungarian there is a similar situation. Although the case inventory of literary Hungarian is quite large, it does not contain any approximate-local cases. In some Hungarian dialects, however, one can find cases such as -nott/-nott 'at', -nol/-nol 'from', -ni, -nyi 'to', which are used only with nouns denoting humans and mean location at, or motion to/from someone's place (Imre 1971: 317-318).
Morphologically, the affixes of the approximate-local cases in Beserman are compositional, consisting of the suffix -n- and the suffixes of local cases. Compare:
[(1).sub.f] a. lud-Sn abi-n-rn field-LOC grandmother-DMS-LOC 'on the field at grandmother's'
b. lud-e abi-n-e field-iLL grandmother-DMS-iLL 'into the field to grandmother's'
c. lud-ds abi-n-[section]s field-EL grandmother-DMS-EL 'from the field from grandmother's'
d. lud-t'i abi-os-n-St'i field-PROL grandmother-PL-DMS-PROL 'through the field; on/to different parts of the field to grandmothers' houses'
e. lud-isen abi-n-isen field-EGR grandmother-DMS-EGR 'from the field from grandmother's'
f. lud-oz abi-n-oz field-TERM grandmother-DMS-TERM 'to (the boundary of) the field (just) to grandmother's house'
It can be seen from (1) that nearly everywhere the local case suffixes in approximate-local forms have the same morphological forms as they do in spatial forms of ordinary nouns. The only difference can be seen in the prolative case, which has an additional prothetic vowel e. However, an independent prolative marker et'i also exists in the dialect, appearing in a small number of cases under certain morphophonological conditions (e. g. in the oblique-stem form of the relational noun pus 'inside', pusk-et'i, as opposed to its bare-stem form, pus-t'i). The functions performed by the case markers in these combinations are the same as when used independently. These case markers denote certain combinations of localizations and orientations, while the suffix -n- compositionally adds to the meaning of the form by denoting the localization DOMUS 'domicile'. For these reasons, we prefer to speak of a Beserman localization suffix, -n-, which has a meaning 'domicile' and can be combined with suffixes of several local cases. Such typologically oriented interpretation simplifies the description of the case system.
The number of the approximate-local cases attested in the Beserman dialect grows from earlier grammatical descriptions to more recent ones. Thus, Tepljasina ([TEXT NOT REPRODUCIBLE IN ASCII] 1970) includes only one such case with the suffix -mn, i.e. personal locative, in her description of the noun declension paradigms; in the appendix to her book, however, one can also find examples of nouns with the suffix -ne (in Tepljasina's terms, one should have labeled this case as personal illative). Four approximate-local cases are mentioned by Kelmakov ([TEXT NOT REPRODUCIBLE IN ASCII] 1998a) and six by Ljukina ([TEXT NOT REPRODUCIBLE IN ASCII] 2008). According to our fieldwork data, four forms were used in Samardan in 2010, namely those which are combinations of the suffix -n with illative
(1), locative (1), egressive (2) and elative (3) suffixes:
[(2).sub.c] Men-i-z, men-i-z, per-i-z podruga-ez-n-e, go-PST-3(SG) go-PST-3(SG) enter-PST-3(SG) friend-P.3(SG)-DMS-ILL es-ez-n-e friend-P.3(SG)-DMS-ILL 'She was walking and walking and came to her friend s (on a visit)'
[(3).sub.f] Masa-n-en todma-c-k-i-m, Masa-n-isen Masa-DMS-LOC get.to.know-CAUS-DETR-PST-1PL Masa-DMS-EGR cos kosk-i-m=ni together leave-PST-1PL=already 'We met at Masa's [place], and we left her [house] together'
[(4).sub.f] Kin-len sestra-ez-n-es ldkt-isko-d? who-GEN sister-P.3(SG)-DMS-EL come-PRS-2(SG) 'From whose sister are you coming?'
In 2014, we attested two more forms, involving -n plus prolative and terminative suffixes. Note that in 2010 all the speakers we worked with rejected such forms and used constructions with the relational noun (2) dor 'neighbourhood' instead:
[(5).sub.f] So jalan gurta-sa vel't-e sestra-os-n-dt'i-z that always be.on.visit-CVB walk-PRS.3SG sister-PL-DMS-PROL-P.3(SG) 'He/she always visits his/her sisters'
[(6).sub.f] Kin-len brat-ez-n-oz so vu-i-z? who-GEN brother-P.3(SG)-DMS-TERM that reach-PST-3(SG) 'Whose brother's house has he/she reached?'
By recessive we understand a case which means 'direction of the Figure from the side of the Ground', thus being semantically opposite of the somewhat more widespread approximative, or directive case ('direction of the Figure towards the Ground'). It differs from other elative cases in that the precise source of movement is unknown or not specified, and the Ground is used only to approximately mark the side where the source of the movement is located. Such cases are not typologically widespread and sometimes do not reach full grammaticalization. For example, in Ossetic the recessive and the directive are not fully grammaticalized and are perceived as postpositions rather than cases by traditional grammarians, although there are arguments for treating them as case markers (Belyaev 2010).
An approximative case with the marker -lan exists both in the Beserman dialect and, at least according to grammars, in literary Udmurt (see more on that in 3.3). On the other hand, a case with a recessive meaning, while existing in the Beserman dialect, has never been attested in the literary language (except for a single dubious example, see 3.3). This case is expressed by the marker -lasen and approximately corresponds to the literary Udmurt recessive postposition lasan in terms of semantics ([TEXT NOT REPRODUCIBLE IN ASCII] 390).
