On the use of Russian and Dutch WH-Questions and Relative Clauses by Bilingual Dutch-Russian Children

(1)

On the use of Russian and Dutch

WH-Questions and Relative Clauses by

Bilingual Dutch-Russian Children

MA thesis by Nadine Hut

University of Amsterdam

Linguistics of European Languages – Slavonic

Supervisor: Mrs. dr. A.V. Peeters-Podgaevskaja

Second Reader: Mr. Prof. dr. W. Honselaar

Contact: nadinehut@me.com

(2)

Acknowledgements

Writing this thesis proved to be a bigger task than expected beforehand. One thing that was clear from the beginning is that I wanted to write this thesis under the supervision of the person who has helped me throughout my study and has been a steady feature in my life for the last decade. Without the help and guidance of Alla Peeters-Podgaevskaja, finishing my education at the UvA and finishing my theses, I stress the plural form here, since this is not the first thesis Alla has supervised me on, would have been something for the future. Her knowledge of bilingual language acquisition as well as her almost unstoppable drive to help me finish this thesis have been of major influence and have given me the strength in striving to produce the best thesis I could have. I would also like to thank Professor dr. Wim Honselaar for being the second reader.

These acknowledgements would not be complete without mentioning another person who has helped me during this time. When Bibi Janssen asked me to assist her on testing multiple children for her PhD project, I did not hesitate for a moment. Throughout my path in writing and finishing this thesis, Bibi has been of immense help. She has given me feedback when it was clearly needed, and due to her great knowledge of the subject, she has been a sparring partner par excellence. She has always encouraged me to finish this thesis as soon as I could and has helped me to understand and to use the statistics needed for my research. Without Bibi, testing these children would have been much harder and less fun.

I would also like to thank the boards and teachers of the Russian Saturday Schools in Amersfoort (Mozaika) and Eindhoven (Buratino) for their help in facilitating test sessions with their bilingual pupils. Last but certainly not least, I would like to thank my parents, Hans and Marja, for their belief in me during all these years, for supporting me in finishing whatever study I had liked to pursue, and for giving me the time and finances to do that.

I would also like to thank my colleagues at Credit Agricole for always giving me the time to visit my supervisor and supporting me in writing this thesis. And finally, my friends who have stuck with me while I had to study and was not always able to join them in doing fun things.

Nadine Hut 21 July 2015, Purmerend

(4)

1 Introduction

One thing that was already clear when I planned writing this thesis was that the main focus would be on bilingual Russian-Dutch children. Due to their different languages, which differ in so many ways, they are a great subject to see how grammatical features are acquired when two languages are not similar. Bilingual children have always interested me, so much that I studied them in my BA thesis as well. However, in comparison to my earlier research, I now wanted to investigate younger children, since they are still fully developing their language proficiency.

From the literature we know that monolingual children in the ages of 4;0 to 6;0 years have fully acquired the basic grammatical features and show adult-like mastery of their mother tongue. We also do know that bilingual children lag behind their monolingual peers when it comes to the acquisition of their languages separately. This fact in itself does not surprise me, but makes it even more interesting to research this specific age group.

In this thesis I want to investigate how bilingual Russian-Dutch children comprehend and repeat questions and Relative Clauses (both subject and object) in their two languages. Wh-questions and Relative Clauses (RCs) are used in both Russian and Dutch. However, since Russian is a Slavonic language and Dutch is a Germanic language, the way these structures are represented in both languages is quite different. It is interesting to find out how bilingual children perceive these features and produce them. Which construction is being learned first? Which language is dominant in producing these constructions? Is the age factor in play? Is it easier for bilingual children to comprehend and repeat Subject RCs or Object RCs? These questions came to my mind when thinking about this subject, and they formed the basis of this thesis.

The task which was used in my study as test material was a Sentence Repetition Task (SRT): the children were given short sentences which they then had to repeat. Their responses were recorded and studied to find out whether or not, and to what extent these constructions were acquired at the age between 4;0 and 6;0 years.

The thesis will start off by giving a theoretical overview of bilingualism and bilingual language acquisition; then a detailed description of Wh-questions and RCs in Russian and Dutch will be given, accompanied by some comments on child acquisition of these features. In chapters three and four, a description of the experiment including hypotheses, methodology, information on participants, and results will be given. In chapter five, a short discussion and the conclusions will be presented.

