3.1 Genetic affiliation and sociolinguistic background

Udmurt belongs to the Permic subgroup of the Finno-Ugric branch of the Uralic language family. Its closest relatives are Komi-Zyrian and Komi-Permyak. Udmurt is mainly spoken in the Udmurt Republic of Russia as a minority language, and in the neighbouring administrative areas of the Russian Federation (Republic of Tatarstan, Republic of Bashkortostan, Mari El Republic, Perm Krai, Kirov Oblast, Sverdlovsk Oblast etc.). According to the Russian Census of 2020–21,³ Udmurt is spoken by 255 877 speakers in the territory of Russia, while the number of the ethnic population is 386 465. Although Udmurt, besides Russian, has the status of official language in the Udmurt Republic, its usage is limited in public spheres. Russian has had a strong influence on Udmurt, which manifests itself at all linguistic levels. Nowadays, practically all native speakers of Udmurt are bilingual, as they also speak Russian at a native or near-native level. Udmurt is part of the Volga-Kama Sprachbund, and has also been in contact with Turkic languages.

3.2 Main typological features and information structure

Udmurt is an agglutinative, pro-drop language with nominative-accusative alignment, differential object marking and SOV basic word order (although, according to Asztalos, Gugán & Mus 2017 and Asztalos (2018, 2021), contemporary Udmurt is undergoing a change towards the SVO type).

The placement of contrastive topics and of focused items is of relevance for the purposes of this study. Surányi et al. (to appear) claim that contrastive topics appear preverbally, on the left periphery of the clause, and they may be either preceded or followed by an aboutness topic as illustrated in (11):

With respect to focused constituents in Udmurt, Asztalos (2020) argues that they do not have a fixed position in the sentence. The results of her questionnaires indicate that the most accepted position for foci is the immediately preverbal one, however, sentence-final and preverbal but not verb-adjacent⁵ focusing is also grammatical for a part of speakers as the example in (12) illustrates (for sentence-final foci, see also Tánczos 2010).⁶

Asztalos (2020, 41–46) finds that focus is interpreted in the same way in all of these (linear) positions, that is, none of the above positions is associated with an obligatory exhaustive or contrastive reading. In a later study (Asztalos 2022), she argues that none of these positions is a syntactically marked (i.e., structural) focus position.

3.3 Sentential negation

The present subsection reviews the morphosyntactic means of sentential negation in Udmurt.

Clausal negation is expressed by the negation of the predicate of the sentence. Depending on the nominal or verbal nature and on the tense and mood of the predicate, the negator may be a negative verb,⁷ a negative form of the copula, a negative affix, or a negative particle.

As in many other Uralic languages, sentences with a verbal predicate are negated in Udmurt in most tenses by means of the complex unit of an inflected negative verb and the so-called connegative stem of the lexical verb (13)–(15). The negative verb always precedes the connegative stem. Person is marked on the negative verb, cf. the pairs in (13a) vs. (13b), and number is expressed on the connegative form, cf. the pairs in (13b) vs. (13c). Past tense is expressed on the negative verb (13). In the present (14) and future (15) tenses, the stem of the negative verb is the same (u-); present tense forms are differentiated from future tense forms by a present tense marker appearing on the connegative stem (14).

For the negation of a nominal predicate or an existential clause, the negator $\underset{̮}{e}$v$\underset{̮}{e}$l is used in the present tense (16)–(17). I take $\underset{̮}{e}$v$\underset{̮}{e}$l to be the present tense negative form of the defective verb v$\underset{̮}{i}$l$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be’ (see also Winkler 2011, 108 and Asztalos 2018, 95), but it can also be analysed as a negative particle (cf. Edygarova 2015, 275, 277).