The Beserman recessive has spatial (7) and non-spatial meanings, while the latter are restricted to kinship terms (8):
[(7).sub.f] abi-je-len korka-ez-lasen skal-jos lekt-o grandmother-P.1SG-GEN house-P.3(SG)-RCS cow-PL come-PRS.3PL 'There are cows coming from my grandmother house's side'
[(8).sub.f] muzik-e-lasen zazaj-os-ez bijagam su-el-i husband-P.1SG-RCS uncle-PL-P.3(SG) uncle-in-law call-ITER-PST(1sg) 'I called the uncles on my husband's side bijagam'
In a corpus of experimental texts recorded and transcribed in 2015 there is a considerable number of utterances where the recessive forms compete with the construction formed by the relational noun pal 'side'. The egressive form of the latter may well be the diachronic source of the recessive case. Arguments in favour of this hypothesis include their very similar distribution and a rare, non-productive declination type of pal (it normally has the egressive in -asen, instead of the productive marker -Isen). The difference in form between the Beserman recessive marker and the corresponding Udmurt postposition lasan 'from the side' can then be explained by the dialectal variation of the egressive marker, which has the form -(i)sen in some Udmurt and Komi dialects and -san in others (including literary Komi Zyrian and Komi Permyak). The variation between the recessive and the forms of pal can be seen in (9) and (10). Each of the examples features two utterances taken from a dialogue, one of the utterances containing the case marker, and the other having an equivalent construction with pal:
[(9a).sub.c] [--] i selt-e masina dor-a-z [--] and stand-PRS.3SG car neighbourhood-LOc/iLL-P.3(SG) kud pal-asen puks-o sofor-jos, so pal-a-z which side-EGR sit-PRS.3PL driver-PL that side-LOc/iLL-P.3(SG) '[--] and stands near the car, at the side where the drivers sit'
[(9b).sub.c] kud-lasen kdce masina lekt-e menem ton which-RCS which car come-PRS.3SG I.dat you(SG).NOM wal'l'o vera-d? earlier tell-PST.2(SG) 'Which car returns from which side, have you told me?'
[(10a).sub.c] bur-lasen, pal'l'an-lasen? right-RCS left-RCS 'From the right side or from the left side?'
[(10b).sub.c] a bur pal-as en lSt-e pios and right side-EGR come-PRS.3SG fellow 'And from the right side the fellow comes'
During the experiment, the speakers were presented with referential communication tasks in which they had to give a detailed description of the movement of several figures across the model of the Samardan area. A similarly designed experiment with the same speakers was carried out in 2010, (3) and the texts recorded and transcribed during it did not contain any recessive forms of nouns. In the contexts typical for the recessive, only constructions with the egressive forms of pal 'side' were used:
[(11).sub.c] a muket pala-sen, muket pal-as en and different side-EGR different side-EGR sur ser-en abi ul-e river across-LOC old.woman live-PRS.3SG 'And on the other side, on the other side, across the river, lives an old woman'
[(12).sub.c] Speaker 1: vot odig-ez es valla pal-a-z, here one-P.3(SG) door upper side-LOC/ILL-P.3(SG) es vel-a-z, es vella pal-a-z=ges door top-LOC/ILL-P.3(SG) door upper side-LOC/ILL-P.3(SG)=CMPR 'One of them is above the door, just a little higher above the door'
Speaker 2: pallan pal-isen, ben=a? left side-EGR yes=Q 'On the left side, yes?'
This difference indicates that the grammaticalization process in the case of the Beserman recessive is quite active.
Cross-linguistically, newly grammaticalized units are usually characterized by a conglomerate of heterogeneous properties, which are partly new and partly inherited from their previous stage on their grammaticalization path. This leads to the current situation in linguistic literature where the same unit in the same idiom is labeled as "case suffix" in one work, as "postposition" in another, as "clitic" in yet another etc.; the same goes for postpositions, relators and nouns. This state of affairs is characteristic for items undergoing grammaticalization and is very vividly characterized by Dryer (2013). It can become a real problem for a typologist who wants to compare units with the same status in different languages. In fact, the present literature demonstrates two different ways of avoiding the problem (it is unlikely to be solved as it goes back to the fundamental collision between the discreteness of terminology and the fuzziness of a constantly changing natural language). The first way is to unite local case suffixes, postpositions and relators into one category, as is done, for example, by Svenonius (2006) and Ashbury (2008), or Spencer (2008) for Hungarian; "flag" is sometimes used as a cover term subsuming both cases and adpositions. However, this approach is debated, at least for certain languages (see Thuilier 2011 for counter-arguments from Hungarian). Alternatively, one can describe the properties of the unit in question and then decide to which class it should be attributed. This approach is chosen by many authors describing cases and adpositions in different languages. It was employed by, among others, Itkin ([TEXT NOT REPRODUCIBLE IN ASCII] 2002) for Veps, Belyaev (2010) for Ossetic, Thuilier (2011) for Hungarian, Grenoble (2014) for Evenki, Jadhav (2014) for Marathi. We also prefer this approach and will now discuss the properties of the newly grammaticalized Beserman cases in order to decide in which morphological class of units they should be included.
3.1. Approximate-local cases
3.1.1. Properties typical for core cases
Forms with the suffix -n- have most of the properties of fully grammaticalized, undoubted markers of nominal inflection.
1. They cannot be attached to a conjoined phrase (13), and nor can most of the fully grammaticalized cases, such as illative, for example (14):
[(13a).sub.f] Menem kat Vana-n-e=no Volod'a-n-e men-ono I.DAT now Vanja-DMS-ILL=ADD Volodja-DMS-ILL go-DEB 'Now I have to go to Vanja's and to Volodja's'
[(13b).sub.f] *Menem kai' [Vana = no Volod'a]-n-e man-ono I.dat now Vanja=ADD Volodja-DMS-ILL go-DEB *'Now I have to go to Vanja's and to Volodja's'
[(14a).sub.f] kolodca-jan=no vedra-jan vu-ez eval well-LOC=ADD bucket-LOC water-P.3(sG) be.NEG 'There is no water either in the well or in the bucket'
[(14b).sub.f] *[kolodca=no vedra]-jan vu-ez eval well=ADD bucket-LOC water-P.3(sG) be.NEG * 'There is no water neither in the well nor in the bucket'
However, it should be noted that there are three case markers which sometimes can be attached to a conjoined phrase: approximative, instrumental and recessive (see 3.2).
Note that the situation with postpositions and relational nouns is alike: some of them can take a conjoined phrase as a dependent, some cannot, and for most of them there is a great variation in acceptability between speakers.
2. Not only nouns, but also personal (15) and reflexive (16) pronouns have approximate-local forms:
[(15).sub.f] so-os-n-as that-PL-DMS-ILL 'from their house'
[(16).sub.f] as-n-a-z as-s-a-n-e REFL-DMS-LOC/ILL-p.3(SG) REFL-3-PL-DMS-ILL 'to/in his/her house to his/her house'
This is an evidence of a relatively high degree of grammaticalization but it does not distinguish n-forms from relational nouns as the latter can also govern personal pronouns and reflexives.