(5)

2 Theoretical framework

This chapter commences with a discussion on bilingualism with regard to key terms and issues connected to the topic that are significant for the ongoing debate on bilingualism on the one hand, and Second Language Acquisition (SLA), on the other. In section 2.3 of this chapter, the focus is on Relative Clauses (RCs), and more specifically on the way children acquire this grammatical feature and the differences in forming RCs in Russian and Dutch. A comparable analysis is conducted in section 2.4, where Wh-questions are discussed, as well as their differences in Russian and Dutch, and, even more significantly the way children acquire this feature.

2.1 Bilingualism from different perspectives

2.1.1 Terms and definitions

There is much debate in the field about what bilingualism truly is. Some scholars state that any individual, who has learnt two or more languages, regardless of at what age the acquisition of the other language took place, can be perceived as a bilingual (Bhatia 2004: 5). Whether one has learnt the „whole” language or just a few words like c’est la vie, or is able to ask for the bill in a restaurant in another country is of no importance in that respect (Edwards 2004: 7). This definition would suggest that there is no real difference between Second Language Acquisition (SLA) and bilingualism.

Other scholars claim that one can only be a true bilingual when two or more languages have been acquired during early childhood. Unfortunately, it is not very clear what the true definition of early childhood is. Connected to early childhood is the claim that a critical period exists, a period, which is of great importance for determining whether a child, who is involved in learning two languages, can become a fluent speaker and to what extent. Some argue that the age of three or four years can be perceived as the end of early childhood, since at that age, children are expected to have acquired most syntactic and morphological features of their mother tongue(s).

The claim that children can learn another language swiftly and fluently before the critical period, next to their mother tongue, is also often connected to Universal Grammar (UG), which states that due to having full access to UG, children can acquire languages quickly and fluently by basically ticking those grammatical features in the brain or cognitive system, which are, according to Lydia White (2007: 42), needed for the other language, since all features are stored in the cognitive centre and just have to be awoken. This also means that a child will learn the language implicitly without being corrected when making errors, which would normally lead to errors in the acquisition. However, in this case, according to Chomsky, this will lead to the complete acquisition of the grammatical rules (White 2007: 42). Other studies have shown that fluency and proficiency

(6)

in both languages can also be achieved before the age of twelve years (Frijn & De Haan 1994: 110). Strengthened by these outcomes, I believe that there is no such thing as a single age boundary. Even though the majority of the features are acquired at the age of three, some major grammatical features will only be acquired when reaching the age of twelve years and beyond.

Bilingual speakers can be divided into successive1 and simultaneous2 bilinguals. A successive bilingual is, for example, someone who is born in an immigrant family where both parents speak the same mother tongue and have moved to another country. It is often said that successive bilingualism can lead to a lower level of linguistic knowledge. This can be assigned to the fact that the younger the child gets acquainted with the two languages, the better the results will be in both languages. Children of immigrants often learn the other language at a later stage in life, which is why it takes them much longer to reach a level of proficiency in comparison to that of children who started learning both languages from birth. Connected to successive bilingualism is the term additive bilingualism, which, according to Appel and Vermeer (2000: 348), means that the first language has enough support at home and a second language has been added to the linguistic palette.

Simultaneous bilinguals have an immediate exposure to both languages, also defined as elite-bilingualism3_{. It has also been mentioned that when a child is a simultaneous bilingual, e.g. the} child learns two mother tongues at the same time, one can not merely speak of language A and B, which means that both languages are equal regardless of being the stronger or the weaker language. This phenomenon is called Bilingual First Language Acquisition (BFLA), which makes the distinction between SLA and bilingualism even more clear (De Houwer 2009: 2).

As I already mentioned in the earlier sections of this chapter, when a child is bilingual, usually one of the languages will be the majority language and the other the minority language. Usually, the majority language will be the dominant language and the minority language will be the weaker one. A child can start with the minority language as a dominant language, due to being a child of immigrants or having a mother who is the speaker of the minority language. When a child goes to school, the majority language can substitute the dominant minority language, because from then on the child will interact in that language most of the time. A reason for the child switching between languages could be that the intensity and the amount of the linguistic input is changing from high to lower, especially in combination with the other, more dominant language.

