TWO NOMINAL CLAUSE-TYPES IN NORTHERN MANSI: AN EXPERIMENTAL STUDY OF LANGUAGE VARIATION

. The paper examines the structure and distribution of two types of nominal/adjectival predicates in the Northern Mansi language. A nominative noun or adjective serves as the predicate in one construction. The other predicate type contains a predicate noun or adjective that takes translative case marking. In both constructions, the stative-like copula ō l - ’be, exist’ can also appear, though under different conditions. In the paper we focus on (a) the licensing conditions of the ō l - copula, (b) the predicate-subject agreement morphology, and (c) concord within the predicate phrase in both predicates. Our findings demonstrate that the two types exhibit systematic structural differences: the copula ō l - is utilized in the nominative construction in the past, while it must be omitted in the present. The ō l - copula is always obligatory in the translative predicate. The nominative predicate noun/adjective takes the morpheme of the subject agreement in number, and we attested inter-and intra-speaker variation in Number concord in this construction when there is an overt copula in the predicate phrase. The translative-marked nominal/adjectival predicate does not take any inflectional suffix, and agreement that indicates both the person and the number of the subject is marked on the ō l - copula. Additionally, we will show that only the translative-type is acceptable in identificational clauses. As a result, the identificational reading/interpretation is where the semantic division of labor between the two constructions lies. Our data come from fieldwork where Mansi native speakers helped us with survey research. Northern Mansi newspaper texts were also used to clarify certain inconsistencies between our findings and the literature.


Иван лēккарыг ōлыс Ivan lÍēkkar-¸ ōl-ǝs
Ivan doctor-TRSL be-PST.3SG'Ivan was a doctor' (4) Ты кол мāниг ōлыс ti kol māń-¸ ōl-əs this house small-TRSL be-PST.3SG'This house was small' We will focus on two types of non-verbal predication in Northern Mansi.In the typological literature, one of them is often referred to as "predicational" copular sentence (cf. Higgins 1979;Mikkelsen 2004), or "true nominal predicate" (e.g.Dryer 2007).The predicate in this sentence-type is a noun phrase that is indefinite or nonreferential, see the Northern Mansi examples in (1a), (2a), and (3).An adjectival phrase functions as the predicate in the other non-verbal sentence-type, which is termed as an "attributive sentence" in the literature (see, e.g., Payne 1997).The adjective predicate in this sentence-type describes a property that is attributed to the subject of the sentence, see the Northern Mansi examples in (1b), (2b), and (4).We put aside the discussion of further types of non-verbal clauses in Northern Mansi, and get back to them in Section 5.2.When used as the predicates in either of the aforementioned constructions, nouns and adjectives have no (known) grammatical and/or structural differences.Therefore we will treat these two syntactic categories as members of a homogeneous group, and we will refer to this group as Nominals.
Several open questions regarding this topic are still to be found, some of which we aim to answer here.In our paper we will systematically examine the complete paradigms of the two types by focusing on the licensing requirements of the ōlcopula, subject agreement, and concord within the Nominal predicate.We describe the two constructions in detail and provide grammatically acceptable and not acceptable constructions based on native speakers' judgements.Our results show that the two types exhibit systematic structural differences in (a) allowing an overt copula, and (b) taking subject agreement inflection: in the nominative construction there is obligatory subject agreement in number in the present tense on the Nominal predicate, and the copula is only used in the past, where we attested both inter-and intra-speaker variation in Number concord.In contrast, the translative-marked Nominal predicate cannot take any (further) inflectional suffix, and agreement that indicates both the person and the number of the subject is marked on the obligatory copula.We assume that the two Nominal predicates have different underlying structures, but the analysis of these two constructions is beyond the scope of this paper.
Additionally, we will investigate the distribution of the two constructions in various contexts.The translative-marked predicate construction is usually regarded as a direct result of Russian contact in Northern Mansi (Скрибник, Афанасьева 2007 : 53;Sipőcz 2017 : 385).This implicitly assumes that the construction with the nominative Nominal predicate is the original Mansi structure, whereas the translative-marked one is an innovation.This claim is partly supported by our findings: we attest to a widespread variation between the two constructions in the past tense, but the translative-marked construction is hardly, and certainly not automatically, available in the present tense.This patterns with the Russian Nominal predicate system, where there is also an alternation in morphological case marking: the Nominal predicate can be marked with nominative case (5) or instrumental case (6) (see e.g.Pereltsvaig 2001).The non-marked nominative and the marked instrumental opposition is only observed in the past (and the future) tenses (Pereltsvaig 2001).We will show, however, that in Northern Mansi there is inter-speaker variation in the availability of the translative-type in the present tense.This observation suggests that the two constructions coexist within the same grammar of certain speakers.The fact that the variation is not the consequence of two competing systems (but more likely the part of a single grammar) is further supported by our data, i.e. we found an additional context in which only the translative type is allowed.In identificational constructions only the translative type is accepted.Thus, the semantic division of labour of the two construc-tions lies along the identificational reading/interpretation of the constructions (we will define the semantic terms in Section 5.2).
In Northern Mansi there are also predicate Nominal-like constructions formed with semi-copulas such as jēmt-, and pat-'become', i.e. dynamic verbs.In these clauses, the Nominal part of the predicate takes the translative ending too, see ( 7) and ( 8 In our paper, we focus on constructions with the stative copula and exclude constructions containing dynamic semi-copulas from our discussion (like those in ( 7) and ( 8)).
The outline of the paper is as follows.In Section 2 we give a brief overview of the Mansi language.Then we discuss our data collecting methods in detail in Section 3. Section 4 focuses on structural aspects of the two Nominal predicate types where we take a closer look at (a) the licensing conditions of a zero/non-zero copula, (b) the marking strategies of subject agreement, and (c) concord within the predicate complex.In Section 5 we focus on the contrastive properties of the two variants in terms of tense and semantic/pragmatic factors.Section 6 contains our concluding remarks.