3. The group "-n- + local case suffix (+ possessive suffix, if present)" cannot be detached from the noun it refers to. Compare (5) with the following ungrammatical example (17), where they were detached from the noun:
[(17).sub.f] *So jalan sestra-os(-len) gurta-sa that always sister-PL(-GEN) be.on.visit-CVB vel't-e n-a t'i-z walk-PRS.3sG DMS-PROL-p.3(SG) * 'He/she always visits his/her sisters'
This property proves that -n-forms cannot be analyzed as NPs headed by relational nouns because the latter can be detached from their dependents (18), the dependents being turned into external possessors (or omitted altogether):
[(18).sub.c] Ot-en pene so-os-len wan dor-a-z-e there-LOC dog that-PL-GEN be.PRS neighbourhood-LOC-P.3-PL 'There is a dog near them'
4. We attested different subtypes of case compounding (4) phenomena in Beserman, among them, double case marking, where case suffixes attach to a stem which already contains a genitive marker. Such kind of compounding is cross-linguistically widespread, because the genitive marks a different kind of relation (Dixon 2010: 45); see an overview of compounding in Uralic languages in (Bartens 2003). Such forms can be used when the head of the NP undergoes ellipsis: (5)
[(19a).sub.f] (kin-len korka dor-oza-z mat-en mene-ne?) who-GEN house neighbourhood-TERM-P.3(SG) close-LOC go-INF Vasa-len-oza-zmat-en Vasja-GEN-TERM-P.3(SG) close-LOC '(To whose house is it closer to go?) To Vasja's house [it is] close'
Some speakers allow the suffix -n- together with the following local case markers also to be compounded with the genitive (although such forms normally do not occur in speech due to their excessive complexity):
[(19b).sub.f] kud-iz-len sestra-n-en Masa ul-e kat, which-P.3(SG)-GEN sister-DMS-LOC Masa live-PRS.3SG now Zoja-ez-len-n-en = a Ol'a-ez-len-n-en = a? Zoja-P.3(SG)-GEN-DMS-LOC=Q Olja-P.3(SG)-GEN-DMS-LOC=Q 'In the house of which sister is Masa living now--at Zoja's sister or at Olja's sister?'
3.1.2. Signs of incomplete grammaticalization
As can be seen from these examples, the suffix -n- 'DOMUS' has certain important traits of quite a well grammaticalized suffix. However, the -n-forms have several properties of a marker which has not yet finished the process of grammaticalization:
1. Its semantics is pretty straightforward and has no metaphorical ramifications. Grammaticalization is often accompanied by case markers acquiring additional meanings due to metaphor; all other cases, even the recessive, have more than one meaning. Besides, the combinations of -n- with the spatial cases are purely semantically compositional, whereas a longer grammaticalization history would probably result in a partial loss of compositionality.
2. The suffix attaches only to pronouns (15, 16; see 3.1.3 below for personal pronouns) and nouns belonging to certain semantic classes: nouns denoting persons (2), personal names (3), and kinship terms (4-6).
3. The degree of grammaticalization is different for different combinations of -n- with local case suffixes, which is probably the reason why some of them were not mentioned in earlier works. The inessive and illative forms are the most productive ones, while the terminative and prolative forms occur rarely and are not accepted by all speakers. Moreover, the speakers rejected personal terminative and personal prolative forms of the 1st and 2nd person pronouns. Such uneven degree of grammaticalization can be explained by the fact that it is an ongoing process whereby more frequent members of the series grammaticalize faster than the rest, since frequency is a primary contributor to this process (Bybee 2007: 336). It is also cross-linguistically common for personal pronouns to have a somewhat different case inventory than nouns, which can be explained by the conserving effect of frequency (forms that are used very often tend to be more resilient to diachronic change). As a means of expressing the DOMUS localization with nouns ('in X's home'), the personal cases gradually force out the construction with the locative forms of the relational noun dor 'neighbourhood', which can be seen in the following examples:
[(20).sub.c] Ot-ce Fel'ik s dor-e =gine kel't-i-z that-ILL Felix neighbourhood-ILL=only leave-PST-3(SG) so pici masina-z-e that little car-P.3(SG)-ACC 'He only left his passenger car at Felix's'
[(21).sub.c] Mama-os-ed dor-oz tab-e, mom-PL-P.2(SG) neighbourhood-TERM go.up-PRS.3SG i so, sel-e, ug men-e and that's.all stand-PRS.3SG NEG go-SG '[The horse] only reached your mom's house, and that's all, it has stopped, it is not moving'
In general, the forms with the suffix -n- have not yet reached a full degree of grammaticalization, but -n- is quite close to fully grammaticalized markers.
3.1.3. Remarks on pronominal forms and morphotactics
The n-cases have several peculiar properties which do not seem to explicitly indicate their position on the grammaticalization path, but still set them apart from other cases.
In the Beserman dialect, just as in literary Udmurt, the mutual order of the case and possession markers in nouns is variable. As the order is fixed for every given case, different cases require different orders: e.g. the locative marker precedes the possessive markers (as in 18), while the genitive marker requires the opposite order. We do not have enough data to fully describe the ordering rules for all combinations containing the suffix -n-, however, it is obvious that the case markers following it are sometimes ordered differently than in the forms where they appear independently. In all our examples of personal locative (-n-en), illative (-n-e), elative (-n-es) and terminative (-n-oz) forms of nouns, the possessive markers precede the case markers, and at least for the first three of these cases we know this is the only possible ordering. For instance, the grammatical sentence (2) cannot be transformed into (22) below. This would not be the case in the forms where these cases appear independently.
[(22).sub.f] *Men-i-z, men-i-z, per-i-z podruga-n-a-z, go-PST-3(SG) go-PST-3(SG) enter-PST-3(SG) friend-DMS-ILL-P.3(SG) e s-n-a-z friend-DMS-ILL-P.3(SG) * 'She was walking and walking and came to her friend's (on a visit)'
On the other hand, the rules are evidently more complex than just "always put possessive markers before -n-", since at least in the personal prolative the possessive marker is allowed to appear to the right of the case marker (5), just as it happens with an independent prolative marker. Besides, the Case-Possessiveness order is acceptable for the P.3SG illative/locative form of the reflexive pronoun (16a). Given that the form with the 3pl possessor has the opposite order (16b), the approximate-local cases, unlike with nouns, appear to behave just like their independent local counterparts with the reflexive pronouns.