(7)

2.1.2 Balanced and unbalanced bilinguals

Bilingualism is more complex than most people think. Scholars can easily state that when a child is brought up speaking both languages, the outcome at the age of approximately 18 years old will be that the child has become fluent in both languages. If this were the case, then most of us would be perfect bilinguals by now. Children that start learning two different languages before the age of three can follow multiple paths, leading to differences in their ultimate attainment of the languages. For example, a child that is born in a bilingual environment where the mother and father are speaking language A, where the majority language is language Alpha and the children are schooled in both languages, would be an ideal situation. In this case, the child will be a balanced bilingual, being fluent and equally proficient in both languages. Amongst others, De Houwer (2009: 86-90) has mentioned that there are many different ways of becoming a bilingual family, with balanced and unbalanced bilinguals as the outcome of the amount of input and attitude of both children and parents.

Only a small amount of bilinguals will reach the state where they have the same command of both languages. There are many factors influencing the final outcome. Children can differ in aptitude, linguistic sensitivity, motivation, and attitude to their second mother tongue. The child does not want to learn the other language, merely because it is not the language of the environment. In this case, it will be quite difficult to teach the child to speak the other language without having to force it onto the child (which leads to incomplete acquisition due to the child’s resistance).

More often the child will turn out to be an unbalanced bilingual. Basically, this means that the child has a different command of both languages. This would suggest that almost all bilingual speakers are unbalanced, since it turns out to be quite complicated to learn both languages to the same degree. There are different types of unbalanced bilingualism. A child can be a native speaker of language A and have a good proficiency of language Alpha. Another example could be children who are fluent in language A and have a passive or limited active knowledge of language Alpha. This would mean that the child is an unbalanced bilingual, but the opposite of this kind of unbalanced bilinguals are the children who are not proficient in their languages at all, neither language A, nor in language Alpha. This phenomenon is called subtractive bilingualism, and, according to Appel and Vermeer (2000: 348), this refers to children who learn language Alpha when language A has not been acquired to a certain extent, or who are not supported enough at home or at school. This can in turn lead to double semilingualism, which means that none of the two languages is fully acquired and that the child is equally poor in his/her command of both (Appel and Vermeer 2000: 348).

(8)

2.2 Factors influencing language acquisition

2.2.1 Input

One of the most important factors contributing to linguistic development is input. In this thesis, input is equal to the amount of linguistic material (in speech and in script, as well as directly and indirectly addressed to the child). Since the input is of vital importance in acquiring the languages it is necessary to know how often the children get addressed in both of the languages and who addresses them in these languages. Even though children do not just imitate their caretakers, they do need input as a reference for their own linguistic upbringing. The caretakers need to correct the children if their sentences and constructions are formulated the wrong way and incorrect (De Houwer 2009: 119-127).

A principle that has been broadly introduced is the ‘One person, one language’ principle (1P/1L). According to De Houwer (2009: 107), many parents stick to this principle in a strict fashion. However, as De Houwer claims, these parents do not stick to this principle all the time (Goodz 1989: 35-44). One could say that this principle is more of an ideal situation than a real fact. This principle would result in children becoming balanced bilinguals, since the amount of input from both languages would be 50%-50%. However, De Houwer claims that this strategy does not necessarily result in active bilingual use. Even though the children seem to be equally addressed in both languages, it still can lead to a situation in which children only prefer to speak one language instead of both languages offered. Actually, it turns out that children, who grew up in the families using the 1P/2L4 principle, had a slightly better chance of speaking both languages (De Houwer 2009: 8). This claim is also supported by the research of Yamamoto (2001), which showed similar results. However, these results do not show why the children with 1P/1L are more likely to speak one language than their 1P/2L peers.

2.2.2 Interaction

Besides the importance of input, one also has to reckon that it is of great importance whether it is the mother or the father who speaks the majority language or the minority language. Research has shown that the mother’s speech is always different from that of the father. Panscofar and Vernon-Feagans (2006) have shown that fathers are of great importance to the language development of their children. Research has also pointed out that parents who address their children in a more academic style have a great influence on the language development of their children, basically because they introduce their children into a world of more complex linguistic structures. The same

(9)

holds for the fathers, they seem to have a different way of addressing their children which will help them develop different features in their languages (De Houwer 2009: 113-114).