The form $\underset{̮}{e}$v$\underset{̮}{e}$l is also used in the negation of certain tenses, e.g., the evidential past, as in (18). The evidential past, however, has a synthetic negative paradigm as well. In those forms, negation is expressed by a bound negative morpheme, as illustrated in (19):⁸

Contemporary Udmurt in some (non-literary) registers also allows for the use of the negative particle ńe (borrowed from Russian) for the purposes of sentential negation. In these cases, ńe is followed by the affirmative form of the verb (20):

In Udmurt, negation does not have an impact on the morphology of sentence constituents, that is, it does not alter the case of the direct object (whether a direct object stands in the nominative or in the accusative is determined by its specificity and definiteness, cf. É. Kiss & Tánczos 2018).

3.4 Previous analyses on negative verbs

A structural analysis of negative verbs in a generative framework has been offered by Mitchell (2006) and Georgieva, Salzmann & Weisser (2021). Mitchell (2006) proposes that negative verbs (described by her as auxiliaries) are heads projecting NegP-s, which are generated immediately below TP.⁹ Georgieva, Salzmann & Weisser (2021) propose that NegP-s are generated below TP and above vP, and they derive the preverbal surface position of negative verbs by means of a post-syntactic operation called Lowering. They show that negation forms a complex unit with the highest verb in the clause and takes high scope in the sentence, i.e., over the highest verb in the clause, as shown by the example in (21):

The authors, on the basis of examples provided by Edygarova (2015, 282), claim that subjects can optionally move out from the vP to Spec,TP. Constituents moved to Spec,TP take scope over negation (22), while elements that stay in the vP, like the quantifier in (23), fall under the scope of negation:

Georgieva, Salzmann & Weisser (2021, 498) mention another case when elements move out of the vP/VP to a higher position (and thus, negation does not scope over them), i.e., that of contrastive elements¹⁰ in polar answers, as illustrated by the answer sentence in (24):

3.5 Previous literature on constructions with a constituent negation reading

Constructions with a constituent negation reading in Udmurt have been discussed so far by Winkler (2001, 54; 2011, 109–10), Edygarova (2015, 285), and Kozmács (2017). The authors treat the constructions under discussion as instances of constituent negation. Edygarova claims that constituents can be negated in a biclausal contrastive structure in which the negated constituent is (immediately) followed by “a negative verb or particle without a lexical verb”, and the second clause contains the affirmative form of the verb (25) (ibid.) and the alternative of the negated constituent:

Thus, according to the data presented in Edygarova (2015),

1. Udmurt uses the same negators for constituent negation as for sentential negation (i.e., negative verbs, negative particles etc.);

2. constituent negation involves a biclausal structure;

3. the negated constituent is followed by the negative morpheme in the first clause;

4. the second clause contains the alternative of the negated constituent.

The same strategy is also exemplified by Winkler (2011, 110):

Winkler (2001, 54, 2011, 109) notes that constituent negation is expressed by the same means as sentential negation, but it can be made “more clear by word order or morphological means” (2001, 54) (he, however, does not make more clear what he means by the above observation). In any case, while the negated constituent immediately precedes the negative verb in Edygarova's (2015, 285) example in (25), in (28) it is not directly preverbal. A further difference compared to what Edygarova (2015) claims is that the example in (28) suggests that the alternative of the negated constituent does not necessarily have to make part of a full clause.

According to Kozmács (2017), contemporary Udmurt applies two strategies for negating constituents. The standard way involves the use of the negative verb and the connegative stem of the lexical verb. Clausal negation and constituent negation, according to Kozmács (2017), 277–278), are differentiated by word order, as illustrated in the examples in (29)–(30): if a given constituent follows the negative verb + verbal stem complex, we face sentential negation (29), whereas if the same constituent (immediately) precedes the verbal complex, constituent negation takes place (30). Kozmács (2017) Kozmács implicitly claims that a negated constituent always immediately precedes the negative verb (cf. 2017, 277–278).

The other strategy (which can be detected, according to Kozmács, 2017, in the Udmurt blogosphere and in texts of contemporary press), involves the negative particle ńe, borrowed from Russian (31):

Ńe precedes the constituent to be negated and is always combined with a verb in the affirmative form, as (31) illustrates.