Another property worth noting concerns 1st and 2nd person pronouns, which have irregular declension. While -n- forms of 1sg and 2sg pronouns appear in [TEXT NOT REPRODUCIBLE IN ASCII] 1970 for the personal illative in Beserman (and in [TEXT NOT REPRODUCIBLE IN ASCII] 1981) for all six approximate-local cases in other Northern Udmurt varieties), these seem only marginally, if at all, acceptable in Samardan and never occur in speech. Instead of singular forms, like ton-n-a-d you(SG)-DMS-LOC/ILL-2, the speakers in Samardan always use plural forms, both in conversation, as in (23), and when asked to translate a Russian stimulus containing a pronoun in the singular:
(23)f So t in -n-a-d kartoska-d-e that you(PL).OBL-DMS-LOC/ILL-P.2 potato-P.2(SG)-ACC puk-t-on-nig-a-z zol zad'-i-z sit-CAUS-NMLZ-NLOC-LOC/ILL-P.3(SG) very get.tired-PST-3(SG) 'He got very tired while planting potatoes at yours'
The -n-cases generally tend to be used with the associative plural, like in (21), much more often than any other case; out of the 22 nominal examples in our corpus, 8 are in plural. However, with personal pronouns this tendency turns into an absolute rule. Since the grammatical number of the pronouns in -n-cases does not influence grammatical forms elsewhere in the sentence and thus can be determined only on morphological and semantic grounds, it would be probably more correct to treat these forms as unspecified for number rather than as plural.
3.2.2. Relational noun-like properties
The recessive case has several properties distinguishing it from more grammaticalized cases; most of these properties also characterize NPs headed by relational nouns. Most other cases do not exhibit them. However, it turns out that the semantic counterpart of the recessive, the approximative case with the marker -lan, despite its long history as part of the nominal paradigm, shares some of these properties to certain extent. 1. Some speakers prohibit the recessive forms of 1SG and 2SG pronouns, which, as was mentioned earlier, have complicated paradigms. Those speakers who admit such forms as grammatical, allow attaching the suffix -lasen to either the nominative or the genitive form of the pronominal stem (different speakers prefer different stems):
[(24).sub.f] mon-lasen/menam-lasen, ton-lasen/tenad-lasen I.NOM-RCS I.GEN-RCS you(SG).NOM-RCS you(SG).GEN-RCS
This variation is also attested for the approximative (lan), at least for some speakers:
[(25).sub.f] mon-lan/menam-lan, ton-lan/tenad-lan I.NOM-APPR I.GEN-APPR you(sG).NOM-APPR you(sG).GEN-APPR
All relational nouns, unlike ordinary nouns and (uninflected) postpositions, also can take pronominal dependents in both nominative and genitive, which makes these cases closer to relational nouns. However, it must be noted that this resemblance is not absolute and has substantial reservations. Apart from the fact that speakers usually allow only one of the options for the cases (both being available for relational nouns), they differ in the possessive marking. Compare different case forms of the relational noun ul 'bottom':
[(26).sub.f] mon ul-t'i/menam ul-t'i-m/*menam ul-t'i I.NOM under-PROL I.GEN under-PROL-P.1(SG) I.GEN under-PROL kosk-i-z paroxod leave-PST-3(SG) ship 'A ship swam (lit. went away) below me'
[(27).sub.f] mon ul-en/menam ul-a-m/*menam ul-en I.NOM under-LOC I.gen under-LOC-P.1(sG) I.gen under-LOC t'imofej ul-e Timofej live-PRS.3SG 'Timofej lives [in the apartment] beneath me'
Whenever the dependent of the relational noun is in the genitive, the relational noun must bear a corresponding possessive suffix. Pronominal forms with both recessive and approximative markers, on the contrary, never take possessive suffixes, according to our data: (6)
[(28).sub.f] *mon-lasen-em/*menam-lasen-em, I.NOM-RCS-p.1(SG) I.GEN-RCS-p.1(SG) *ton-lasen-ed/*tenad-lasen-ed you(SG).NOM-RCS-P.2(SG) you(SG).GEN-RCS-P.2SG
[(29).sub.f] *mon-lan-em/*menam-lan-em, I.NOM-APPR-p.1(SG) I.GEN-APPR-p.1(SG) *ton-lan-ed/*tenad-lan-ed you.NOM-APPR-P.2(SG) you.GEN-APPR-P.2(SG)
2. Unlike the markers of other cases, the recessive marker can be attached to a noun in the genitive form, i.e., be a part of a form representing a subtype of derivational case compounding (Noonan 2008). This compounding, however, is only attested in the oldest speakers:
[(30).sub.f] muzik-len-lasen suan-e tros men-o husband-GEN-RCS wedding-ILL many go-PRS.3PL 'From the husband s side many people are coming to the wedding'
[(31).sub.f] muzik-lasen suan-e tros men-o husband-RCS wedding-ILL many go-PRS.3PL 'From the husband's side many people are coming to the wedding'
Note that the forms with and without the genitive marker from (30) and (31) are completely synonymous. We must establish the fact that at least in several idiolects the recessive (and also the approximative (see (33) below) forms of nouns are based upon either nominative (bare stem) or genitive forms. All other case markers can be compounded with the genitive suffix under certain circumstances (when the NP head undergoes ellipsis, in comparative constructions etc.), but not in cases like (30-31). 3. Suspended affixation is a situation where only one affix is attached to a coordinated noun phrase consisting of several nouns (see, e. g., Kabak 2007 for a detailed account of this phenomenon in Turkish). In coordinate constructions the suspended affix, while attached to only one of the nouns, is marking the case for all of them, as in the following Eastern Armenian example (32):
(32) Eastern Armenian: (7) [erg u par]-ov song and dance-INS 'with song and dance'
Generally, Beserman and literary Udmurt case markers never allow being used as suspended suffixes in any of the several coordinating constructions that exist in the language. However, speakers occasionally agree to suspend the recessive marker in nominal coordinating constructions, attaching it only to the rightmost member of the conjoined phrase, in both the construction with no 'and' (33a) and the comitative coordinating construction (33b):
[(33a).sub.f] Mi kat vicak rodna-os-te we.NOM now every relative-PL-ACC.PL [kesno=no muzik = no]-lasen wife=and husband=and RCS nem-en nem-en vera-sko-m name-INSTR name-INSTR say-PRS-1PL 'Now we call all the relatives, both on the wife's side and on the husband's side, by name'
[(33b).sub.f] Mi kat vicak rodna-os-te we.NOM now every relative-PL-ACC.PL [kesno-en muzik-en]-lasen wife-INSTR husband-INSTR RCS nem-en nem-en vera-sko-m name-INSTR name-INSTR say-PRS-1PL 'Now we call all the relatives, both on the wife's side and on the husband's side, by name'
Another exception is again the approximative (-lan), which was allowed by one speaker to be suspended in a similar fashion (34b), along with the non-suspended usage (34a):
[(34a).sub.f] zor kosk-e Bagurt-lan=no Jezgurt-lan rain leave-PRS.3SG Abasevo=ADD Jozevo-APPR 'The rain is going away to the direction of Abasevo and Jozevo [nearby villages]'
[(34b).sub.f] zor kosk-e [Bagurt = no Jezgurt]-lan rain leave-PRS.3SG Abasevo=ADD Jozevo-APPR 'The rain is going away in the directions of Abasevo and Jozevo'
Not only almost all case markers, but also some relational nouns do not take a conjoined phrase as their dependent. For example, for vel 'top, surface' the vast majority of speakers excluded contexts with coordination between dependents (35a). At the same time, some relational nouns allow it; for example, for vis 'interval; between' the comitative construction of dependents is the default variant and the coordination construction, quite as expected, is also allowed (35b):
[(35a).sub.f] *skap = no zek = no/*skap-en zek-en cupboard=ADD table=ADD/cupboard-INS table-INS vel-ist-ed kopot' cus-el surface-EL-P.2(sG) dust wipe-ITER.IMP 'Wipe the dust from the cupboard and the table'
[(35b).sub.f] gibi bud-e kes.pu=no kez = no visk-en mushroom grow-PRS.3SG birch=ADD fir=ADD interval.OBL-LOC 'A mushroom grows between a birch and a fir tree'
For most other relational nouns, the opinions of the speakers were polarized.