Even though the father can be the caretaker at home or the stay-at-home-dad, the way a father addresses his child will always differ from the way the mother does. In that sense, many scholars (amongst others, De Houwer (2009: 113), claim that one can speak of motherese, a specific way of addressing a child. There is no evidence that the mothers’ talk contributes to a faster linguistic development of their child. Of greater importance is the amount of time spent in interaction. It is known that mothers spend more time with their children because they often work less or take care of their children full-time. Another feature, which is not supported by research, is that the mother’s language will always be the stronger language. There is no research that can back this up. In contrast, it is more often the environmental language that turns out to be the stronger language. This is, according to De Houwer (2009: 116), the result of the combined effect of peer pressure and the higher frequency of input of the societal language.

2.2.3 Corrective strategies/Negative feedback

Just like there are different strategies in teaching children both languages, there seem to be different approaches to correcting children. Feedback turns out to be of great value for both monolingual and bilingual children. According to Ortega (2009) and De Houwer (134-38), due to ‘negative’ feedback, a child playfully notices that what has been said is not correct and can start imitating the correct structure. De Houwer (2009: 134-138) claims that there are multiple strategies for correcting the utterances produced by children. For example, when a young child in conversation with the mother says: I want a tizza, where the child is basically asking the mother for a pizza; the mother can react in multiple ways. For example, she could say: It is not tizza, you must mean pizza! However, it could result in the child not knowing what the mother is saying and he/she will continue to use this word. Another way of correcting the child would be that the mother would repeat the question to provoke the child to respond correctly, this is better known as a recast. Ortega (2009) showed that this type of feedback/correction is popular among teachers/parents since it gives an example of the correct utterance but it does not make the child feel bad, since he/she will rarely experience this to be a correction. As exemplified in You want a pizza? the child can then respond with: Yes, I want a pizza. Children seem to acquire words and structures according to a U-shape strategy or the Skill Acquisition Theory. According to De Keyzer (2007: 97-114), this means that the child will imitate the parent and use the correct word, after a while he/she will start playing with the word or structure and will formulate his/her own versions, which are incorrect. This is the period when the child needs to receive negative feedback in order to get back to the right track and start using and sticking to the correct form.

(10)

Another feature of correction/negative feedback can arise when the child starts interacting in both languages. It can often lead to code switching or code mixing, when the child uses the words and structures from language Alpha when speaking language A. For example, the child could say: Mum, can I have some snoepjes? where snoepjes is the Dutch word for sweets. The mother can correct the child by saying: It’s not snoepjes but sweets, or she could say: I don’t know what you’re talking about, could you please tell me again what you want? The child clearly shows that the translation for sweets is not on the tip of the tongue and uses his/her other language to fill in the gap, which is not that bad, because due to this strategy, the child maintains communication, which is important. Another way of tackling the problem could be to reformulate the child’s question correctly (the so-called recasts): for example, the mother could say: Would you like some sweets? In this way, the child will be confronted with the correct word or utterance without feeling that what he/she had said was incorrect. Note that children can stop using the language, when they receive too much negative feedback, which can result in the lack of interest and motivation, fear to fail and negative emotions with regard to their 1L2.

2.2.4 Incomplete acquisition

Polinsky (2008b: 40) states that incomplete competence in the first language seems to be an interesting matter, that would indicate interrupted acquisition, delayed acquisition, and even language loss. Even though we are not yet able to conclude whether the children studied in this thesis are subject to incomplete acquisition, it does provide us with the background information, which is needed to clarify the results.

In order to be able to understand acquisition, one has to understand the nature of incomplete acquisition, mainly because the heritage speakers or low proficient bilinguals seem to provide a crucial link between competent L1 learners, balanced bilinguals, and presumably L2 learners (Polinsky; 2008b: 40). According to Polinsky (2008b: 40), an incomplete learner or heritage speaker of language A has learned or has only heard language A in the country of birth, but has moved to another country at a young age, more than often, before this speaker has been attending any kind of educational institution. It has to be noted that the group of heritage speakers as such cannot be portrayed as a homogeneous group, since they vary from passive speakers to near fluent speakers of the language. The baseline language of heritage speakers is not comparable to that of monolingual speakers, since the command of heritage speakers’ language is based on the input they were exposed to as a child (Polinsky 2008b: 40).