Demography
Mansi is a severely endangered Uralic language spoken in Western-Siberia.According to the 2020 census data of the Russian Federation, there are 12,228 Mansi in Russia (Census RF 2020 5/1), the majority of whom reside in the Khanty-Mansi Autonomous Okrug, Yugra.The remaining roughly 1,000 people live mostly in the neighbouring administrative districts of the Russian Federation: 526 people in other parts of the Tyumen region, 345 people in the Sverdlovsk area, and 5 people in the Komi Republic (Census RF 2020 5/17).2,093 Mansis stated that they had Mansi as their mother tongue (Census RF 2020 5/20).A total of 1,346 people stated that they spoke Mansi (Census RF 2020 5/4), and 1,236 of them were of Mansi ethnicity (Census RF 2020 4/19).The Khanty-Mansi Autonomous Okrug has a population of 1.7 million, the majority of whom are Russian (51.9%).The actual indigenous population of the Okrug forms approximately 2% of the total population, while the Mansi represent only 0.65%.According to the Ukrainian census data from 2001, 43 Mansi live in Ukraine, of whom 5 named Mansi as their mother tongue (Census Ukraine 2001).
Four Mansi dialect groups were documented in the nineteenth century: Northern, Eastern, Southern, and Western Mansi, each of which had several (sub-)dialects.The southern and western dialects are already extinct, the eastern dialect is either extinct or moribund (cf.Riese 2001 : 7; Ромбандеева, Вахрушева 1984 : 3).Henceforth in this paper where the Mansi language is mentioned, we refer to Northern Mansi.Mansi is used in both spoken and written form.It is spoken most often in private life with relatives and childhood friends.The written language is used primarily in the monthly newspaper Lūimā Sēripos.
The level of speakers' proficiency in Mansi is typically related to their age: the older the speakers are, the more likely they are to have native competence in Mansi.This general tendency is often counterbalanced by the speaker's place of birth and residence: younger speakers born and raised in smaller Mansi settlements also often have good command of Mansi language (cf.Horváth 2020).

Preliminaries on the structure of Northern Mansi
Mansi is an agglutinating, transitive-accusative language low in fusion and inflection.The basic constituent order in a neutral clause is S(O)V.The finite verb is in clause-final position in both active and passive sentences.
Nouns in Mansi are inflected for number, person, possession, and case.There are three numbers: singular, dual, and plural.There are six cases: nominative, lative, locative, ablative, instrumental, and translative.The translative case is generally not used in the dual and plural.Verbs have the inflectional categories of subject person, subject and object number, tense, mood, and voice.
The verb ōl-'to be' is used both in the meaning of 'to be, to exist' and 'to live, to reside', it is inflected in all persons and numbers (e.g.Virtanen, Horváth 2023 : 678).

Research methods
In order to get a better understanding of the contemporary representations and distribution of the constructions, we decided to gather more information by conducting survey research with the help of Mansi native speakers.
The survey research, completed by Mansi consultants with the assistance of the authors, took place during fieldwork conducted in Khanty-Mansijsk in 2018, as well as independently by the consultants in 2019.
We set up our surveys by using standard data collection techniques, especially stimulus-driven and target-language-manipulation elicitation techniques.We approached the consultants in two stages.First, we contacted a group of native Mansi speakers and requested them to fill in our exploratory survey.Then, based on the results, we manipulated the Mansi data, i.e. we designed a grammatical and contextual acceptability judgment survey and got some of our consultants to fill out this second, control survey.
The consultants participating in the research were Mansi specialists working with the Mansi language on a daily basis (mostly journalists and teachers).They are all native speakers of the Mansi language, bilingual in Mansi and Russian, and alumni of the Institute of the Peoples of the North. 3 The total number of questionnaires completed is seven.

