Russian differential object marking in Spanish speakers learning Russian as a second language

Full Text

Introduction

The term Differential Object Making DOM (Bossong, 1991) comprises a range of morphological, syntactic and pragmatic phenomena affecting the distinctive marking of direct objects.

In Russian, DOM surfaces as a genitive case in the place of the accusative/nominative marking on specific animate direct objects. DOM also encodes every adjective which modifies a noun marked with DOM. Example (1a) illustrates an animate direct object noun phrase (NP) marked with DOM, while (1b) illustrates an inanimate object NP displaying accusative/nominative case:

(1)          a.            Я             вижу     нового студента.           

                              Ja           vižu        nov-ogo               student-a.           (Russian – DOM)

                              I              see         new-acc/gen.m.sg           student- acc/gen.m.sg

                              ‘I see the/a new student.’            

b.             Я           вижу      новый      учебник

                              Ja           vižu        nov-yj    učebnik.               (Russian – No DOM)

                              I              see         new-acc/nom.m.sg          handbook.acc/nom.m.sg

                              ‘I see the/a new handbook.’

The phenomenon of DOM in Russian is independent of other syntactic mechanisms operating in the language, such as verbal agreement and word order. Here, we remain agnostic about a potential contrast in the structural position of Russian DOM vs. non-DOM objects. For the purpose of this study, we just assume that the presence or absence of an additional syntactic value, an [animacy] feature, differentiates the two types of objects^[1].

From the point of view of its morphology, Russian DOM is not homogeneous, i.e., it only affects masculine animate singular nouns of the first noun class, as well as plural animate nouns of all three declension classes existing in Russian (and the adjectives and low numerals that modify those nouns). From now on, we will gloss as DOM Russian accusative/genitive direct objects, as well as Spanish a-pre-positional direct objects.

In order to evaluate the vulnerability of incomplete acquisition of Russian DOM in the realm of the acquisition of Russian case, the present study measured the acquisition accuracy through an experiment, whose subjects were Spanish-speaking (young adult) learners of Russian as a second language (L2), living in Spain^[2].

The phenomenon of the acquisition of L1 Russian DOM in (monolingual and bilingual) children has been studied together with the acquisition of grammatical case in general by, i.a., A.N. Gvozdev (Gvozdev, 1961), M.D. Voeikova (Voeikova, 2015), N. Meir, M. Avramenko & T. Verkhovtceva (Meir, Avramenko, Verkhovtceva, 2021), A. Chrabaszcz, N. Ladinskaya & A. Lopukhina (Chrabaszcz, Ladinskaya, Lopukhina, 2023), etc. The only specific studies on the acquisition of DOM by L1 learners (Hrzica et al., 2015; Galkina, 2015) show that the age of acquisition is around 1:9 – 1:10^[3], analysing Liza and Vania from the CHILDES database. Only occasional errors are still found at a later age (3–9% between 2–4-year-olds). The errors are “both ways” (omission and commission), that is, inanimate objects marked with DOM (2a) as well as animate objects lacking DOM (2b), the latter type being much more frequent:

(2)          a.            Вот        этого     мячика надо     помыть обязательно.

                              Vot         èt-ogo   mjačik-a              nado      pomyt’  objazatel’no.

                              so           this-dom             ball-dom              needed  wash      necessarily

                              ‘We need to wash this ball.’ (adult = ètot mjačik)               (Liza 3:3)

       b.            Давай   медведь             причешем.

                              Davaj                    medved’              pričešem.                                         

                              let’s                      bear                      brush

                              ‘Let’s brush the bear.’ (adult = medvedj-a)                           (Liza 2:4)

Early acquisition of DOM is also observed in the case of L1 speakers of other languages, such as Spanish (Rodríguez-Mondoñedo, 2008), Hebrew (Uziel-Karl, 2015), Croatian (Hrzica et al., 2015), Lithuanian (Dabašinskiene, 2015) and Estonian (Argus, 2015)^[4].

There are several studies on the acquisition of Russian case by L2 learners (Kempe, MacWhinney, 1998; Artoni, Magnani, 2015; Cherepovskaia et al., 2022; Ceytlin, Kruglyakova, 2024), but none of them specifically studies the phenomenon of DOM. The most recent study, Cherepovskaia, Slioussar & Denissenko (Cherepovskaia, Slioussar, Denissenko, 2022) is an analysis of the whole trajectory of acquisition of the six Russian cases by L2 learners as compared to L1 learners. This study reports that the genitive case (which underlies the DOM syntactic position) displays the longest and most complex acquisition trajectory due to its special complexity and multiple uses.

To the best of our knowledge, there are no specific works on L2 acquisition of Russian DOM. However, this topic has often been addressed in the case of Spanish and Romance (Gerards, 2023, von Heusinger, Duarte, García, 2024) and, to a lesser extent, other languages (see e.g., the papers in (Mardale, Montrul, 2020)). Most of these studies have evidenced that even intermediate and advanced L2 students of Spanish exhibit certain difficulties in the use of DOM.

These difficulties have been specifically explored by Sorace (Sorace, 2011) in her Interface Hypothesis, who claims that the acquisition of those interface phenomena combining syntactic effects and pragmatic properties is vulnerable in incomplete acquisition, as compared to monolinguals, contrasting with purely syntactic phenomena, which are fully acquirable. As argued by, i.a. Montrul (Montrul, 2011, 2022), Spanish DOM is a complex phenomenon which lies at the interface of syntax, semantics and pragmatics. Spanish object marking is intertwined with a complex array of different semantic categories and pragmatic criteria.