There is a difference between incomplete acquisition and language attrition. Incomplete acquisition is the lack of mastering a language, whereas language attrition is the loss of language

(11)

less grammatical sentences as a result (Polinsky 2007: 176). Language attrition is not applicable to the group of our participants because they have not achieved a complete mastery of their 2L1 yet. However, both phenomena do share some features that are noteworthy. One of the features is that both result from insufficient access to a given language input, according to Polinsky (2007: 176), impoverished and heterogeneous input.

In what follows, a brief overview of incomplete acquisition of nominal morphology in Russian and Dutch will be presented. Because the focus of this thesis is on the analysis of the use of Wh-words and relative and interrogative pronouns by young Dutch-Russian bilinguals, verbal morphology stays beyond the scope of our survey.

2.2.4.1 Russian

It’s well known that gender and case marking is problematic in 2L1 of bilingual children with low language proficiency (amongst others Polinsky 2008b: 46). They make a distinction between masculine (words ending in a consonant) and feminine words (ending in {a}) of the first and second declensional classes, since their endings are clear to the children. However, feminine nouns ending in palatalized consonant and neuter stem-stressed nouns are difficult to acquire. According to Polinsky (2008b: 46), feminine nouns ending in a soft consonant are often treated as masculine ones. Neuter stem-stressed words are experienced as feminine, which is not surprising, since stressless endings in {o} are reduced to a kind of [a-] sound, close to a schwa, due to vowel reduction of unstressed /o/, /a/, and /e/, which is one of the important features of Russian phonology. Note that these features are not something specific only for bilinguals. In the literature on child acquisition of Russian, the same mistakes are also observed in monolingual children of a younger age (Cejtlin, 2009: 151-152).

When it comes to the use of the Russian case system, as Peeters-Podgaevskaja states (2008: 614), Russian-Dutch unbalanced bilinguals at age 5-7 are still not able to produce simple utterances with correct case markers. The children use the regular nominative, accusative, and locative cases singular, and certain forms of the genitive. However, the second locative is not used; and the stress pattern in the accusative singular of feminine nouns with shifted stress is regularised (for example, v vodu instead of v vodu). The dative and instrumental are practically missing in children’s speech and are replaced by the default nominative or genitive cases. The plural case system consists of three different endings: nominative, genitive, and some mixed forms of the dative-instrumental-locative cases (Peeters-Podgaevskaja 2008: 616). The situation is even worse with low proficient heritage speakers of American-Russian, who usually have only two case forms, one being unmarked (corresponding to the nominative), and the other based on the accusative case (Polinsky 2006: 214). According to her (2008b: 45), the knowledge of the case system is based on the

(12)

knowledge of declensional classes, which is absent in heritage speakers and unbalanced bilinguals (for instance in American-Russian bilinguals).

When we look at the acquisition of number and gender agreement in bilingual Russian children, we can see that there seem to be non-significant errors. However, there seem to be some issues in acquiring features that are not embedded in the structures formed by analogy. For example, adjectival agreement with immutable words or nouns containing mixed declination is problematic: želtyj taksi 'yellow-m.nom taxi-n.nom', and/or tvoja imja 'your-f.nom name-n.nom' (Peeters-Podgaevskaja 2008: 615).

According to Peeters-Podgaevskaja (2008: 623), there seems to be an interesting use of the pronoun kakoj, where it appears to take over the functions of the relative pronoun kotoryj in Relative Clauses, as exemplified:

1) Ja vstretil mal’čik, kakoj tri goda. I met boy-m.sg.nom which-inter.m.sg.nom three years ‘I have met a boy who is three years old.’

2) Malen'kie rakety, kakoj nado v pistolet vstavljat'. Little rockets-f.pl.nom which-inter.m.sg.nom need in pistol inserted. 'Little rockets, that need to be inserted into the pistol.'

Note, in monolingual Russian children, the acquisition of case forms starts at the age of two. However, the mastery of the complete case system takes much longer, and ends around the age of 6-7 (Slobin 1965: 96).

2.2.4.2 Dutch

When we look at the incomplete acquisition of Dutch, we can see that there has not been much research conducted on this matter. The only ‘bilingual’ research that has been reported is on the use of Dutch by Arabic/Turkish-Dutch bilinguals in The Netherlands. This is not fully comparable to Dutch as a 2L1 somewhere.