Structural differences between the two Nominal predicates
This section focuses on two parameters of Nominal predicates: the conditions that license the presence/absence of the ōl-copula in the nominative and the translative Nominal predicates, and the subject agreement inflection realised on the predicate of the Nominal clause (incl.Concord).
It is claimed in the literature that these predicates differ in the case of the predicate Nominal and whether they appear with or without a copula.It has been shown that a copula is usually missing in the present tense in the nominative type in any number and person (e.g.Kálmán 1976 : 66;Keresztes 1998 : 411;Riese 2001 : 29).Our findings are in line with the literature and show that the nominative predicate construction does not contain the copula in the present tense.Examples ( 9)-( 11) illustrate construction with a 3rd person singular, dual, and plural subject respectively, and in (12) the subject is a 1st person singular subject.None of the sentences below contain the ōl-copula. 3The authors find it necessary to note that they have never felt obliged to follow the outdated, but still widely used standards for choosing supposedly ideal research participants, which are frequently used not only in dialectal research but also in the study of indigenous languages.Contrarily, the authors believe it is important to interact and work with consultants who are not NORMs (nonmobile, old, rural, male, cf.Chambers, Trudgill 1998).Taking into account the influence of education, the authors do not think that Mansi consultants who hold degrees in Mansi philology would have less proficiency in their first language or less established judgement about the grammaticality of the sentences than other speakers.On the other hand, the fact that these consultants have Mansi as their first language, work with Mansi on a daily basis, and are literate in Mansi, made their involvement in our research project an enormous benefit and addition to our research.
Our survey further revealed that the copula is invariably absent in the present tense in the nominative construction (13).Thus whenever the copula is missing from the construction, it must be obligatorily absent.
(13) Кол мāнь (*ōлы) kol māń (*ōl-i) house.SG small.SG be-PST.3SG'The house is small' In the past tense, the nominative predicate construction requires the ōlcopula that takes the tense inflection ( 14).Tense feature morphology is not indicated on the predicate Nominal.
(14) Кол мāнь ōлыс kol māń ōl-ǝs house.SG small.SG be-PST.3SG'The house was small' As already mentioned, the translative predicate Nominal construction is normally available in the past tense, where the ōl-copula is obligatory, and it takes agreement and tense inflection, see (15).The translative predicate Nominal does not take any morphological marker other than the translative marker.As example (15) further suggests, the copula cannot be absent in this type of predicates.
(15) Кол мāниг *(ōлыс) Kol māń-ɣ *(ōl-ǝs) house small-TRSL be-PST.3SG'The house was small' In the translative predicate construction, nevertheless, the copula ōl-appears even in the present tense ( 16)-( 18).(In Section 5.1, we go into further detail on the availability of the translative Nominal predicate in the present tense.) The copula cannot be absent in the present tense either.
( It is demonstrated in ( 21) that, in the case of non-third person subjects, the person feature of the subject is also not realized on the Nominal predicate.
In the past form, where the copula is also there, the subject agreement relation surfaces on the copula, too.Note that the copula takes both person and number inflection via a verb suffix ( 22).
(22) Колыт мāнит ōлсыт kol-ǝt māń-ǝt ōl-s-ǝt house-PL small-PL be-PST-3PL 'The houses were small' In addition, the number feature of the subject may be realised by number inflection on the nominative Nominal predicate.Thus, the predicate construction may show partial concord ( 22).We consider it as partial concord as the Nominal predicate does not agree in person, but in number.Interestingly, we attested both intra-and inter-speaker variation in partial concord in Number.In the production of some speakers an alternative construction appeared where the partial concord in Number is missing (23).
(23) Колыт мāнь ōлсыт kol-ǝt māń ōl-s-ǝt house-PL small.SG be-PST-3PL 'The houses were small' We were not able to reveal any difference between the concord and nonconcord constructions in terms of their use and/or semantics.Still, there were no speakers who only produced the non-concord variant.Therefore, we assume that the non-concord construction is an innovation in the language.
In the translative Nominal predicate, it is only the copula that takes agreement inflection of the person and number features of the subject via the regular verbal agreement marker.The translative Nominal does not take any agreement inflection (24).

Table 1
The structural differences between the two predicate Nominal constructions

The distribution of the two Nominal predicates
This part of the paper focuses on two aspects of the distribution of the two Nominal predicates.On one hand we discuss a dimension along which one can account for the existence of the variation, which is the dimension of past tense.On the other hand, we consider possible interpretational differences that may be found between the two Mansi Nominal predicate types.