In formal linguistics studies, grammatical case marking is assumed to be a double-sided phenomenon. Older generative approaches assumed a first operation, in which a syntactic head assigned an abstract case to an NP, and a second operation materialising that abstract case as a specific morphological marker (Chomsky, 1981 and subsequent work). In the minimalist programme, case valuing is parasitic on φ-feature valuing (Chomsky, 2000; Pesetsky, Torrego, 2001), speci-fically for Russian (Bailyn, 2004)^[5]. Thus, an NP enters a derivation with an unspe-cified case feature, later valued by a Probe in an Agree operation, which results in φ-feature valuing of the uninterpretable features of the probing head as well as valuing of the uninterpretable case feature of the NP.

Under these approaches, abstract or “syntactic” case can be realised phonologically or not as morphological case (Sigurðsson, 2003; Legate, 2008); specifically for Russian learning (Meir, Mitrofanova, Tomas, 2025). In languages like Russian, with a rich system of overt case morphology, the second operation is at work when performing case marking. On the contrary, in languages like Spanish, with almost no overt realisation of case, the second (morphological) operation is not relevant.

The twofold nature of case is not a universally accepted hypothesis and, rather, characterises the generativist viewpoint on grammar. Other, mostly usage-based approaches acknowledge the existence of case marking only when we “see” it, i.e., when it is realised morphologically (see e.g., (Nichols, 1992)), and do not accept the existence of an abstract level of case.

In fact, native speakers of languages with overt case morphology homogeneously use syntactic and morphological case as indistinguishable units. The aforementioned authors, who described the acquisition of Russian case by first language (L1) learners, concluded that syntactic and morphological case go hand-in-hand amongst native speakers from their first productions.

In the realm of this discussion, the first goal of this study is to explore the idea that incomplete acquisition of a language with overt case morphology can reveal asymmetries corresponding to the twofold nature of case marking in the language, which are undetectable in regular native speakers. If such asymmetries exist, they should reflect the two operations involved in Russian case marking: namely, the acquisition of the syntactic features (detecting a syntactic position) and the acqui-sition of the corresponding morphological marker(s).

Experimental work evidencing an abstract level of syntactic operations has been conducted before, mostly concerning subject-verb agreement and nominative case matching. In the same vein, in this paper, we aim to contribute experimental evidence to the idea that there is a contrast between abstract and overt operations in case marking in incomplete acquisition.

Furthermore, we aim to relate the twofold nature of case to the interface phenomena, the incomplete acquisition of which has been addressed before by Sorace (Sorace, 2011), in her Interface Hypothesis, and later work. Thus, as a second goal, the present study aims to check whether acquisition (in a broad sense) of Russian DOM is especially vulnerable in L2 learners due to the complex syntax-morphology interface that characterises this phenomenon (for a similar approach in the case of the syntax-semantics interface, cf. (Gondra, 2022))^[6].

We split this idea into two more fine-grained hypotheses, in order to explore this idea in specific ways. Our first hypothesis is based on Lardiere’s (Lardiere, 2017) Feature Reassembly Hypothesis, also partially present in Perez-Cortes, Putnam & Sanchez’s (Perez-Cortes, Putnam, Sanchez, 2019) Feature Activation Hypothe-sis. Following these hypotheses, Spanish speakers, learners of Russian as a second language (L2) should be aware of the special syntactic status of animate objects, because the same feature is also relevant for direct object marking in Spa-nish, i.e., is syntactically active in their L1, and should be acquired and put to use more readily. In this sense, we should not expect any contrast in Spanish speakers, learners of Russian as a second language (L2) in the detection of the syntactic position of DOM vs. other syntactic positions, which are also available in Spanish.

Our second hypothesis is based on Haznedar’s (Haznedar, 2003) Missing Surface Inflection Hypothesis, according to which the variable use of morphological forms in L2 grammars reflects a problem with the realization of surface morphology, rather than an impairment in the domain of features or functional projections. In line with this hypothesis, we expect that, albeit the feature [animacy] should be readily available for Spanish speakers, learners of Russian as a second language (L2), its exact mapping into the actual morphological form corresponding to each specific Russian word should not be straightforward. This is not only because of the variety of forms in Russian, but also because of the cue reliability problem (Kempe, MacWhinney, 1998), which DOM inflection implies. On the one hand, DOM inflection is multifunctional, i.e., genitive case markers do not characterise solely DOM, but also other syntactic positions (adnominal genitive, genitive of negation, quantificational / adnumeral genitive, etc.). On the other hand, direct cases in Russian produce nested paradigms across noun classes, which are expected to hinder the acquisition of the actual morphological exponents, even if the syntactic function of the morpheme can be quickly detected. In other words, even if the acquisition of Russian case by L2 learners must posit difficulties in general terms, because of its complex morphology, the especially complex interface between syntax and morphology in the specific case of DOM should make these difficulties more evident.

Methods and materials

In order to assess the level of vulnerability of the acquisition of Russian case in L2 young adult learners from Spain (university students) as compared to native speakers of Russian, we measured the accuracy in the two groups independently.

The participants performed a production task. The task was a standard exercise in the Russian as a Second Language class, ensuring that students encountered no difficulties in completing it and required no specific prior training. The instruction given was: “Fill in the gaps by selecting the correct preposition, verbal form or aspect, and the appropriate case and number.” Thus, the subjects were asked to give the right case form of several nouns in different syntactic positions (the target stimuli), the right aspectual form of various verbs and the use of prepositions (the last two tasks as fillers). In the target stimuli, a noun phrase was given in brackets in the default nominative case, and the students had to (i) detect the right syntactic position of the case; and (ii) choose the right morphological exponent for the relevant grammatical case.