Some research has been conducted on the language use of Dutch in three generations of immigrants in New Zealand. However, this group can hardly be compared to the Russian diaspora in the United States of America because their population is much smaller than the Russian diaspora. Since Dutch is closely related to English, the Dutch immigrants were much more likely to start using English, instead of preserving their spoken Dutch, was due to their cultural identity. Dutch as

(13)

in their L2, English, but still speak some Dutch. The third generation can be perceived as monolingual in English (Hulsen 2000: 14).

Like their Russian counterparts, Dutch immigrants also differ from their original countrymen in the way they have retained the language and have been using it until now. The test group used by Hulsen in her dissertation, greatly differed from the Russian-English research in the sense that the subjects where tested at a much later stage in their life, and that they had emigrated to Australia and New Zealand at a later age than the subjects in the Russian-English research. It does, however, give us insight in how Dutch is maintained over the years in a bilingual environment.

According to Keijzer (2009: 2), the maintenance of Dutch is strongly connected to the level of education in connection to accuracy rates; this also counts for the Dutch control groups. A research conducted on Canadian émigrés from the Netherlands showed that there was no significant difference in the answers between the test group and control group. Errors were mostly found in the regularization of irregular forms (for example, *ik helpte mijn moeder, instead of ik hielp mijn moeder ‘I helped my mother’); the misuse of Dutch definite articles de and het, and the misuse of diminutive forms of standard words (for example, *raam-tje instead of raam-pje ‘little window’), which are comparable to the data by Dutch monolingual children. Other research, amongst others Montrul (2008), has shown that the percentage of ‘marked’ elements in Dutch is lower than 5% when it comes to language interference from English. Most problems occur at the lexical and morpho-syntactical level, for example, the overgeneralised use of the article de ‘the’, which can be seen as a direct influence from the English article ‘the’.

Interestingly, Keijzer’s other research on attrition in Dutch also showed that the acquisition of Dutch is frequently not complete until the age of 18 (Keijzer 2007: 74).

2.3 Relative clause agreement in Russian and Dutch

2.3.1 Russian

According to Praktische Grammatica van de Russische taal (Podgaevskaja & Honselaar 2007: 503), a Relative Clause (RC) is used to give additional information on a constituent from the main clause. By combining sentences with a RC, the text becomes more fluent and the repetition of certain words can be avoided. Depending on the relationship between a RC and a word in the main clause, and the meaning and functions of the constituents within a RC, two types of RCs can be distinguished: a Subject RC (SRC) and an Object RC (ORC). In SRCs, a subject is expressed by a relative pronoun and is mostly directly followed by a verb (for instance, in English). In ORCs, a relative pronoun refers to another constituent and has no subject function.

(14)

According to Polinsky (2008a: 338), Russian relative clauses are formed by using the gap strategy5, and involve a relative pronoun (kotor-), which agrees with the extracted constituent in gender and number. The case marking depends on the function of kotoryj in RCs. Next to kotoryj, kakoj, čto, and kto can be used, although less frequently. Kotoryj, kakoj, and čto are not sensitive for animacy and refer to animate and inanimate constituents as well. Kto can only be used with animate referents.

Kotoryj

The pronoun kotoryj acts as an independent pronoun within the dependent clause. According to the Praktische Grammatica van de Russische taal (Podgaevskaja & Honselaar 2007: 176), kotoryj substitutes a noun and has case marking that is required in the dependent clause. For example6:

3) Čelovek, kotoryj prodal mne etu kvartiru, uexal v Moskvu. man-m.sg.nom that-rel.m.sg.nom sold me this apartment went to Moscow 'The man who sold me this apartment, went to Moscow.'

4) Čelovek, o kotorom my tol'ko čto govorili, uexal v Moskvu. man-m.sg.nom about whom-rel.m.sg.loc we just talked went to Moscow 'The man, who we just talked about, went to Moscow.'

As opposed to case, gender and number are decisive for kotoryj, and are determined in the main clause. This is shown below:

5) Ženščina, kotoroj ja prodal etu kvartiru, uexala v Moskvu. woman-f.sg.nom to whom-rel.f.sg.dat I sold this apartment went to Moscow 'The woman whom I sold this apartment to, went to Moscow.’

