Symposium B6: Uralic nominal structures: noun phrases and nominal sentences

Organizer: Barbara Egedi

Thursday, August 25, Department of Slavic Studies (Slawistik), room 1

(Original symposium call)


Timofey Arkhangelskiy & Maria Usacheva:

Grammaticalization of relational nouns in Permic languages

Most inflected postpositions, or relational nouns, in Permic languages are historically derived from (ordinary) nouns. While this fact is rather obvious and has long been known, little has been said about particular stages of this process.

It is usually assumed in grammars that ordinary nouns (labeled N below) and standard relational nouns (labeled RelNPst) constitute two distinct morphosyntactic classes in the standard languages, without any further syntactic gradations. For Udmurt, their properties can be summarized as follows:

- nouns can have genitive modification (1a), where any NP can be the modifier and the head must bear a possessive suffix;

- nouns can also have nominative modification (a construction called “composition” by Winkler (2011:90), analyzed as compounding by Fejes (2005) and, on Beserman Udmurt material, as incorporation by Simonenko and Leontyev (2012:323)); the dependent cannot be specific and cannot have any morphology.

- relational nouns can only have nominative modification and disallow genitive modification (2)

(1a)      korka-len      ukno-ja-z                     (1b)        korka     ukno-je

            house-gen    window-ill-p.3sg                    house     window-ill

‘to the window of the house’                       ‘to a/the house window’

(2)       korka     dor-i̮

            house     near-ill

‘to the house’

Based on corpus data and our own fieldwork in multiple Permic dialects, we will show that apart from these two classes there are actually further subclasses with different sets of syntactic properties (some of which exist only in dialects). Their properties, listed in Table 1, allow us to view them as diachronic stages of a single grammaticalization cline.










N,spec + N







N,pl + N







N,poss + N























PRO + N-Px







N + N-Px







Table 1. Distributional subclasses of Udmurt (relational) nouns

Based on the data of Table 1, the following diachronic cline can be reconstructed for Udmurt relational nouns:

1 (N+). An ordinary noun with spatial semantics acquires the ability of having specific unmarked nominal dependents.

2ʼ (RelN-). It allows more nominal structure on nominal dependents in the nominative: plural and, in some cases, possessiveness. If possessiveness is allowed, personal pronouns in the nominative are also allowed as dependents, and vice versa.

2ʼʼ (Nlight). Alternatively, the item with a meaning like “time” or “place” can become a “light noun”, allowing pronominal dependents in the nominative and plural (but not the possessive) on nominal dependents in the nominative.

3 (RelNgen). The item gradually loses the ability of having nominal dependents in the genitive.

4 (RelNst). The item gradually loses the ability of having pronominal dependents in the genitive (this can happen at different rates for different combinations of person and number).

In our talk, we will present a similar cline for Komi and provide examples for each of the proposed subclasses.


Fejes, László. 2005. Összetett szavak finnugor nyelvekben, Ph.D. thesis, Eötvös Loránd Tudományegyetem, Budapest.

Simonenko, Aleksandra & Alexey Leontyev. 2012. Morfosintaksis imennogo kompleksa v finno-permskix jazykax: analiz v ramkax gipotezy minimalizma. In Ariadna Kuznetsova (ed.), Finno-ugorskie jazyki: Fragmenty grammaticheskogo opisanija, 262–340. Moscow: Jazyki slavjanskix kul’tur.

Winkler, Eberhard. 2011. Udmurtische Grammatik. Veröffentlichungen der Societas Uralo-Altaica / Band 81. Wiesbaden: Harrasowitz Verlag.


Mariia Privizentseva:

Nominal ellipsis and concord in Moksha

In Moksha Mordvin, modifiers generally do not agree with the noun, but are infected if the noun is elided. All data come from my fieldwork with native speakers.

(1) mon   and-in'ə                   akšə     katə-t'       /     akšə-t'        /     *akšə-t'            katə-t' 

    I           feed-pst.3.o.1sg.s   white   cat-def.gen    white-def.gen     white-def.gen  cat-def.gen

I fed the white cat / the white one.

This pattern is typically analyzed as cliticization of stranded nominal features to the linearly closest host (Ruda 2016; Saab and Lipták 2016; Saab 2019). This approach does not derive Moksha data, because inflection appears on the head of a closest nominal modifier, which is not always the linearly closest host for affixes, as in (2).​

(2) mon   rama-jn'ə               ti-f                      keluv-ən'    lopa-stə     nastojka-t'                   /

     I         buy-pst.3.o.1sg.s    make-ptcp.res  birch-gen    eaf-el         liquor-def.gen    

    ti-f-t'                                  keluv-ən'     lopa-stə   /   *ti-f                       keluv-ən'         lopa-stə-t'

    make-ptcp.res-def.gen     birch-gen    leaf-el            make-ptcp.res    birch-gen        leaf-el-def.gen

I bought the liquor / the one which is made from birch leafs.