The task design involved a careful selection of a balanced number of different instances of Russian DOM, emphasizing those most relevant to Spanish-speaking learners (see “Discussion” below). First, we determined that the task would include 40 stimuli, with half (20) serving as fillers. Of the remaining 20 stimuli, 10 were designated as target stimuli (sentences involving the direct object case), while the other 10 functioned as control items to compare participants’ proficiency in using the direct object case versus other grammatical cases.

The 10 target stimuli included 6 instances of DOM. The selection of nouns was based primarily on frequency and simplicity from the perspective of foreign learners. Due to occasional discrepancies in lexical proficiency among students, we opted for words at the A1–A2 levels of the Common European Framework of Reference for Languages (CEFR), following Leksičeskij minimum po russkomu jazyku kak inostrannomu. Bazovyj uroven’. Obščee vladenie (2013, Saint Petersburg: Zlatoust). There were two exceptions: the words for large and small animals (volk and tarakan), which appear only at the B1 level (pervyj sertifikacionnyj uroven’), as outlined in Leksičeskij minimum po russkomu jazyku kak inostrannomu. Pervyj sertifikacionnyj uroven’. Obščee vladenie (2014, Saint Petersburg: Zlatoust). This exception was made because no masculine-gender animals are introduced at the A1–A2 levels. We aimed to assess participants’ knowledge of this particular feature, as it represents one of the key points of divergence between Spanish and Russian DOM.

The remaining 10 stimuli involved other syntactic positions requiring other grammatical cases, namely, accusative, dative and genitive case after prepositions, genitive of quantification, dative of recipients and experiencers, instrumental denoting instruments.

Finally, 20 additional items referred to other grammatical and morphological phenomena (use of prepositions and verbal forms), i.e., they fulfilled the function of fillers and helped to assess the participants’ general command of Russian grammar.

The participants: the control group: The control group included native speakers of Russian living in Russia (31 subjects, mean age M_age = 42.3, standard deviation of age SD_age = 12.5). The control group was surveyed using the crowdsourcing platform Toloka (URL: toloka.ai). The task performed by the native speakers and the L2 students of Russian was exactly the same. The L2 students filled out the task by hand on paper, while the control group filled it out online, accessing the task through a link. The participants signed informed consent for personal data collection.

We took into account 31 surveys out of a total of 32; specifically, we discar-ded the responses of one participant, who was not a native speaker of Russian. The vast majority of the participants in the control group consisted of Russian monolinguals, while only 3 people reported knowing English (as an L2) and one Belarusian.

The participants: the target group (L2 learners of Russian): The experiment was carried out among third-year undergraduate Spanish-speaking students of the Russian language at the University of the Basque Country in the course of three scholar years (2021–2022, 2022–2023, 2023–2024). There were 46 participants in total; 10 surveys were discarded due to an extremely low accuracy or lack of answers, so 36 participants were eventually taken into account.

The participants were 20–22-year-old native speakers of Spanish (36 subjects, mean age M_age = 20.4, standard deviation of age SD_age = 0.5). The dominant language of these speakers was Spanish, but some of them also knew Basque at different levels. In the preliminary questionnaire 17 (out of 36), i.e. 47,2% of the participants declared themselves as speakers of Basque. In addition, all the participants were students of the degree in Translation and Interpretation, and had English as their first L2 language, in which all of them were highly proficient. The knowledge of Basque or English was irrelevant for this study as these languages lack DOM or any similar phenomenon.

Russian was the second language for these speakers; they had started learning Russian from zero level in their first year of university. When the task was carried out, they had studied the Russian language for approximately 85 weeks (a total amount of about 340 hours), and had on average an A2 level, with a few exceptions. The differences in proficiency which we observe in the responses must be attributed to the personal linguistic abilities of each student and their personal commitment to studying the language.

Results

The criterion to measure accuracy in the use of case was the total number of convergences with the linguistic norm, confirmed by the answers of the control group of monolingual native speakers of Russian, living in the Russian Federation.

The total number of target responses obtained was 719, distributed in the following way: 216 responses corresponding to the DOM condition and 503 to other case positions (table 1). Here are the counts and raw percentages of (in)correct answers in the two groups of subjects, L2 students and native speakers^[7].

Table 1
Accuracy — L2-students

Position	Condition			Control group: position + morpheme correct
	Only syntactic case position correct	Both case position and morpheme correct	Incorrect answers
DOM objects	26 (12.04%)	99 (45.83%)	91 (42.13%)	216 (100%)
Other positions	68 (13.49%)	289 (57.34%)	147 (29.17%)	503 (99.78%)
Total	94 (13.06%)	388 (53.88%)	238 (33.06%)	719 (99.86%)

Source: completed by N. Madariaga using Microsoft Excel.

The accuracy rates in the group of L2 students are lower than the control group, most notably in the case of full convergence with both the syntactic positions and morphological exponent (almost 54% on average). The contrast between the “only case position” and “case position + morphology” criteria suggests that the students sometimes know which case corresponds to a syntactic position, even if they do not know which form is the correct one (number of convergences only in the syntactic position), albeit not very often (13% on average).

Discussion

Detailed description of the experiment: The experiment was designed according to the idiosyncrasies of the Russian case system. First, we describe the difficulties that it posits to L2 learners, speakers of Spanish (asymmetries and morphological difficulties), and then, the differences in the use of Russian DOM as compared to Spanish DOM.

The noun classes / numbers affected by DOM in Russian do not have their own accusative case form; thus, accusative is syncretic either with the genitive (on animate nouns = DOM) or with the nominative case (on inanimate nouns); see example (1) above. In other words, Russian accusative syntactic case is a morphologically dependent case (Zaliznyak, 1973) as it does not correspond to morphological segments that would be unique in the Russian case paradigm (see also (Shvedova, 1980)). These syncretisms produce the nested paradigm represented in Table 2 below. Within plural number, the morphological exponents of genitive case are especially complex, and vary according to declension paradigms and other additional considerations.