6) Ozero, o kotorom my tol'ko čto govorili, ser'ezno zagrjazneno. the lake-n.sg.nom about which-rel.n.sg.loc we just spoke seriously polluted 'The lake that we just spoke about, is seriously polluted.'

A relative clause involving kotoryj is always preceded by a comma, regardless of whether the sentence is a restrictive or defining RC.

(15)

Kakoj

The RC with kakoj refers to the qualities and/or features of objects that are mentioned in the main clause (Podgaevskaja & Honselaar 2007: 503). Kakoj shares the same grammatical features as the abovementioned kotoryj. The case is determined by the structure of the RC and the function of kakoj. The gender and the number correspond with those of the antecedent in the main clause. For example7:

7) Ego prigovorili k samoj strašnoj kazni, kakaja tol'ko možet he was convicted to most hideous death.penalty-f.sg.dat that-rel.f.sg.nom just can 'He was convicted to the most hideous death penalty that

byt'. be exists.'

8) On vsë smotrel na tu dver', v kakuju ona ušla s vedrom. he all looked at the door-f.sg.acc through which-rel.f.sg.acc she went with the bucket 'He looked the whole time at the door, through which she went with the bucket.'

There is one exception between kakoj and kotoryj, being that kakoj can be used as a plural in contexts, which emphasize general usage or a common tendency, whilst the word in the main clause it is referring to is singular, for example:

9) Svet padal v komnatu čerez okno, kakix ja nikogda ešče ne videl.

light-m.sg.nom fell in room through window that-rel.m.pl.acc I never even not saw

'The light fell through the window, like I had never seen before.' Kto

According to Podgaevskaja & Honselaar (2007: 503), the relative pronoun kto is connected to animate referents that are further defined in the main clause with a pronoun or an adjective. For example8_:

7_{The Russian examples 7 to 9 are extracted from Praktische Grammatica van de Russische taal (Podgaevskaja &}

Honselaar 2007)

8

The Russian examples 10 to 12 are extracted from Praktische Grammatica van de Russische taal (Podgaevskaja & Honselaar 2007)

(16)

10) Vinovny te, kto oxranjal, i te, kogo guilty those-pl.nom who-rel.m.sg.nom guarded and those-pl.nom who-rel.m.sg.acc 'Both the guarded and those who was guarding are guilty.'

oxranjali. guarded

When kto is the subject of the RC, the verb is always associated with masculine (in the past tense) and singular (in the present/future tense). Kto is usually accompanied by the deictic pronoun tot in the main clause. This means that there is a specific connection between kto and the animate referent that is referred to by tot.

Čto

A RC with čto can be seen as the counterpart of kto: it refers to objects and is connected with words such as to, vsego, etc. in the main clause. For example:

11) Larisa osuščestvila to, o čëm mečtala polžizni. Larisa made that-n.sg.acc about what-rel.n.sg.loc dreamed half life 'Larissa realised that, what she had dreamed of half of her life.'

In informal language use, čto can be used instead of kotoryj. It is however remarkable that when čto is perceived as the topic of the relative sentence, the predicate agrees in number and gender with the noun to which čto refers and not just with čto itself. For example:

12) On napravilsja k dveri, čto naxodilas' daleko vperedi. he went to door-f.sg.dat that-rel.n.sg.acc located far ahead 'He moved towards the door, that was located far ahead of him.'

(17)

It is also remarkable that čto as a relative can refer to people, as can be seen in the next example: 13) Te, čto sprava stojat, stojat, no te kto

those-m.pl.nom that-rel.n.sg.nom right stand stand but those-m.pl.nom who-rel.n.sg.nom 'Those who are on the right -stand still, but those on the move must

idut, vsegda dolžnyj deržat'sja levoj storony! go always must hold on left side always keep on the left!'9

In this case, čto replaces kotoryj; however it should be noted that this use is limited to the informal style.