4. As was noted in 3.1.3, the linear position of the possessive suffixes with respect to the case marker is different for different case markers, but fixed for each case marker. The recessive marker, however, is a unique case in the Beserman nominal paradigm, for which this order is not fixed. The noun from (36), for example, allows for both p.3sg-rcs and rcs-p.3sg orders:
(36)f korka-ez-lasen/korka-lasen-dz/*korka-lasen-ez house-P.3(sG)-RCS house-RCS-P.3(sG) house-RCS-P.3(sG) 'from the side of his/her house'
Phrases headed by relational nouns, again, show the same variation. Thus, the possessive marker in them (and also in NPs formed by other types of nouns, see Arkhangelskiy, Usacheva 2015) can appear either on the dependent, or on the head noun. Compare (38) and (39) below:
[(37).sub.f] so teb-i-z korka-je dor-e/ he go.up-PST-3(SG) house-P.1sG neighbourhood-ILL korka dor-a-m uza-ne house neighbourhood-ILL-P.1(SG) work-INF He went up to my house to work
In this respect, the recessive marker structurally resembles the relational noun phrase. But although such variation is a feature that moves the recessive marker closer to the relational nouns than other cases, it nevertheless does not allow us to actually treat it as an independent relational noun or postposition like the Udmurt lasan. The ungrammatical alternative form *korka-lasen-ez in (36) is a clear indication thereof. The P.3SG marker in Beserman normally takes the form -ez when used before case markers or in absence thereof, with the exception of a closed list of nouns which mainly include kinship terms. After the case markers, however, it can only appear as -ez (unless in compounded forms). Therefore, the ungrammaticality of the form *korka-lasen-ez indicates that -lasen is treated like a case affix here.
For all other cases, the approximative included, there is no such variation. However, it must be noted that some speakers refuse to combine the approximative with possessive markers altogether.
5. There are recessive forms of adjectives like bur 'right, pallan 'left' and muket 'other'. They function as adverbs: burlasen 'from the right side', pal'tanlasen 'from the left side', muketlasen 'from the other side'.
[(38).sub.f] so bur-lasen suj-z-e kuja-na kucik-e that right-RCS soil-P.3(SG)-ACC throw-INF begin-PRS.3SG He starts throwing the soil from the right side'
The adverbial function is standard for local case forms in Beserman. In an adverbial function, local cases of relational nouns, such as vel-e 'top-ILL', can be used without dependents. However, this is typical for locative case forms of nouns rather than adjectives. As a rule, adjectives in Beserman attract case suffixes in the case of substantivation or (quite seldom) in an attributive position under certain conditions (contrast, focus, etc.), copying the case of the noun they modify. Moreover, the recessive form of adjectives is semantically different from other case forms. It looks like the adjective is semantically modifying the meaning 'side' expressed in the recessive suffix. Such behaviour is generally not characteristic for fully grammaticalized cases. However, there exist similar approximative forms, while 'right' and 'left' also have illatives.
6. The semantics of the recessive case is much narrower than that of all other cases. Apart from the primary spatial meaning, there is only one other meaning expressed metaphorically with this case, '(guests/relatives) from someone's side' (see (8)). There are no verbs that govern the recessive or constructions that require it.
3.2.2. Case-like properties
On the other hand, there are properties distinguishing the recessive from postpositions and relational nouns.
1. The o-series of deictic pronouns has a recessive member olasen:
[(39).sub.f] gord masina lakt-em Kwaka-lasen, a red car come-PST2 Bird.street-RCS and ted'e masina pumet'-a-z lekt-e o-lasen white car opposite-ILL-P.3(SG) come-PRS.3SG there-RCS 'The red car came from the side of the Bird street, and the white car is coming in the opposite direction from the other side'
Unlike the so- and ta- series derived from the corresponding personal pronouns which exist in isolation and can be used with postpositions and relational nouns, the o- series does not have a non-marked member that could be followed by a relational noun and it does not even have core case forms.
2. The recessive marker cannot be detached from the noun it refers to (40), just as any other case marker, including the approximative:
[(40).sub.f] *abi-je-len korka-ez lekt-o lasen skal-jos grandmother-P.1SG-GEN house-P.3(SG) come-PRS.3PL RCS cow-PL *'There are cows coming from my grandmother house's side'
Although we found only two properties that strongly suggest treating -lasen as a case marker, note that most of the properties in 3.2.1 do not explicitly suggest otherwise. While most of them indicate that the item in question still possesses a number of features non-typical for well grammaticalized cases and resembles relational nouns in some respects, some of them in fact provide evidence against such treatment.
3.3. Remarks on the approximative and standard Udmurt
The approximative is quite an old and well-established case in Permic languages, dating back to Common Permic ([TEXT NOT REPRODUCIBLE IN ASCII] 1963: 17-18; Csucs 2005: 184). However, as the above examples show, it still shares with the recessive many properties that suggest its incomplete grammaticalization. Therefore, the Beserman approximative is an interesting example of a case that has existed for a long period, but still exhibits properties characteristic of undergrammaticalized units.