To sum up, the observations of previous works on constituent negation raise some questions, yet some of the observations lead to contradiction. The literature suggests that

1. The native expression of constituent negation involves in Udmurt the same means as sentential negation (negative verb + connegative stem, or negative particles). According to Edygarova (2015), a native speaker, constituent negation involves biclausal constructions that mention the alternative of the negated constituent in the second clause (25), and Winkler's (2001, 2011) examples suggest the same (26)–(28). Kozmács (2017), however, reports a monoclausal example (30), claiming that word order alone is apt for differentiating between constituent negation and sentential negation (which would imply that there is no need for specifying the alternative(s) in order to get a constituent negation reading), and suggesting that a negated constituent always immediately precedes the negative verb (cf. 2017, 277–278). However, when taking into consideration Edygarova's (2015) and Winkler's (2011) examples, it is not obvious whether the constituent that the negation refers to has indeed a fixed position: in (25), it directly precedes the verb, thus not appearing in its base position (which would be the position preceding školaje ‘to school’), while in (28), it appears in its base position but not immediately before the verb. Furthermore, the examples in (25)–(28) suggest that in biclausal structures, the lexical verb may be absent either from the first (25)–(27) or from the second clause (28); however, these phenomena are not discussed by the authors themselves.

2. The contact-induced way of expression of constituent negation involves the negative particle ńe, borrowed from Russian. Ńe immediately precedes the constituent to be negated and is combined with a verb in the affirmative form (31), but it is not clear whether the alternative of a constituent negated by ńe has to be necessarily present in the sentence.

Finally, none of the above mentioned authors provides a syntactic analysis of the examples presented above, nor do they deal with the information structural role of the constituent that negation refers to. Thus, the syntactic structure of the constructions under discussion remains unclear, as well as the question whether they are to be analysed as instances of constituent negation in a syntactic sense.

3.6 Further considerations about the constructions under discussion

The picture that arises from the literature is further complicated by some observations I made on the basis of some preliminary data collected through consultations with a native speaker of Udmurt. As presented in the previous subsection, Kozmács (2017, 277–278) proposes that there is a one-to-one relationship between word order and sentential vs. constituent negation: when a constituent follows a negated predicate, clausal negation takes place, while when it precedes it, we face constituent negation. The data in (32), however, indicate that this is not necessarily the case: a negated constituent may also follow the negative verb + connegative stem complex, and there is no one-to-one relationship between the scope of negation and the position of a constituent in a negative sentence. Depending on the context, both (32a) and (32b) can get either a sentential negation or a constituent negation interpretation, and it is exactly the presence of the alternative(s) that ensures a constituent negation reading to the sentence. This implies that if the context is not specified (e.g., no alternative is given for the constituent kinoje ‘to the cinema’ in (29)–(30)), the same ambiguity should also hold for Kozmács's examples above.

It thus seems that the position of the negated constituent is not fixed, and a research on constructions with a constituent negation reading cannot be confined to sentences with the negated constituent in the immediately preverbal position.

Another question that arises on the basis of the previous literature and the preliminary data concerns the information structural role of the constituent that negation refers to. In fact, a linear sequence of elements of the type XP NegAux V¹¹ (where XP, in a first approach, is taken to be a constituent with an undefined role), can be associated with multiple readings, and may only be disambiguated by the context. Thus, a sequence like Śibiŕ-e $\underset{̮}{e}$-j m$\underset{̮}{i}$n$\underset{̮}{i}$ (Siberia-ill neg.pst-1sg go.cng.sg), depending on the context, may have the interpretation of constituent negation (33a), sentential negation (33b); furthermore, it may also be interpreted as a sentence with a contrastive topic (33c), or as a sentence with a focus taking scope over negation (33d) (the question whether the different readings are associated with different prosodic properties has not been studied yet). The main point to be clarified is whether the constituent that negation refers to is information structurally neutral, or bears the role of contrastive topic, or that of focus.

The example in (28) suggests that ellipsis in the second clause is possible, which would be evidence for the focus status of the negated constituent (this will be discussed in Section 4.1 in more detail). If indeed this is the case, the negated constituent should be able to stand in any of the positions that foci in contemporary Udmurt can stand (i.e., immediately and not immediately preverbal and sentence-final, as outlined in Section 3.2).