Inflection under ellipsis is possible with all modifiers that are not marked for case. Modifier with case fall into two groups, as shown in table (4). Here is the generalization:

(3) If the noun is elided, its φ-features are obligatory expressed on its modifier unless the modifier has its own φ-features.

Empirical support comes from non-verbal predication. Adjectives in the predicative position tend to agree with the subject and nouns rather do not agree (Stassen, 1992, 2005). This is because nouns have their own φ-features which intervene and block agreement with a different noun (Baker 2008). If (3) is correct, inflecting modifiers should also agree in the predicative position and non-inflecting modifiers should not. This is indeed the case:

Under ellipsis

Number agreement

in predicative position




Indefinite genitive









Definite genitive



Definite dative



Indefinite dative






In languages with concord modifiers with their own φ-features don't show concord, others do (Baker 2008). I propose that inflection under ellipsis is an instance of nominal concord and take concord to result from Agree (see Carstens (2001, 2018); Baker (2008); Toosarvandani and van Urk (2014)). Agreement remains invisible if the noun is present and it is phonologically realized if the noun is elided.

This is derived as follows. First, I argue that Spell-out to PF occurs once a node has no features that can induce operations in syntax (cf. Epstein and Seely (2002); Wojdak (2008) on local Spell-Out). I further suggest that features that induce syntactic operations (i.e. features triggering Merge, Agree and ellipsis), are never  morphologically realized. Second, I argue that after Valuation Probes undergo Conversion. Conversion makes a Probe indistinguishable from a feature inherit to a node. Unconverted Probes do not prevent Spell-out of a node, but are not morphologically realized themselves.

A modifier agrees with the noun, case and φ-features probe together. It is the last operation triggered by features on a nominal modifier, so that it is sent to PF right after Probes are valued, but before their Conversion. Feature [E] triggers ellipsis (Merchant 2001). Its presence makes agreement with the noun not the last operation, postpones Spell-out and allows for Probe conversion.

Under this approach nominal ellipsis makes a general but otherwise invisible properly of Moksha syntax apparent.


Maria Usacheva:

Plural marking of nouns and adjectives in Beserman Udmurt 

In Udmurt there are two non-verbal plural markers: -(j)os and -. According to general information from Udmurt grammars (Winkler 2001: 15, 40, Perevoschikov et al. 1962: 73, 127), -(j)os is used with nouns and - with adjectives:

1.Udmurt [Perevoschikov et al. 1962: 73]:

Magaźin-len ukno-os-i̮z badӡӡi̮m-es.
shop-gen1 window-pl-poss.3sg large-pl
ʼThe shopʼs windows are large.ʼ

It is also usually stated that -ešʼ is obligatory with predicates and optional with adjectives used as attributes (Winkler 2011: 81-84, Perevoschikov et al. 1962: 128):

(2) Udmurt [Winkler 2001: 40]:
badӡӡi̮m(-eś) gurt-jos-i̮n
large(-pl) village-pl-in
ʼin large villagesʼ

With numerals plural marking of nouns is also optional.

In Udmurt dialects plural marking of nouns and adjectives is not identical. Winkler (2001: 40, 2011: 83) states that adjectives can attach both nominal and adjectival plural suffixes if they are marked by POSS.3SG suffix:

(3) Udmurt [Winkler 2001: 40]:
badӡӡi̮m-es-jos-a-z gurt-jos-i̮n
large-pl-pl-ill-poss.3sg village-pl-in
ʼin the large villagesʼ

Saparova (2018: 162) claims that speakers of central Udmurt dialects treat such examples as ungrammatical.

Our work is devoted to plural marking in Beserman Udmurt. We consider noun phrases, numeral phrases, adjectives in predicative position, substantivation. The data were taken from Beserman corpora (, and gathered during series of referential communication tasks experiments (the method described by Krauss and Weinheimer (1966)) in the Beserman village of Shamardan (Yukamenskoye district, Udmurtia, Russian Federation) in 2017. We claim that in Beserman Udmurt plural marking in nominal structures is influenced by the following hierarchy of factors (the strongest factors are on the left):

Information structure (contrast, topic/focus)>syntactic position (subject vs direct object vs dative vs predicate)>activation cost>animacy hierarchy, comparative clitic ges (present vs absent)>order of adjectives>lexical semantics


Krauss, Robert M., and Sidney Weinheimer. 1966. Concurrent feedback, confirmation and the encoding of referents in verbal communication // Journal of Personality and Social Psychology 4. Pp. 343–346.