In standard Udmurt, meanwhile, the approximative seems to have been evolving along a different trajectory. All Udmurt grammars describe it alongside other cases without additional comments (e.g. Winkler 2000: 25, [TEXT NOT REPRODUCIBLE IN ASCII] 1962: 111), citing examples like surlan 'in the direction of the river'. While such a description would be undoubtedly correct for the Beserman dialect, inflectional approximative is all but nonexistent in contemporary literary Udmurt. In the Udmurt corpus, (8) out of the 1837 occurrences of the suffix -lan in a word-final position, there were only three genuine examples of the approximative inflection with full-fledged nouns (namely, gurt 'village', kisnomurt 'woman' and limsor'south'). Moreover, all of them occurred in verse, which is more likely to retain obsolete forms than prose. Only a handful of other approximatives out of this sample were formed from full-fledged nouns (zt 'evening' and sizil 'autumn'), but these seem to have reached the stage of lexicalized adverbs. The rest were all combinations of a stem from a closed list of 19 relational nouns, pronouns and adjectives, and the approximative marker, which seems to be of a derivational rather than inflectional nature. The hypothesis about the derivational status of -lan in contemporary literary Udmurt is supported by the fact that in order to express a lative or locative orientation, these words are very frequently combined with other locative cases, most often with the illative and locative. Thus, instead of the form az-lan front-APPR 'ahead (e.g. moving ahead)', the combination az-lan-e front-APPR-ILL is used, along with another combination, such as az-lan-in front-APPR-LOC 'ahead (e.g. located ahead)'. According to Nekrasova ([TEXT NOT REPRODUCIBLE IN ASCII] 1990), the approximative suffix in Komi has followed a similar development path.
It is worth noting that nearly the same set of relational nouns, pronouns and adjectives are also used with -lasan and are considered adverbs by Udmurt grammars and dictionaries (and written as a single word), unlike combinations of full-fledged nouns with lasan. There are no combinations of -lasan with case markers in the corpus, just as there are no combinations of it with the possessives (except a single example of NP head ellipsis, after which possessives can end up on any part of speech). The corpus data does not enable drawing conclusions regarding the possible grammaticalization or morphologization of lasan. However, there are still a couple of observations which might indicate that this in fact might be the case.
First, there is a peculiar complementary distribution between lasan and the egressive marker -isen. There is a handful of town/village names in -a, which have irregular declension in the illative (-la instead of -je), the locative (-lan instead of -in), the elative (-las instead of -is) and the egressive (lasen instead of -isen). While lasan can be encountered with virtually any noun, including a wide range of place names (like Rossija 'Russia' or Moskva 'Moscow'), we could not find any instances of it with any of the place names from this closed list, either in the corpus, or with Google (which includes search on at least two OCR-ed book collections in Udmurt). However, we found a number of occurrences of those place names in the egressive (like Mozga-lasen 'from Mozga' or Purga-lasen 'from Purga'), which formally coincides with the Beserman recessive. Since the meaning of the egressive with place names is very close to that of lasan, it seems tempting to suppose that at least some of the speakers see them as equivalents with a complementary distribution, which would gradually attract lasan to the case status. Of course, these preliminary findings should be checked with native speakers and could turn out to be wrong due to the scarcity of the available data.
Second, we have attested the pronominal form olasan (cf. (39) in 3.2.2) in two early 20th-century Udmurt newspapers. (9) As we argued above, such forms suggest a morphologization of the Udmurt -lasan, as the o-stem deictic pronoun does not have a free form and does not co-occur with postpositions. However, these findings should also be checked, as we did not find occurrences of olasan in more recent texts, and their appearance in a literary Udmurt text might have resulted from the influence of the dialect of the author. Apart from this form, we have also found in the book collections one occurrence of the form talasen derived from the pronoun ta 'this' ([TEXT NOT REPRODUCIBLE IN ASCII] 1949: 66). A normal form for the Beserman dialect, it is, based on the corpus data, hardly acceptable in literary Udmurt and we have no explanation for its appearance in a text written in literary Udmurt.
3.4. Comparison and conclusions
Let us now compare the properties of the forms in question. Apart from the forms with the suffix -n and the recessive, we are going to also include the approximative in the comparison, as the latter demonstrates several recessive-like properties. Before summarizing we need to give some explanations.
The next remark concerns the degree of productivity. As we have shown in (3.1.1), n-forms exist only for limited classes of nouns and pronouns, whereas the recessive (like most well grammaticalized cases as well as relational nouns) has a wide distribution. Thus, there are recessive forms of nouns denoting persons, including kinship terms (8), proper names (41), inanimate nouns (7), relational nouns (42), reflexives (43), numerals (44), deictic pronouns (39), deverbal nomina loci (45) etc.
[(41).sub.f] masina-os lakt-o Bagurt-lasen = no Jezurt-lasen car-PL come-PRS.3PL Abasevo-RCS=ADD Jozvo-RCS 'The cars are coming from the sides of Abasevo and Jozvo'
[(42).sub.c] tare sed masina-ez anaj-en ataj-en donga-lo ber-lasen then black car-ACC mother-INS father-INS push-PRS.3PL back-RCS 'Further, it is a black car which the father and the mother are pushing from behind'
[(43).sub.f] odig-e aslam-lasen lakt-e, ot-e one-P.3(SG) refl.1PL-RCS come-PRS.3sG there-LOC puk-e peres d'edus sit-PRS.3SG old old.man 'One [car] is coming from our side, there is an old man sitting in it'
[(44).sub.f] gord masina-le es-jos-s-e kek-na-lasen pas=ni, red car-GEN2 door-PL-P.3-PL two-OBL-RCS open=already kek-na-lasen pas masina-z-e two-OBL-RCS open car-P.3-PL 'The doors of the red car are open on two sides, their car is open on two sides'
[(45).sub.f] zivot l'uka-n-nig-lasen lakt-i-z traktor cattle gather-NMLZ-NLOC-RCS come-PST-3(SG) tractor 'From the side of the place where cattle is gathered a tractor came'
The final comment concerns possessive suffixes in pronominal forms. In approximative and recessive forms they must be absent, as we have shown earlier. In n-forms and in the forms of grammaticalized cases the presence or absence of the suffixes in question is a matter of the concrete case. Compare the fragments of 1sg and 2sg pronominal paradigms in (46):
[(46).sub.f] a. min-n-a-m tin-n-a-d we.OBL-DMS-LOC/ILL-P.1(SG) you.OBL-DMS-LOC/ILL-P.1(SG) 'at/from our place at/from your place'
b. min-n-ist-em tin-n-ist-ed we.OBL-DMS-EL-P.1(SG) you.OBL-DMS-EL -p.1(sg) 'from our place from your place'
c. min-n-isen tin-n-isen we.OBL-DMS-EGR you.OBL-DMS-EGR 'from our place from your place'
The properties of the forms containing the domus suffix n-, the recessive and approximative markers in comparison with those of relational nouns and grammaticalized cases are summarized in Table 1. One can see that n-forms are quite close to forms of grammaticalized cases (5 properties out of 8) and very far from relational nouns (only one common property). This fact leads to the suggestion that these forms probably have not arisen in Beserman independently but have been borrowed from northern Udmurt vernacular. It might have been the case that only combinations of -n- with locative and illative suffixes were borrowed, while combinations with other spatial case markers could have arisen by analogy. This hypothesis is supported by the fact that their most probable ancestor, the Permic postposition din-/din-/dyn- 'near', which, in turn, arose from the noun din 'basis, proximity, neighbourhood' ([TEXT NOT REPRODUCIBLE IN ASCII] 2008: 84), has not been attested in Beserman. Whatever its source, the suffix in question demonstrates a good degree of grammaticalization.