As presented in Section 3.5 and also mentioned at the beginning of this section, neither the linear nor the structural position of the negated constituent is made clear in the literature: in Edygarova's (2015, 285) example in (25), the negated constituent has the syntactic function of subject and it stands in a non-base position, immediately preceding the verb (in a neutral sentence, the subject would precede the goal argument of the verb, cf. Glezdenev 1921, 45–47; Asztalos, Gugán & Mus 2017, 51). This suggests either that the negated constituent is focused, or that the goal argument is topicalized. However, in Winkler's (2011, 110) example in (28) the negated constituent does not stand in the immediately preverbal position, yet in (32b), it follows the verb.

Constituents with the role of focus, negators and quantifiers are regarded in theories of formal syntax and semantics as operators (see, e.g., Horvath 2013). Operators are scope-taking categories, i.e., the operation they encode is to be interpreted within a certain sentence domain. Two (or more) operators in a sentence enter into a scopal interaction, i.e., one of the operators takes scope over the other one. When the scope relations between two operators are reflected by the linear ordering of the operators (i.e., the operator with wide scope precedes the one with narrow scope), transparent (or linear) scope reading takes place, while the opposite case gives rise to an inverse scope reading.¹³ Scopal ambiguity arises when a sequence allows both readings. Thus, if it turns out that (33a) has the same structure as (33d), that is, both sentences contain a focused element which linearly precedes negation, the question arises how the same sequence of focus and negation may give rise both to a transparent (Foc > ¬, cf. (33d)) and to an inverse (¬ > Foc, cf. (33a)) scope reading.

A sequence of the type NegAux V XP like $\underset{̮}{E}$-j myny Śibiŕ-e (neg.pst-1sg go.cng.sg Siberia-ill) (cf. 34a) can also be associated with more than one reading depending on the context: besides the constituent negation reading (34a), it can also express sentential negation (34b), or the argument of the verb, similarly to what we face in (33d), can be interpreted as a focus taking wide scope over negation (34c).¹⁴ In this latter case, inverse scope reading is taking place, as the negative verb precedes the focused element (Foc > ¬), whereas with the constituent negation reading (34a) (if the negated constituent bears the role of focus in the sentence) scope relations are expressed transparently (¬ > Foc). Scope relations in (34a) and (34c) would thus show the mirror image of what we face in the case of (33a) and (33d), respectively.

Constructions with the negative particle ńe, however, can only have one interpretation, since ńe always scopes over the constituent immediately following it. Thus, sentential negation reading could only arise if ńe were followed by the predicate (cf. the sentence in (20)). A wide scope focus reading (with focus taking scope over negation) would also involve a negated predicate. It also remains a question whether constructions with ńe, similarly to “standard” constructions, necessarily require the presence of the alternative(s) of the negated constituent. All in all, the question whether a constituent negated by ńe fulfils the role of focus, and, if so, whether it is a syntactically marked focus, is worth being investigated.

Summing up, the main questions that arise on the basis of previous work and preliminary data concern the information structural role of the negated constituent and its scopal interaction with negation (in case it is a focus). These are also the main questions that the present study aims at investigating. Syntactic “variables” that have to be taken into account while studying the constructions include the type of negator, the position of the negated constituent, the possibility of elliptical second clauses and, in case of negative verb constructions, the ellipsis of the lexical verb in the first clause.

4.1 Grammaticality judgements on constructed test sentences

As anticipated above, a part of the data came from the results of a grammaticality judgement test. The informants were two native speakers of Udmurt, both women, aged between 30 and 40, holding a university degree and residing in Hungary. The consultants had to evaluate the grammaticality of 20 test sentences using a 5-point Likert-scale, and they had to correct each sentence that they rated below 5. The test sentences were constructed by the author of the present paper and they were embedded in short dialogues. The sentences were given in a written form and the informants had to fulfil the task individually. The list of test sentences did not contain any filler items.