Table 2
Morphological paradigm of Russian direct cases (masculine singular first noun class and all plural)

Gender, number, animacy		Case
		Nominative	Accusative	Genitive
Masculine singular 1st noun class	Inanimate	шø		-a/’a
	Animate	ø	-a/’a
Plural (all noun classes)	Inanimate	-y / -i / -a /’a		-ov / -ev/ -ej / ш
	Animate	-y / -i / -a /’a	-ov / -ev/ -ej / ш

Source: completed by N. Madariaga using Microsoft Word.

As opposed to the forms in Table 2, we show in Table 4 that singular nouns of the second noun class have their own accusative case (mam-u ‘Mum-ACC’), which is non-syncretic with either nominative (mam-a ‘Mum-NOM’) or genitive case (mam-y ‘Mum-GEN’), whereas the accusative singular of third class is always syncretic with the nominative (mat’ ‘mother-NOM/ACC’), regardless of the animacy of the noun.

Table 3
Morphological paradigm of Russian direct cases (feminine singular)

Gender, number, animacy		Case
		Nominative	Accusative	Genitive
Feminine singular 2nd noun class	Inanimate	-a/’a	-u/’u	-y / -i
	Animate
Feminine singular 3rd noun class	Inanimate	ø	ø	-i
	Animate

Source: completed by N. Madariaga using Microsoft Word.

Nested paradigms of object marking emerge not only across noun classes but also across grammatical functions. In practice, a DOM position imposes on learners an additional “problem”, besides the matching of the animate feature and the correct morpheme, which is the choice of the correct grammatical case, nominative or genitive, in contrast to dedicated accusative.

As for grammatical cases related to other syntactic positions, let us consider the range of available forms in dative case as an example (Table 5). From the point of view of its morphological exponents, the picture here is similar to dedica-ted accusative case, represented in Table 4; only one or two forms corresponding to one class & number, and no animacy effects.

Table 4
Morphological paradigm of Russian dative case

Noun class	Number
	Singular	Plural
1st noun class	-u/’u	-am/’am
2nd noun class	-e
3rd noun class	-i	-’am

Source: completed by N. Madariaga using Microsoft Word.

This paradigm, of course, displays morphological complexity, as compared to Spanish, but here the array of available forms is not as wide as in the case of DOM.

In order to understand the asymmetry in the number of available forms of DOM vs. other positions, the most important consideration is that, in the DOM position, the speaker has to know not only which case must be used (i.e. identify the syntactic position), but also the correct morpheme (i.e. the morphological exponent), depending, again, on the gender/class and number of the noun (nominative, genitive or dedicated accusative), whereas in the other case syntactic positions only one non-syncretic case can be used, be it dative, genitive, or instrumental (cf. (Plungian, 1998) on the difference between syntactic and semantic understanding of case).

In Spanish, the phenomenon of DOM affects animate (definite) direct objects, regardless of the gender and the number of the noun. The morphological pattern is very simple; DOM merges animate direct objects (3a) with indirect objects, typically introduced by the preposition a ‘to’. This pattern contrasts with inanimate direct objects and indefinite objects, which generally lack the preposition a (3b).

(3)          a.            Veo        al            nuevo    estudiante.                         (Spanish – DOM)

                              I see       a-the     new       student               

                              ‘I see the new student.’ 

               b. .            Veo        el            nuevo    manual.                              (Spanish – No DOM)

                              I see       the         new       handbook                          

                              ‘I see the new handbook.’

Spanish and Russian have in common the presence of DOM in (at least certain) animate objects. However, the correspondence of patterns is not univocal.

The first difference concerns the semantic conditions of DOM. Whereas the crucial semantic feature conditioning DOM in Russian, i.e., animacy, is shared with Spanish DOM, the other relevant semantic features which overlap with animacy in Spanish DOM are absent in Russian.; most notably, definiteness (DOM: Busco a-l secretario nuevo ‘I search dom-the new secretary’ ≈ No-DOM: Busco un secretario nuevo ‘I search a new secretary’)^[8]. We will disregard these additional semantic features here because of their absence in the Russian pattern; thus, in the corresponding experiment we tested only definite direct objects.

The second difference concerns animacy itself. In Russian, every animate object which fulfils certain specific morphological requirements, surfaces with DOM. Animate nouns range from proper names to small animals (including insects) (4a), while inanimate nouns include everything else (4b).

(4)          a.            Я вижу Ивана / студента /          медведя /          таракана.

                              Ja vižu   Ivan-a  /               student-a  /        medvedj-a  /       tarakan-a.

                              I see                      Ivan-dom             student-dom       bear-dom            cockroach-dom

                              ‘I see Ivan / a/the student / a/the bear / a/the cockroach.’

             b.             Я             вижу     дуб /     сахар.

                              Ja           vižu        dub  /    sahar.

                              I              see         oak        sugar

                              ‘I see an/the oak / (the) sugar.’  

In contrast, Spanish allows optionally not applying DOM in the case of non-domestic animals (5a), and it does not apply DOM to small animals or insects, if they are not emphatically individualized (5b) (Aissen, 2003):

(5)          a.            He          matado               a-l                         / el         lobo.

                              aux         killed                     dom-the              / the      wolf

                              ‘I killed the wolf.’

             b.            He          matado ^??a la                    / la         cucaracha.

                              aux         killed                     dom the              / the      cockroach

                              ‘I killed the cockroach.’