Perevoschikov, P. N., V. M. Vakhrushev, Vasily I. Alatyrev, A. A. Pozdeyeva, I. V. Tarakanov. 1962. Grammatika sovremennogo udmurtskogo yazyka: fonetika i morfologiya [The grammar of modern Udmurt: phonetics and morphology]. Izhevsk: Udmurtskoye knizhnoye izdatelstvo.

Saparova, Daria A. 2018. Soglasovaniye po chislu prilagateljnykh v udmurtskom yazyke [Number agreement of adjectives in Udmurt] // Problemy yazyka, Sbornik nauchnykh statej po materialam Shestoy konferencii-shkoly “Problemy yazyka: vzglyad molodykh uchenykh” (12 – 14 marta 2018). Moscow: Institut yazykoznaniya RAN, Kanzler. Pp. 160 – 179. Winkler, Eberhard. 2001. Udmurt. Languages of the World/Materials 212. Muenchen: Lincom Europa.

Winkler, Eberhard. 2011. Udmurtisch. Muenchen.


Suzanne Lesage:

Gradient constraints on the realisation of Estonian possessives


Most Finno-Ugric languages show a rich possessive suffix system. In Finnish for instance, the suffix expressing the possessor is mandatory and added to the possessed noun. If the possessor is a first or second person, a genitive pronoun can precede the possessed noun. If the possessor is a third person, it is impossible for the genitive pronoun to refer to the subject, it is mandatory that it refer to a nonsubject argument. 

Estonian stands out, employing a very different system to express the possessor. The possessor, if it does not refer to the subject, is expressed by a short or long form of genitive pronoun preceding the possessed noun.  If the possessor refers to the subjects, the possessor is expressed by the reflexive possessive oma. More importantly, the overt expression of a possessor is not mandatory. 

















I forgot your/my book at home.

The constraints on the realisation of possessives are noncategorical and multifactorial. Previous corpus analyses focused on the distribution of nonreflexive and reflexive possessives, without taking into account sentences where the possessive is not realised, as it is impossible to identify implicit possessors automatically from a raw corpus and a hand annotation would be very costly. To overcome this difficulty, I relied on the parallel Estonian-French corpus CoPEF ( I looked for the Estonian equivalents of French1 segments containing a possessive. I randomly picked 50 examples for each subcorpus and obtained 900 examples. After removing false positive examples, I annotated the 733 remaining ones by hand. The statistical analysis of the data reveals that the realisation of Estonian possessive is significantly affected by:

  • the text type
  • the language of the original text
  • the antecedent’s syntactic function and its person
  • the semantic class of the possessed noun.

One main finding is that there is a strong correlation between subject antecedent and covert possessive. This means that reflexive possessive tends to be overtly expressed less often than the nonreflexive possessive. The result can be explained by the fact that the subject is more salient than other arguments and that we tend to refer to a salient argument by a reduced referent (Givon, 1983).

I also analysed examples with an overt nonreflexive possessive focusing on the form (short or long) used. I observed a significant effect of the text type, antecedent person, and the semantic class of the possessed noun.

There is no correlation between the factors having a significant effect on the possessives’ realisation and the ones having an effect on their form. We could conclude that the short possessive, though shorter, doesn’t allow reference to a more salient referent than a long possessive.

1 In French, the rate of overt possessives is very high. 




Irina Burukina:


Referential DPs and reflexive pronouns: a case study of Hill Mari

I focus on the morphosyntactic properties of reflexive škӛ-POS-CASE pronouns in Hill Mari, comparing them to possessed referential DPs. Adapting the typology of reflexives developed by Déchaine and Wiltschko (2012), I argue that škӛ- anaphors should be analyzed as full DPs, in contrast with structurally reduced anaphoric elements in some other languages (phi-Ps, nPs, etc.). Examining similarities in the behavior of possessed DPs and reflexives contributes to the study of anaphors in Uralic languages (Savatkova 2002, Volkova 2014, a.o.) and helps to better understand the structure of nominal elements. The data under consideration were collected during my field work in Kuznetsovo, Mari El, Russia. Hill Mari utilizes two kinds of reflexive pronouns: (i) agreeing (ӛ)škӛ- reflexives (1a), and (ii) the unmarked (ӛ)ške form, often used with postpositions (1b).

(1) a. tӛn' škӛm-et-ӛm jarat-et.
‘You love yourself.’

b. tӛn' ške giš-än-et šajәə št-at.
you REFL about-LAT2-POS.2SG talk-NPST.2SG
‘You talk about yourself.’

As proposed by Déchaine and Wiltschko (2012), reflexive markers typologically differ in morphosyntactic behavior and structural properties. The two most widespread types of reflexives are D-reflexives, whose distribution is similar to that of referential DPs, and phi-reflexives behaving like clitics. I argue that the properties of škӛ- reflexives in Hill Mari are parallel to those of referential DPs. First, reflexives in Hill Mari can both saturate various arguments and function as predicates (cf. in English you are not yourself). Second, reflexives allow various kinds of modifiers, including postpositive adjuncts and appositive constructions (2).