As anticipated in Section 3.6, in order to find out whether the negated constituent fulfils the role of focus, the syntactic test illustrated in (35) was applied. If the clause containing negation can be followed by a coordinated elliptical clause containing the alternative(s) of the negated constituent, the negated constituent bears the focus role.

If, however, such sentences turn out to be ungrammatical and only VP-coordination structures like the ones in (25)–(27) are grammatical, there are good reasons to assume that clausal negation is taking place: more precisely, the negation either of a neutral sentence (cf. 36b vs. 36a), or of a sentence with a contrastive topic (37).

In order to decide between the two latter options (more precisely, in order to check whether the possibility of a contrastive topic reading can be rejected), I used in half of the test sentences (that is, in 10 sentences in total), a verb that requires an adjunct with an exhaustive reading, i.e., vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’. The meaning of vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’ can only hold true for one locative phrase (38), thus, the locative phrase cannot be interpreted as a contrastive topic, as illustrated by the ungrammatical Hungarian example with contrastive topics in (39). In other words, if the constituent that negation seems to refer to bears the role of contrastive topic, sentences with vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’ should give ungrammatical results in biclausal constructions with VP-coordination. If they give grammatical results, a contrastive topic analysis should be excluded. Grammatical sentences with an exhaustive complement thus support the focus analysis.

Test sentences were constructed along the following parameters:

1. Type of negator: negative verb or particle ńe;

2. Position of the constituent that negation refers to (or seems to refer to): immediately preverbal/not immediately preverbal/postverbal;

3. In case of negative verb + preverbal constituent: only negative verb/negative verb + connegative stem in the first clause;

4. The presence of the verb in the second clause: elliptical second clause/VP-coordination.

Test sentences were embedded in question-answer contexts of the type illustrated in (40). Half of the test sentences, as mentioned above, contained the verb vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’, a subject and a locative adjunct that was intended to be negated. In order to make the test dialogues less repetitive and combat speaker fatigue, the subject and the locative adjunct were expressed by a different proper name or lexical item in each dialogue. The intended meaning of the dialogues/possible answer sentences is rendered by (40a–b), while two concrete test items are given in (41)–(42).

The other half of the test sentences (10 sentences in total) contained the verb $\underset{̮}{e}$t'$\underset{̮}{i}$n$\underset{̮}{i}$ ’to invite’, a direct object and a subject that was intended to be negated. As in the case of the test dialogues with the verb vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’, proper names and common nouns were different in each dialogue. The intended meaning of the dialogues is rendered by (43a–b) ((43a i) and ii) illustrate two theoretically possible interpretations of the same sentence), while a concrete test item is given in (44).

Table 1 summarizes the configurations of negators, verbal predicates, negated constituents and other elements that were tested.

4.2 Corpus data

Corpus data were retrieved from Udmurt corpus (which is the main corpus of the Udmurt corpora),¹⁶ by searching for the particle ńe.¹⁷ The total number of relevant hits amounted to 93 (examples similar to (20), with ńe preceding the predicate and not a lower sentence constituent, were excluded).

4.3 Further consultations with a native speaker

In a third phase of the research, I consulted one of the informants about further questions that had emerged during the analysis of the collected data. The task of the consultant was to translate from Hungarian to Udmurt sentences similar to (45)–(47).¹⁸ The test sentences aimed at investigating i) the grammaticality of double negation constructions, i.e., sentences containing both sentential negation and constituent-level negation, cf. (45), and ii) whether the occurrence of negation and of focused numerals in the same clause gives rise to sentences with an ambiguous scope reading, cf. (46)–(47).

5.1 Grammaticality judgement tasks

Tables 2 & 3 summarize the judgements of the consultants. As mentioned in Section 4.1, speakers had to correct each sentence that they rated below 5. As set forth in Sections 3.6 & 4.1, the research was concerned with the grammaticality of a predefined set of parameters (type of negator; word order in the first clause; ellipsis in the second clause; ellipsis of the connegative stem of the lexical verb in the first clause). Thus, corrections that did not affect these parameters were not taken into consideration from the point of view of the grammaticality of the sentences. To put it another way, when a speaker rated a given test sentence below 5 but her correction clearly did not affect the above parameters (e.g., she only deleted the conjunction of the elliptical second clause but otherwise left the sentence intact), her rating was taken to be 5.