The recessive marker combines relational noun-like properties with those of grammaticalized cases (3 vs 4, respectively). Still, we believe its case-like properties meet stronger criteria than the properties it shares with relational nouns, and the latter are generally expressed to a limited extent. Therefore, we prefer to treat this marker as an incompletely grammaticalized case rather than an independent relational noun. One possible alternative to this analysis might be treating -lasen as a morphologicalized relational noun (this interpretation is proposed in Belyaev 2015 for Ossetic directive, recessive and comitative markers, which demonstrate many postpositional properties, just as the Beserman recessive and approximative do). We should conclude that in the case of the Beserman recessive the process of morphologization has outrun the process of grammaticalization. The source of the marker in question seems to be the construction X + palasen (X + pal-EGR) 'from X's side (see 2.2), while the egressive form of certain proper names in -lasen (3.3) has probably also influenced its development.
We have also cited several examples with the approximative case, which demonstrates two properties of relational nouns (together with six of well grammaticalized cases). As it is quite an old and stable case, relational noun-like properties might be the result of the influence of the recessive, which is cognitively close to the approximative as it has just the opposite meaning. On the other hand, the existence of the recessive in Beserman might have contributed to its existence as a productive case, while in standard Udmurt it is no longer used productively.
In the present article we described the properties of orientation forms, recessive and approximative markers. We can see that in the Beserman nominal inflection system there are three different phenomena involving grammaticalization: grammaticalization itself (orientation suffix), morphologization (recessive marker), and possibly partial demorphologization under internal influence (approximative marker). The preconditions for those phenomena are contained in the history and current development of Beserman grammar.
Lomonosov Moscow State University
National Research University
Higher School of Economics
ACC--accusative; ADD--coordinative clitic; APPR--approximative case; CAUS--causative; CMPR--comparative; CVB--converb; DAT--dative; DEB--debitive; DETR--detransitive; DMS--approximate-local series marker; EGR--egressive case; EL--elative case; FUT--future tense; GEN--genitive case (used for marking dependents of NPs in all syntactic positions except that of direct object); GEN2--objective genitive case (used for marking dependents of NPs in a direct object position); ILL--illative; IMP--imperative; INF--infinitive; INS--instrumental case; ITER--iterative; LOC--locative case; NEG--negation, negative; NLOC--locative nominalization (in -nig); NMLZ--nominalization in -on; NOM--nominative case; OBL--oblique; P--possessive marker; PL--plural; PROL--prolative case; PRS--present tense; PST--first (direct) past tense; PST2--second (evidential) past tense; Q--interrogative clitic; RCS--recessive case; REFL--reflexive pronoun; SG--singular; TERM--terminative case; 1-1st person; 2-2nd person; 3-3rd person.
Arkhangelskiy, T., Usacheva, M. 2015, Syntactic and Morphosyntactic Properties of PPs in Beserman Udmurt as Part-of-Speech Criteria.--SKY Journal of Linguistics 28, 103-137.
Ashbury, A. 2008, The Morphosyntax of Case and Adpositions, Utrecht [unpublished doctoral dissertation].
Baker, R. 1985, The Development of the Komi Case System. A Dialectological Investigation, Helsinki (MSFOu 189).
Bartens, R. 2003, Kahden kaasuspaatteen jonoista suomalais-ugrilaisissa kie lissa.--Unnepi kotet Honti Laszlo tiszteletere, Budapest, 46-54.
Belyaev, O. 2010, Evolution of Case in Ossetic.--Iran and the Caucasus 14, No. 2, 287-322.
--2015, "Incorporating Case" in Ossetic and the Typology of Case Markers. Report at the Department of Typology Seminar, Institute of Linguistics, RAS, March 31, 2015 [unpublished].
Bielecki, R. 2012, On Gradualness of Grammaticalization of Desinential Case Exponents in Finno-Ugric Languages.--Poznan Studies in Contemporary Linguistics. PSiCL 48, 537-564.
Bybee, J. 2007, Frequency of Use and the Organization of Language, Oxford.
Comrie, B. 1999, Spatial Cases in Daghestanian Languages.--Sprachtypologie und Universalienforschung 52, No. 2, 108-117.
Comrie, B., Polinsky, M. 1998, The Great Daghestanian Case Hoax.--Case, Typology and Grammar. In Honor of Barry J. Blake, Amsterdam, 95-114.
Creissels, D. 2009, Spatial Cases.--The Oxford Handbook of Case, Oxford, 609-625.
Csucs, S. 2005, Die Rekonstruktion der permischen Grundsprache, Budapest.
Daniel, M., Ganenkov, D. 2008, Case Marking in Daghestanian. Limits of Elaboration.--The Oxford Handbook of Case, Oxford, 668-685.
DeLance y, S. 1997, Grammaticalization and the Gradience of Categories. Relator Nouns and Postpositions in Tibetan and Burmese.--Essays on Language Function and Language Type. Dedicated to T. Givon, Amsterdam, 51-69.
Dixon, R. M. W. 2010. Basic Linguistic Theory. Volume 1. Methodology, Oxford.
Double Case. Agreement by Suffixaufnahme, New York--Oxford 1995.