The experiment was designed taking into account all these features. The 10 target stimuli included 6 instances of DOM: (i) Masculine: one singular noun denoting a large animal, one singular noun representing a small animal (insect), one singular personal noun, one plural noun with a palatalized ending, and one plural noun with a non-palatalized ending. (ii) Feminine: one plural noun. Additionally, there were 4 non-DOM objects: 2 masculine nouns (one singular, one plural) and 2 feminine accusative singular nouns (one inanimate, one animate). The specific lexical items selected were: volk ‘wolf’, tarakan ‘cockroach’, prepodavatel’ ‘teacher’, roditeli ‘parents’, student ‘student’, sëstry ‘sisters’, rasskaz ‘story’, karandaši ‘pencils’, rodina ‘homeland’, mama ‘Mum’.

Here are some examples of the stimuli testing singular masculine human DOM (6a), plural human DOM (6b) and animal DOM (6c).

(6)          a.            Студенты            внимательно    слушают             преподавателя.

                              Studenty              vnimatel’no        slušajut prepodavatelj-a.

                              students               carefully              listen                    teacher-dom

             b.            Вчера я видел   его                        сестёр.

                              Včera ja               videl      ego                       sestër.

                              yesterday I          saw        his                         sisters.dom

             c.            Я             поймал маленького                      таракана.

                              Ja           pojmal                  malen’k-ogo                      tarakan-a.

                              I              captured              small-dom                         cockroach-dom

A couple of examples of stimuli testing non-DOM positions: inanimate masculine nouns (7a) and regular accusative direct objects of second noun class, singular number (7b):

(7)          a.            Студенты            вернули              на место             все         карандаши.

                              Studenty              vernuli na mesto             vse         karandaš-i.

                              students               returned              to place               all.nom pencils-nom

             b.            Все        дети      любят   маму.

                              Vse         deti        ljubjat    mam-u.

                              all           children                love       mum-acc

Measuring accuracy in the answers: As explained before, Russian DOM is characterised by an [animacy] φ-feature, which must be matched in the course of the derivation. We consider that performing this operation successfully equates to “matching the syntactic value of DOM”. However, together with identifying the syntactic position of DOM, speakers must also choose the correct grammatical case plus the specific morphological exponent or case ending within the corresponding noun class. We define this second step as “choosing the correct DOM morpheme” in each specific instance or word.

This “two-step” procedure was used to assess accuracy of use of case also in the other syntactic positions tested, namely: accusative, dative and genitive case after prepositions, genitive of quantification, dative denoting recipients (indirect objects), dative denoting experiencers, non-DOM direct objects, and instrumental denoting the instrument. All these positions are shared with Spanish from the syntactic point of view, but the Russian morphological exponents are much more complex.

Let us take as an example the dative case of experiencers. Spanish and Russian share the corresponding syntactic position, but the morphological array of variants is wider in Russian. As shown in (8a), all experiencers are marked in Spanish with the preposition a ‘to’, whereas Russian displays several forms, depending on gender, number and noun class (8b):

(8)          a.            A la niña              / a-l niño              / a las niñas       les          gustan    los gatos.

                              to the girl            / to-the boy        / to the girls                      them     like.3pl the cats

             b.            Девочке              / мальчику         / матери             / девочкам         нравятся               кошки.

                              Devočk-e             / mal’čik-u           / mater-i            / devočk-am       nravjatsja                koški.

                              girl-dat / boy-dat             / mother-dat / girls-dat like.3pl cats

                              ‘The girl / the boy / the mother likes / the girls like cats.’

Spanish and Russian share the basic syntactic values for DOM and the other case positions tested, but Spanish lacks case morphology and noun classes. Spanish-speaking learners must overcome the challenge of picking the right morphological form for each syntactic position.

In the specific case of DOM, three challenges must be overcome by L2 learners: they must (i) detect the presence or absence of an animacy feature in the direct object; (ii) know that only animate direct objects of certain numbers and noun classes surface as genitive case, as opposed to nominative or dedicated accusative case; and (iii) be able to choose the specific morphological exponent encoding genitive case of the corresponding noun class (as opposed to other genitive endings, corresponding to other noun classes). In other words, the morphological complexity of Russian case is intertwined in the case of DOM with an additional choice of the correct grammatical case corresponding to the specific ([animacy]) syntactic position.

Additionally, Spanish speakers, learners of Russian as a second language (L2) have to know where the animacy hierarchy is “cut off” in the case of Russian DOM, i.e. that every animate object (including little insects) undergoes DOM, unlike in Spanish, in which DOM is optional or dispreferred with some animals; see example (5). This effect was also tested in our experiment.

In regard to the three challenges discussed above, we were able to control only two of them, with help of the following types of answers: (i) the participant chooses a genitive form for a DOM position, but not the right morpheme; and (ii) the participant chooses a genitive form for a DOM position + the right morpheme for the specific word at stake, achieving the objectives outlined successfully. That is, we can affirm with certainty that a participant carried out successfully the three operations described above (correct syntactic position + morphological form) or that the participant carried out successfully only the first two operations described (correct syntactic position + correct morphological case – genitive).

Taking into consideration only these two scenarios implied discarding those answers with anything other than these two possibilities, i.e., random endings (cases other than genitive, nominative and accusative). Therefore, we took into account three types of answers:

1. When a subject chooses genitive case for a DOM position (I see parents-gen), no matter whether the genitive form is morphologically correct or not, they were considered to be successfully matching the “syntactic” value of DOM. In case the genitive morpheme chosen was not felicitous (e.g., roditel-ev or roditel-ov instead of the correct roditel-ej ‘parents-gen’), we ticked the answer as “only matching the syntactic value”, but not the morphological form. In case the NP included an adjective, and the adjective was correctly marked as genitive, but not the noun (e.g. malen’k-ogo tarakan-u ‘small-gen cockroach-dat’), we ticked the answer as matching only the syntactic position.
2. Answers with the correct genitive form in DOM position (e.g., roditel-ej ‘parents-gen’) were marked as succeeding at both syntax and morphology.
3. Finally, a nominative form in a DOM position (e.g., roditel-i ‘parents-nom’) was marked as failing at both syntax and morphology.