As the tables illustrate, Speaker A was much more permissive and had much lower variability in her judgements than Speaker B.¹⁹ Sentences that were given at least 4 points by both speakers were interpreted as grammatical, while those where the two speakers' evaluations differed at least by 2 points were considered as subject to interspeaker variation.

Almost all test sentences with an elliptical second clause (i.e., configurations ending in [but YP]) were given at least 4 points by both speakers,²⁰ which suggests that ellipsis in the second clause is grammatical both in negative verb (48)–(50) and in ńe particle-constructions (51)–(53), independently of the position of the negated constituent (immediately preverbal (48), (51)/not immediately preverbal (49), (52)/postverbal (50), (53)). This implies that the negated constituent i) fulfils the role of focus in both construction types, and ii) can appear in any of the positions in which focused elements occur in contemporary Udmurt (immediately preverbal, not immediately preverbal and sentence-final, as presented in Section 3.2).

In negative verb constructions, the analysis of the negated constituent as a focus is also supported by the fact that both informants considered the VP-coordination structure as grammatical not only with the verb $\underset{̮}{e}$t’$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to invite’ but also with vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’, at least when the locative adjunct was in the immediately preverbal position (54)–(55) (i.e., in the position which, as presented in Section 3.2, is the most accepted position for foci according to Asztalos 2020). As argued in Section 3.6, a contrastive topic analysis of the locative phrases is unlikely in these sentences, since the meaning of a verb like ‘to be born’ requires that the locative adjunct had an exhaustive reading. The variant with the locative adjunct in postverbal position (56) received a less favourable judgement by Speaker B, who placed the negated constituent in her correction into the immediately preverbal position, while the variant with the locative phrase preceding the verb non-immediately (57)²¹ was least favoured by both speakers.

As Tables 2 & 3 show, all of the sentences with an immediately preverbal negated constituent and a non-elliptical first clause (i.e., those with the configurations [S/O XP NegAux V] [YP V] and [S/O XP NegAux V] [but YP], cf. rows 1 and 3 of the Tables and examples (48), (54)), were given 5 points by both speakers. This may be related to the “standard” nature of negative verb constructions, and is also in line with the proposed analysis of the negated constituent as a focus, since, as mentioned above, the immediately preverbal is the most preferred position for foci in Udmurt. The immediately preverbal position was highly accepted even when it was clearly not a base position for the negated constituent. In constructions with elliptical first clauses (i.e., with the connegative stem ellipted), Speaker B gave higher points for sentences with an immediately preverbal negated constituent (configuration [S/O XP NegAux] [YP V], 4 points) than for those with a sentence-initial negated constituent (configuration [XP S/O NegAux] [YP V], 2 points), cf. rows 2 and 6 of the Tables. In the case of the minimal pair with $\underset{̮}{e}$ť$\underset{̮}{i}$n$\underset{̮}{i}$, this meant that the OSV ordered sentence (58) was judged more favourably than the sentence showing the basic SOV pattern (59). The fact that the negated subject was preferred in a non-canonical position, identical to the position in which foci are most preferred (cf. Asztalos 2020), also suggests that the negated constituent bears the focus role in the sentence.

Finally, a couple of more general remarks have to be said about the overall evaluation of sentences with an elliptical first clause (i.e., with the connegative stem ellipted). Speaker B gave lower points to these sentences than to their non-elliptical counterparts, as illustrated by the sentences in (55) vs. (54). Throughout the whole test, both in the vordsk$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to be born’ and in the $\underset{̮}{e}$t’$\underset{̮}{i}$n$\underset{̮}{i}$ ‘to invite’ sentences, it was the configuration [XP S/O NegAux] [YP V] (with elliptical first clause and sentence-initial negated constituent, cf. the examples in (57) and (59)) and row 6 in the Tables) that received the lowest points (i.e., 2) by Speaker B, and this was also the configuration that received the lowest point on average (i.e., 3). It thus seems that the ellipsis of the connegative stem of the lexical verb is somewhat less preferred than its spell-out.