Dryer, M. S. 2013, Order of Adposition and Noun Phrase.--The World Atlas of Language Structures Online, Leipzig. http://wals.info/chapter/85.
Grenoble, L. A. 2014, Spatial Semantics, Case, and Relational Nouns in Evenki.--Typology of Languages of Europe and Northern and Central Asia, Amsterdam, 109-132
Imre, S. 1971. A mai magyar nyelvjarasok rendszere, Budapest.
Jadhav, A. 2014, Marathi Postpositions.--Language in India, Vol. 14 (9), 37-56.
Kabak, B. 2007, Turkish Suspended Affixation.--Linguistics 45, 311-347.
Kahr, J. C. 1975, Adpositions and Locationals.--Working Papers on Language Universals 19, 21-54.
Korhonen, M. 1996, Remarks on the Structure and History of the Uralic Case System.--Typological and Historical Studies in Language by Mikko Korhonen. A Memorial Volume Published on the 60th Anniversary of His Birth, Helsinki (MSFOu 223), 219-234.
Lehmann, C. 2002, Thoughts on Grammaticalization. Vol. 2 (revised edition), Erfurt (Arbeitspapiere des Seminars fur Sprachwissenschaft der Universitat Erfurt 9).
Malchukov, A. 2009, Rare and Exotic Cases.--Oxford Handbook of Case, Oxford, 635-651.
--2010, "Quirky Case": Rare Phenomena in Case-Marking and Their Implications for a Theory of Typological Distributions.--Rethinking Universals. How Rarities Affect Linguistic Theory, Berlin, 139-169
Noonan, M. 2008, Case Compounding in the Bodic Languages.--Case and Grammatical Relations, Amsterdam, 127-148.
Spencer, A. 2008, Does Hungarian Have a Case System?--Case and Grammatical Relations. Studies in Honor of Bernard Comrie, Amsterdam (Typological Studies in Language 81), 35-56.
Starosta, S. 1985, Relator Nouns as a Source of Case Inflection.--For Gordon H. Fairbanks, Honolulu (Oceanic Linguistics Special Publication No. 20), 111-133.
Svenonius, P. 2006, The Emergence of Axial Parts.--Nordlyd: Tromso Working Papers in Linguistics, Vol. 33.1. Special Issue on Adpositions, Tromso, 50-77.
Thuilier, J. 2011, Case Suffixes and Postpositions in Hungarian.--Proceedings of Head-Driven Phrase Structure Grammar 2011 Conference, Stanford 209-226. http://web.stanford.edu/group/cslipublications/cslipublications/ HPSG/2011/thuilier.pdf.
Winkler, E. 2001, Udmurt, Munich (Languages of the World, vol. 212).
[TEXT NOT REPRODUCIBLE IN ASCII]
* This research was supported by the Russian Foundation for Basic research (grant No. 16-24-17003).
(1) Examples marked by the index c are taken either from the corpus of Beserman texts available on-line at http://beserman.ru or from texts recorded and transcribed during a series of referential communication experiments, which are not yet available in the corpus. Examples marked by f are taken from our own field notes and are mostly produced by elicitation.
(2) In most Udmurt grammars dor and the units with the same properties are considered to be postpositions. For the reasons to treat them as relational nouns in Beserman see Arkhangelskiy, Usacheva 2015 and also subsection 3.2 of the present paper.
(3) The experiment carried out in 2010 is described in detail in [TEXT NOT REPRODUCIBLE IN ASCII] 2012. In 2010 there were much fewer figures than in 2015, and the scenario of movement was different, but the three-dimensional models of the area were very much alike.
(4) Following Noonan (2008), we use case compounding as a term which covers all phenomena involving the presence of more than one case marker in one word: double and multiple case-marking, case layering, case stacking, Suffixaufnahme etc. (for typological observations see Malchukov 2009; 2010). The notion of double case-marking goes back to "Double case. Agreement by Suffixaufnahme" (1995).
(5) Following Testelec ([TEXT NOT REPRODUCIBLE IN ASCII] 2011), we treat ellipsis as a phenomenon by which certain elements of a syntactic structure are not expressed by the antecedent, which creates context redundancy.
(6) According to Tepljasina ([TEXT NOT REPRODUCIBLE IN ASCII] 1970: 185), the approximative forms allow the possessive marking with the nominative stem and with -a- instead of -e- (mon-lan-am, ton-lan-ad). The allowing of possessive marking with the nominative (rather than the genitive) stem makes the approximative marker even less relational-nounlike in the varieties of Beserman where it is possible.
(7) Example from the Eastern Armenian National Corpus (http://eanc.net/).
(8) The corpus is available online at http://web-corpora.net/UdmurtCorpus/.
(9) [TEXT NOT REPRODUCIBLE IN ASCII] [News from the war], January 31, 1916; [TEXT NOT REPRODUCIBLE IN ASCII] [The new village], No. 19 (26), May 19, 1928. The newspapers are available at the page of Fenno-Ugrica, the National Library of Finland's digital collection of Finno-Ugric publications (https://fennougrica.kansalliskirjasto.fi/).
(10) We have not attested any forms with variable order in our data, but we have not checked all possible combinations and thus we do not have explicit prohibitions for some of them.
Table 1 Properties of Beserman n-forms, the forms with recessive and approximative markers in comparison with those of relational nouns and well grammaticalized cases relational grammaticalized forms with nouns case markers -n- 'domus' attachment depends on * * to a conjoined phrase particular noun detachment + * * from the head noun derivational attachment + * * to nouns in gen form of the nom or nom for oblique pronominal stem gen non-core cases, (1sg, 2sg) special oblique forms for gen, gen2 and dat high productivity + + * possessive suffix if pronoun depends on depends on in pronominal forms is in the particular case particular genitive case variation of the + * * (10) linear position of the possessive suffixes narrow semantics + * + recessive approximative attachment +? +?? to a conjoined phrase detachment * * from the head noun derivational attachment +? * to nouns in gen form of the nom or nom or pronominal stem gen gen (1sg, 2sg) high productivity + + possessive suffix never never in pronominal forms variation of the + * linear position of the possessive suffixes narrow semantics + +
|Printer friendly Cite/link Email Feedback|
|Author:||Usacheva, Maria; Arkhangelskiy, Timofey|
|Date:||Mar 1, 2017|
|Previous Article:||Lexical innovations in the Erzya-Mordvin translations of the Lord's prayer.|
|Next Article:||Adnominal possessive constructions in Narym, Vasjugan and middle-ob dialects of Selkup.|