Other case markers in DOM position (locative, instrumental and so forth) were almost inexistent in the tests; the corresponding participants showed an extremely low command of the language, and their tests were discarded.

Equally, the other case positions were assessed under this three-part system. Following with the example of dative case marking of experiencers, we distinguished three types of answers:

1. Successful matching of case position (syntactic case) when the participant chooses a dative exponent (teacher-dat likes teaching), but not the one that corresponds to the gender, number or noun class needed (e.g. prepodavatel-e or prepodavatel-i instead of the correct prepodavatel-ju ‘teacher-dat’).
2. Successful matching of both case positions and morphological exponent (e.g. prepodavatel-ju ‘teacher-dat’).
3. Unsuccessful matching of both criteria when the case chosen was not the correct one, in our example, any case other than dative (e.g. prepodavatel’ ‘teacher.nom’, prepodavatel-ja ‘teacher-gen’, etc).

As a final remark, we must not lose sight of the fact that, while in other case positions there is a univocal correspondence between the syntactic position and the grammatical case used (e.g. all recipients are marked with dative case), the DOM position demands an additional effort from learners. Namely, the presence of an animacy feature on a direct object does not automatically imply genitive case but, depending on the class and number of the noun, it fluctuates between nominative, genitive and dedicated accusative cases. The errors made by L2 learners in this last association were undetectable in a separate way, and fall under the graph “incorrect answer” in our study.

The answers obtained in the exercises were disaggregated according to the two criteria explained before, and were ticked as one of the following three types:

(i) “Both case position and morpheme correct”: the participant predicted correctly the case position and the morphological form chosen had to be the right one, as in example (9a), in which the form chosen was the DOM zero-morpheme (genitive case).

(ii) “Only syntactic case position correct”: this type of answer evidenced syntactic accuracy with infelicitous morphology. The errors made by the participants in the choice of the morpheme, while the syntactic position is correct, are illustrated in (9b).

(iii) “Incorrect answer”: the participant failed to predict the correct case position and, therefore, also failed to use the correct morpheme. An answer of this type is illustrated in (9c).

(9)          a.            Вчера    я             видел                  его         сестёр.

                              Včera    ja            videl                    ego        sestër

                              yesterday            I              saw                      his          sisters.dom

             b. Вчера    я             видел                  его         *сёстр / *сестров / *сестрей.

                              Včera    ja            videl                     ego        *sëstr / *sestr-ov / *sestr-ej.

                              yesterday            I              saw                       his          sisters-dom       

             c. Вчера    я             видел                  его         *сёстры.

                              Včera    ja            videl                     ego        *sëstr-y.

                              yesterday            I              saw                       his          sisters-nom       

                              ‘Yesterday I saw his sisters.’

In other words, when counting the deviant answers, for every syntactic position, we dissociated overall accuracy (syntactic position + morphological exponent correctly detected) vs. syntactic accuracy only. In the specific case of DOM, there was no way to detect in an independent way the failure in associating animacy and genitive case only in certain noun classes and numbers. Therefore, the potential errors in making this association fell under the graph “incorrect answer”.

We discarded 10 responses; among them, 4 participants seemed to choose randomly the case endings, e.g. non-objectual morphological forms for DOM positions, as in Včera ja videl ego *sëstr-ami ‘yesterday I saw his sisters-inst’, a type of answer absent in the other participants^[9]. This, together with an extremely low command of Russian grammar in general (regarding both the use of case and other grammatical aspects, included in the experiment as fillers), led us to discard these participants. Another 6 participants, who left most of the items unanswered, were also discarded.

Statistical analysis: To analyse the data, we first fit a multinomial logistic regression model with the three possible answers (“correct case position only” / “correct case and morphology” / “incorrect”) as the outcome variable with fixed effect of syntactic position (DOM position / other case positions). The overall model is significant (LR-statistic = 11.43, df = 2, p = 0.0033), however, the model fit is far from decent in terms of Nagelkerke’s R² (0.018). We could conclude that DOM vs. non-DOM distinction does not influence whether the learner correctly identifies the case-related syntactic position or morphology taken separately. Remember that Spanish speakers, learners of Russian as a second language (L2) equally have the syntactic equivalent of DOM and the other case syntactic positions in their own language, and that Russian morphology is equally complex for every condition as compared to Spanish. However, we examined the model further: following (Gries, 2021), we computed the accuracy of the model which is unsatisfactory as the model always predicts that subjects provide both correct syntactic position and correct morphology.

Next, we analysed the error rates in a greater detail. In non-DOM positions we observe a higher number of fully correct answers and a lower number of totally incorrect answers than in DOM position (χ2 (1, 626) = 10.7, p-value = 0.0011). In other words, most participants either know well the use of DOM or do not know it at all, whereas the knowledge of other syntactic positions is better in general terms, and not so polarized among the participants.

Likewise, looking at the variation in accuracy among the individual L2 learners, the distribution of answers with only correctly identified syntactic position among speakers is different in DOM vs. other case positions. This type of answers corresponds to a correct grammatical case in terms of syntax and incorrect morphological form. As shown in Table 5, in the DOM condition, subjects are more likely to have errors of this type, while for other cases the number of errors in a single subject’s response is typically lower (χ2 (4, 72) = 17.72, p-value = 0.0033). That is, in case of DOM errors in morphology with correct syntax are less spread among the L2 learners than in case of other positions at an individual level (whereas the average global figures are similar in both conditions).