(2) škəm-em-əm plat'j-an-әə m už-am.
‘I see myself in a dress.’

Third, a constituent headed by a reflexive can be independently negated (3) and / or focused, which implies that reflexive pronouns are morphosyntactically independent from the predicate.

(3) tӛn' olma-m škӛ-län-et agәə l näl-ӛn-ät.
‘You bought an apple not for yourself.’

Finally, reflexives and referential DPs in Hill Mari can be coordinated, which supports the claim that škӛ- anaphors are full DPs (cf. Kayne 1975).

In addition to this, reflexives are similar to possessed DPs (e.g. pört-em-län house-POS.1SG-DAT) and contrast with personal pronouns in allowing order alternation between a possessive marker and the dative suffix. Furthermore, if a reflexive pronoun is selected by a postposition, many speakers accept a possessive marker added either directly to the anaphor or to the postpositive element (examples are omitted due to the limitations of space); this behavior is also identical to that of possessed DPs but not pronouns or reduced nominal phrases. I will further discuss inter-speaker variation with regard to the positioning of affixes and how it suggests (partial) grammaticalization of anaphoric items.


Chris Lasse Däbritz & Beáta Wagner-Nagy:


Existential, Locative and Possessive Clauses in Kamas

Existential, locative and possessive clauses form an essential group of non-verbal/nominal predication in the sense of Payne (1997: 112ff.): Existential clauses predicate the existence of an entity, the theme, at some location; locative clauses, in turn, predicate the location of the theme at a certain place. Possessive clauses, finally, predicate possession, i.e., simply spoken, the belonging of one entity (possessee) to another entity (possessor).

Already Lyons (1967) observes that in many languages these clause types share several morphosyntactic features. Most Uralic languages are no exception in this respect, cf. the studies of Bartens (1996), Wagner-Nagy (2011), Laakso & Wagner-Nagy (forthc.). In Kamas, existential (1) and locative (2) clauses seem to differ only in their word order: In both cases the location is expressed by an adverbial marked with locative case, whereas the theme is the unmarked subject of the clause. Possessive clauses (3) syntactically resemble the existential type, the possessor being marked with genitive case, the possessee being the subject of the clause:

(1)       A          băra-gən          ši          i-bi.

            and      sack-loc          hole     be-pst.3sg

            ‘And there was a hole in the sack.’


(2)       Măn                 kaga-m                        văjna-gən         i-bie.

            1sg.gen           brother-poss1sg          war-loc           be-pst.3sg

            ‘My brother was in the war.’


(3)       Dĭ        nüke-n             nʼi-t                 i-bi.

            this      woman-gen     son-poss3sg    be-pst.3sg

            ‘This woman had a son.’


The talk shall investigate into the formation of both affirmative and negative existential, locative and possessive clauses in Kamas. Most attention will be drawn to the coding of the theme, location and possessor/possessee respectively as well as to the syntactic structure of the clauses. The empirical base of the study will be the INEL Kamas Corpus (Gusev & Klooster 2018), which contains both early recordings of Kamas (cf. Joki 1944) and recordings of Klavdiya Plotnikova’s Kamas.


Bartens, R. 1996. Die positive und negative Existentiale in den finnisch-ugrischen Sprachen. UAJb NF 14, 58–97.

Gusev, V. & Klooster, T. 2018. INEL Kamas Corpus. Version 0.1. Publication date 2018-12-31. Archived in Hamburger Zentrum für Sprachkorpora. In: Wagner-Nagy, B. & Arkhipov, A. & F., Anne & Jettka, D. & Lehmberg, T. (eds.). 2018. The INEL corpora of indigenous Northern Eurasian languages.

Joki, A. 1944. Kai Donners Kamassisches Wörterbuch nebst Sprachproben und Hauptzügen der Grammatik. Lexica Societatis Fenno-Ugricae 8. Helsinki: SUS.

Laakso, J. & Wagner-Nagy, B. (forthc.). Existential, locational and possessive clauses. In: Bakró-Nagy, M. & Laakso, J. & Skribnik, E. (eds.). Oxford guide to the Uralic languages. Oxford: OUP.

Lyons, J. 1967. A note on possessive, existential and locative sentences. Foundations of Language 3(4), 390–396.

Payne, T. E. 1997. Describing morphosyntax. A guide for field linguists. Cambridge: CUP.

Wagner-Nagy, B. 2011. On the typology of negation in Ob-Ugric and Samoyedic languages. SUST 262. Helsinki: SUS.

[1] The examples come from the INEL Kamas Corpus (Gusev & Klooster 2018).