Turning to ńe-constructions, as it was anticipated at the beginning of this section, elliptical second clauses proved to be grammatical in their case as well, independently of the position of the constituent negated by ńe, cf. the sentences in (51)–(53) and rows 8–10 of the Tables. This shows that constituents negated by ńe are also foci, and that they can also appear in any of the positions in which foci can occur in contemporary Udmurt (cf. Asztalos 2020).

5.2 Corpus search

As mentioned in Section 4.2 corpus data were only retrieved for the ńe particle but not for the negative verb constructions. Data show that ńe constructions themselves may appear in elliptical clauses (60), and that, contrary to negative verb constructions, they do not necessarily require the presence of the alternative(s) of the negated constituent (61):

Corpus data also confirm the findings of the grammaticality judgement task that constituents negated with ńe can occur in different positions with respect to the main clausal predicate, e.g., not immediately before the verb, cf. (31), or postverbally (62):

Since the particle ńe is a functional element borrowed from Russian, the question may arise whether it is more frequently used with Russian than with Udmurt lexemes (or than with older loanwords more adapted to Udmurt). In the data retrieved from the corpus, 20.4% of constituents negated with ńe were composed of Russian lexemes, 17.2% of Russian and Udmurt lexemes, and 62.4% of Udmurt lexemes. Thus, negation with ńe does not seem to be more frequently used with Russian than with Udmurt lexemes (or than with older loanwords more adapted to Udmurt).

6.1 Negative verb constructions

6.2 Constructions with ńe

The results of the ellipsis test (cf. Section 5.1) show that constituents negated with the particle ńe also fulfil the role of focus in the sentence. Although, similarly to negative verb constructions, constituents negated with ńe do not have a fixed position in the sentence (they can immediately or non-immediately precede the verb or stand postverbally, cf. Section 5.1), the following considerations suggest that foci negated with ńe are not necessarily of a merely pragmatic nature:

1. i)The unambiguous interpretation of the constructions: Contrary to negative verb constructions, ńe-constructions have an unambigous interpretation: as mentioned in Sections 3.5 & 3.6, the particle ńe always negates the constituent immediately following it.
2. ii)The non-obligatory presence of the alternative(s) of the focus: Contrary to negative verb constructions, foci negated with ńe do not necessarily require the presence of the alternative(s) of the negated constituent: even in the absence of the alternative(s) it is obvious that negation refers to the constituent immediately following the particle ńe, as it was illustrated by the example in (61), cf. Section 5.2.

Thus, there are reasons to assume that constituents negated with the particle ńe stand in a syntactically marked focus position. The arguments below suggest that what we face with ńe-constructions is focus negation in the sense as it was defined for the purposes of this study (schematizable as [¬ Foc] [_presupp …], see Section 2 and point 1. of Section 6.1.2), and not sentential negation as with negative verb constructions:

1. i)When ńe negates a constituent, the main predicate of the clause is in the affirmative form as presented in Sections 3.5 & 5. In those sporadic cases when ńe immediately precedes the verbal predicate, negation refers to the verbal predicate and not to a lower sentence constituent and thus expresses sentential negation, cf. (20).
2. ii)The particle ńe is suitable for negating a constituent subordinated to a negative predicate, giving thus rise to a double negation construction as in (75)–(76). Negative verb constructions are not suitable to express double negation in the same simple clause.

I propose that the borrowing of the functional element ńe from Russian leads to the development of a focus projection in Udmurt. Thus, constituents negated with ńe are to be analysed as syntactic foci, appearing in the functional projection FocP dominated by a right-branching NegP projection as illustrated in (77):

Since the negative particle ńe always precedes the focus, it gives rise to transparent-scope constructions. An inverse scope reading of the string ńe + focus is out.