Table 5
Distribution of different numbers of answers with only syntactic position correct among subjects’ responses

Numbers of answers showing correct case (only syntactic position)		0	1	2	3	4	5+
How many subjects gave this number of correct answers	DOM	17	14	4	0	1	0
	Other	6	9	10	6	4	1

Source: completed by N. Madariaga using Microsoft Excel.

Let us consider further the variation in accuracy among the speakers (the distribution of answers). As shown in Figures 1 and 2 below, the correct answers and errors among the L2 learners are unequally distributed, with many differences between the participants and atypical values, especially in the case of the DOM condition (Table 5 and Figure 1). In particular, we observe more variance in the way the average accuracy of speakers is distributed for answers with only correct syntactic position (SD = 7.27 in DOM condition vs. SD = 3.62 in Other positions condition). In case of incorrect answers there is also more variance in case of DOM (SD = 3.53 vs. SD = 2.53), while the distribution of accuracy in case of totally correct answers does not show much difference (SD = 1.57 in DOM condition vs. SD = 1.88 in Other positions condition).

Figure 1. Average accuracy (percentages) in DOM position
Source: completed by A. Gerasimova & E. Lyutikova using RStudio.

Figure 2. Average accuracy (percentages) in other positions
Source: completed by A. Gerasimova & E. Lyutikova using RStudio

Figures 1 and 2 show that the distribution of answers among the participants is rather uneven, evidencing the differences in proficiency between them. These figures suggest that the number of errors only in the DOM condition is more evenly distributed among the subjects than in other case positions. In plain words, more students, even those who performed better in general terms, could have errors in DOM, whereas the errors in the morphology in other case positions were more “concentrated” in those subjects who performed worse in general terms.

Observations about the results: (a) In regard to the vulnerability of incomplete acquisition of Russian case, in general terms, it is compromised in the group of speakers under analysis, as compared to the control group of native speakers of Russian living in Russia. If we compare the case positions under study here, in the matter of both the purely syntactic matching and the choice of the right morphological exponent, taken separately, L2 learners perform equally worse than native speakers, regardless of the syntactic position (DOM vs other positions) The differences between the control group and the group of L2 learners is more prominent in the case of DOM positions (less so in other syntactic positions) when we focus on the number of completely incorrect answers as compared to the number of completely correct answers, as well as the distribution of morphologically incorrect answers among the speakers. This can be attributed to the fact that the Russian DOM constitutes a syntax-morphology interface phenomenon of special complexity. The diversity of morphemes, nested paradigms and homophonic forms of Russian case in DOM position, limits cue reliability for the acquisition of this specific phenomenon in L2 learners. Therefore, our hypothesis of the “extended” versions of Lardiere’s (Lardiere, 2017) Feature Reassembly Hypothesis and Sorace’s (Sorace, 2011) Interface Hypothesis is borne out. Russian case, especially the DOM phenomenon, as a morphology-syntax interface phenomenon makes acquisition vulnerable in L2 learners.

(b) As for the twofold nature of case, the asymmetries in accuracy under conditions of incomplete acquisition of Russian are partially confirmed. For these asymmetries, we had to pay attention to the participants’ ability to choose the right morphemes, as compared to their accuracy in choosing only the right syntactic case position or value. On the one hand, in line with Haznedar’s (Haznedar, 2003) Missing Surface Inflection Hypothesis, Spanish speakers, learners of Russian as a second language (L2) perform worse concerning morphology compared to syntax in every condition. The number of “only syntactic” case matching instances is not very high (on average, 13% of all the answers; almost 25% of all correct answers). However, the mere existence of this type of answers, its consistent and recurrent character in the surveys, and its evenly distribution between the participants, contrasting with the total absence of this asymmetry in the control group, suggests that L2 learners can display difficulties in matching syntactic positions with the corresponding morphological exponent. That is, sometimes L2 learners cope with the task of detecting a syntactic position, but fail in choosing the corresponding morphological exponent. The shared existence of all the syntactic positions under analysis in Russian and Spanish, including the DOM position, makes irrelevant the syntactic position for the distinction between syntactic and morphological case in these learners, and renders a similar percentage of answers of the type “only syntax correct” for both DOM and other positions in the surveys.

(c) For further research, we would like to explore an idea implied by the results presented here. The statistical analysis has shown that in the case of DOM, Spanish speakers, learners of Russian as a second language (L2) are more likely to make errors both in general terms and in the case of individual, more proficient learners, as compared to other case positions. This result suggests that the L1 background does not necessarily contribute to speeding up progress along the developmental path, and the potential transfer effect from the majority language is probably not relevant. In fact, their L1/societal language, Spanish, is a language endowed with a pragmatically and semantically complex system of DOM, but very poor morphology. The observed rates of accuracy in using DOM correlate with the speakers’ morphological proficiency rather than the existence of DOM in their L1. To verify this intuition, it seems promising to compare Spanish speakers, learners of Russian as a second language (L2) with learners with a different first language (L1), namely, an L1 without DOM, in order to check whether the presence or absence of DOM in the L1 is significant for its acquisition in Russian as an L2.

Conclusion

We provided experimental evidence about the twofold nature of the grammatical case by revealing certain asymmetries in L2 learners when it comes to accuracy in detecting a syntactic position / value and matching it with the correct morphological exponent. In line with Haznedar’s Missing Surface Inflection Hypothesis, Spanish speakers, learners of Russian as a second language (L2) may display difficulties in producing morphological case, due the lack of rich morphological case in their own language.