Variation in past tense
As mentioned in Section 4, the literature accounts for the variation between the two Nominal predicates both in the present and in the past tenses (Sipőcz 2015;2017).Our results show that for the consultants the alternation is linked only to the past tense.It means that the consultants produced the translative-marked construction as an alternative to the nominative one only in the past tense.This suggests that Northern Mansi displays a similar (or even the same) system as Russian discussed in Section 1 (see e.g.Pereltsvaig 2001).Since the translative construction in Northern Mansi is regarded as acquired from the Russian language, we presumed that the Russian system was adopted into Northern Mansi via the same rules, i.e. the non-marked and marked alternation is only observed in the past in Northern Mansi.Interestingly, corpus data contradict the results of our survey and our preliminary hypothesis.We found examples of the translative-type construction in the present tense as the ones in ( 25)-( 27).In order to resolve this contradiction, we controlled for this parameter in our second survey that was filled by a different consultant.The results of this second survey show that the translative-marked predicate nominal is indeed grammatical in the present tense for this consultant (28).
(28) Ольга ханищтан н г ō лы OlÍga ˛ańiśtan nē-ɣ ōl-i Olga teaching woman-TRSL be-PRS.3SG'Olga is a/the teacher' Based on this finding, it is reasonable to account for the alternation both in the present and in the past tenses.We assume the existence of an interspeaker variation, and suggest that some Northern Mansi speakers' grammar still contains the Russian rule, and they accept the translative construction only in the past tense, whereas for other Northern Mansi speakers the alternation between the two predicates is not limited to the past.

Variation in terms of semantics/pragmatics
As was already mentioned in the Introduction, the predicate in sentences with adjectival predicates, i.e. in attributive clauses defines a quality that is assigned to the subject of the sentence (Payne 1997; Dryer 2007).We did not find any noticeable semantic difference between the nominative and translative constructions with attributive function.Therefore, our focus in this Section will solely be on potential semantic variations in sentences containing NP predicates.
The semantic distribution of the nominative and translative type with NP predicates is rather controversial in the literature.Riese (2001 : 29, 61) and Sipőcz (2015 : 384), for instance, do not account for any semantic distribution between the two types, while Virtanen and Horváth (2023 : 694)  We adopted several parameters discussed by Roy (2005) to test for potential semantic and/or pragmatic differences between the two constructions.We set up contexts and tested whether both or either of the target constructions are (and can be) used.
The two predicate-types do not contrast in terms of the semantic concept of "lifetime effect", see ( 30)-( 31).This finding partly conflicts with Virta-nen and Horváth (2023 : 694) in that the term "lifetime effect" can be used to refer to a permanent state.The translative predicate appears to be appropriate in a non-temporary context as well, as shown in (31).However, Virtanen and Hor váth's (2023 : 694) examples show the relatively rare present tense translative construction, whereas our instances of the translative-type predicate construction have past tense reference.Therefore, we are unable to confirm whether the present tense reference may activate the temporary reading of the translative-type.In the previously indicated contexts, the two constructions are found to be semantically equal.It is clear from these contexts that the two constructions express predicational sentences.The general literature often distinguishes at least two types of NP predicates, as we noted in the Introduction (see, e.g., Higgins 1979;Mikkelsen 2005;Dryer 2007).A non-referential noun phrase serves as the predicate in predicational sentences, also known as "true nominal predicates".Thus in examples (30)-( 35), the predicate NP is non-referential/indefinite.The terms "equative" or "(true) equational clauses" or "identificational" clauses refer to NP predicates in which the subject and predicate complement are both referential expressions (see, e.g., Higgins 1979;Mikkelsen 2005;Dryer 2007).In a context implifying identificational reading it is the translative-marked construction that is accepted in Northern Mansi, compare (36) with (37).The sign # is used to mark semantically strange/unacceptable but grammatically well-formed sentences.
Situation: The old man had three sons, Peter, Ivan, and Igor.

Conclusions
In this paper we discussed certain aspects of two Nominal predicate constructions that exhibit case alternation in Northern Mansi.We have shown that the morphological realisation of past tense is the condition of the copula support in the nominative construction.In the present tense, the copula is obligatorily absent in this type of Nominal predicate.In the translative construction, the copula is obligatorily present.In the nominative one, subject agreement in number appears on the predicate Nominal in cases when there is no copula.When the copula is present, agreement takes place both in person and number, and the predicate shows partial concord in Number (that seems to be optional).In the translative-type, there is no subject agreement on the predicate Nominal.Instead, the copula takes the agreement marker that indicates the person and the number of the subject.Finally, we found that the interpretation of the two Nominal predicate-types is distributed along the line of identification.It is only the translative-type that appears in identificational sentences.This semantic distinction is subtle, but it is there.
Test sentence: Ivan was a doctor, but unfortunately he is no longer alive.In addition, the actual practice of the activity expressed by the predicate Nominal is not a relevant distributional criterion either, see (32)-(33).