Even though DOM is present in both Russian and Spanish, the corresponding patterns display morphological and pragmatic differences in the two languages. The most salient differences are (i) the complex array of morphological exponents (similarly to other cases in Russian), and (ii) the non-univocal correspondence between the syntactic position of direct objects and one grammatical case (unlike other case positions in Russian), which interface with the specific syntactic criteria of Russian DOM. The experiment that we conducted shows that acquisition of Russian case, especially DOM, is significantly vulnerable in Spanish speakers, learners of Russian as a second language (L2) when we focus on the overall number of errors, and the errors are more evenly distributed among learners, as compared to other case positions, which concentrate in less proficient learners.

Even if accuracy in the use of case is conditioned by the general language proficiency of the participant and her/his general command of Russian morphology, the DOM position was slightly more difficult for our participants. We observed this effect (i) when we focused on variation of the distribution of errors among the subjects and (ii) when we compared the completely correct answers and the completely wrong answers. The analysis of the data according to criterion (i) shows that errors in assigning the correct morphological case to the DOM position are found more frequently than in other syntactic positions. The analysis corresponding to criterion (ii) shows that L2 learners have more errors when using DOM in general terms, as compared to other syntactic positions. This is consistent with the hypothesis that the syntax-morphology interface can compromise incomplete acquisition of DOM, because DOM displays an especially complex morphology-syntax interface.

Thus, we suggest that we could extend Lardiere’s Feature Reassembly Hypothesis to the syntax-morphology interface; in other words, not only the syntax-pragmatic interface but also an especially complex morphological layer can compromise the acquisition of a phenomenon whose syntax is more or less easily acquired by incomplete learners.

 

 

^{1 Within formal approaches to syntax, structural asymmetries related to a positional contrast between DOM vs. non-DOM objects have been reported for other languages, such as Tatar (Lyutikova, Pereltsvaig, 2015), Hindi (Bhatt, Anagnostopoulou, 1996) and Spanish (Camacho Ramírez, 2022). In the case of Russian DOM objects, a structural asymmetry is more difficult to grasp, but still, some contrast suggests that DOM and non-DOM objects may be embedded into distinct syntactic layers. For example, non-DOM plural objects can be easily combined with stranded Floating Quantifiers, while DOM objects not so readily (Madariaga, 2007: 271–272).}

^{2 Throughout the paper, we will refer to these speakers as “L2 learners / speakers” or, more specifically, Spanish speakers, learners of Russian as a second language (L2), although all the subjects have a “first L2 language” (English), and some have Basque as a second L1 (see Section 4.2). We stick to the denomination “L2” rather than “L3/L4/L5…” learners for the sake of simplicity.}

^{3 The age of acquisition is expressed according to the notation “year:month”.}

^{4 The idea that morphological complexity and form frequency affect the acquisition of DOM has been tested in the case of L1 learners of Estonian (Vihman, Theakston, Lieven, 2020), a flexional language with a similar DOM system to that in Russian.}

^{5 The φ-features are considered to be person, number, gender and animacy. The last feature will be most relevant for this study.}

^{6 We often refer to the participants in this study as representative of “incomplete acquisition”. We use this denomination in line with the tradition that focuses on the differences these speakers display from a baseline control group of L1 monolinguals.}

^{7 The 31 adult native speakers that formed the control group performed without a single error in all the trials including direct object marking (100% of convergence). As for other grammatical cases, there was only one error made by one speaker, namely, an accusative form instead of the correct dative: Devočk-u nravjatsja koški ‘girl-ACC like cats’. In a few cases, the participants wrote the correct form, both syntactically and morphologically, but in a different grammatical number. These cases are not taken as errors, because the case forms chosen were correct.}

^{8 Also other semantic features, such as agentivity, telicity, disambiguation, left-dislocation and the presence of a modifier (see (Fabregas, 2013) for a complete review).}

^{9 A similar effect was observed by Artoni & Magnani (Artoni, Magnani, 2015: 188–189) in their study on the acquisition of Russian case by L2 learners. In their experiment, they had one example of DOM, the animate noun volk ‘wolf’ in direct object position, and they reported than only one learner (out of 8) used a case different from the nominative volk or the genitive volka.}

About the authors

Nerea Madariaga

University of the Basque Country UPV/EHU

Author for correspondence.
Email: nerea.madariaga@ehu.es
ORCID iD: 0000-0003-4732-4144
SPIN-code: 9800-6563
Scopus Author ID: 42861993600
ResearcherId: K-9871-2013

PhD in Linguistics, Associate Professor at the Department of Classical Stu-dies, Coordinator of the Section of Slavic Philology

6 Paseo de la Universidad, Faculty of Arts, Vitoria-Gasteiz, 01006, Spain

Anastasia A. Gerasimova

Lomonosov Moscow State University

Email: anastasiagerasimova432@gmail.com
ORCID iD: 0000-0003-4686-5598
SPIN-code: 1327-5805
Scopus Author ID: 57194502949
ResearcherId: Q-6419-2016

Ph.D. in Philological Sciences, Researcher at the Research Computing Center

1 Leninskie Gory, bldg. 4, Moscow, 119234, Russian Federation

Ekaterina A. Lyutikova

Lomonosov Moscow State University

Email: lyutikova2008@gmail.com
ORCID iD: 0000-0003-4439-0613
SPIN-code: 2108-4296
Scopus Author ID: 55660727100
ResearcherId: P-2563-2016

Prof. Dr. Habil. in Philological Sciences, Professor at the Department of Theoretical and Applied Linguistics at the Philological Faculty

1 Leninskie Gory, bldg. 51, Moscow, 119234, Russian Federation

References

Supplementary files