In Russian, relative pronouns refer to the last mentioned referent, an antecedent, and follow the long-distance dependency principle which consist of two important characteristics, first, there is a difference in articulation in the expressions filling the head and tail points of the dependency, and second, the positions are separated by a number of unregulated segments (Polinsky 2008: 2). Sentences in which the relative pronoun refers to some component at the left edge of the main clause are ambiguous and mostly judged as incorrect:

14) Devočka xodit v školu, kotoraja xorošo učitsja. girl goes to school-f.sg.acc that-rel.f.sg.nom good teaches. ‘The girl goes to school, that teaches well.’10

In example (14), the information comes from the first referent of the sentence, being devočka, one would usually expect kotoraja to connect with škola due to the close distance dependency principle, which means that the last mentioned referent is the constituent to which the relative pronoun refers. However, in this context, the verb in the RC is also connected to the first referent in the main clause. This is grammatically correct, but semantically is perceived as an error (Polinsky 2008a). Despite the fact that Russian is a free word order language, the relative pronoun occupies an obligatory fronting position in RCs. Other constituents can occupy any position within RCs, although a different word order is preferred. However, it is still not clear whether the concept of “default” word order is appropriate for Russian RCs.

9_{Okudžava, Bulat – Pesenka o metro (1957-1961).} 10_{Russian example 14 is not taken from other research.}

(18)

2.3.2 Processing of RCs

The acquisition of RCs in Russian occurs fairly early, typically in the beginning of the third year of life (Gvozdev 1961). It even turns out that SRCs appear earlier, are produced on a larger scale, and children make fewer errors when comprehending. It also seems that by the age of 4;0 the amount of errors in the choice of the head of a relative clause becomes marginal. According to earlier research (Polinsky 2008a: 348), young six-years-old speakers seem to have full adult-like mastery of RCs. However, the developmental path for Relative Clauses in Russian is not fully investigated yet.

In order to be able to see how both monolingual and bilingual speaker’s process, comprehend and produce RCs, Polinsky set up an experiment. The results of the experiment are relevant for my thesis, which is why I have chosen to discuss these results in the following section (Polinsky 2008a).

The goal of the experiment was to determine possible differences in the comprehension of SRCs and ORCs in monolingual baseline speakers and heritage speakers. Besides, the test was also used to determine the frequency effect in all speakers, and the influence of English on the heritage speakers. Polinsky also hypothesised that if heritage speakers, who have moved to the USA before the age of four, would be able to produce and comprehend the relatives like their monolingual peers, that would suggest that relatives are acquired quite early and might be part of UG. It would be a totally different case when the children seemed to have a command of the relatives and adult heritage speakers did not. This would mean that there is a true loss of their grammar, which has been acquired in childhood, and this fact might be ascribed to attrition.

The test groups consisted of ten monolingual children with a mean age of 6;6 and nine heritage speakers with a mean age of 7;5. The other two groups consisted of seven monolingual adults with a mean age of 28;7 and 12 adult heritage speakers. In the test, the participants were given 36 pairs of pictures portraying reversible actions. For each pair, the subject heard a relative clause within a question and had to choose the picture matching the description.

15) Gde koška [kotoraja sobaku dogonjaet]? where cat that-rel.f.sg.nom dog-f.sg.acc is.catching.up ‘Where is the cat that is chasing the dog?’

16) Gde koška [kotoraja dogonjaet sobaku ]? where cat that-rel.f.sg.nom is.catching.up dog.f.sg.acc ‘Where is the cat that is chasing the dog?’

(19)

Object relatives, with both orders are illustrated below:

17) Gde sobaka [kotoruju dogonjaet koška]? where dog-f.sg.nom that-rel.f.sg.acc is.catching.up cat-f.sg.nom ‘Where is the dog that the cat is chasing?’

18) Gde sobaka [kotoruju koška dogonjaet]? where dog-f.sg.nom that-rel.f.sg.acc cat-f.sg.nom is.catching.up ‘Where is the dog that the cat is chasing?’11

The results of the test showed that the monolingual children and adults had no trouble connecting the correct answer to the pictures. Young heritage speakers also showed a high level of accuracy, and their responses did not deviate compared to the monolingual groups. According to Polinsky (2008: 353), this also means that the young heritage speakers seem to show no traces of attrition and transfer. The biggest surprise was in the adult heritage group. These speakers seemed to have almost no trouble with comprehending SRCs but scored less on the ORCs. Polinsky explains these outcomes: the two main questions were whether the knowledge of RCs was still available in adult heritage speakers, and whether English seemed to be influencing the L2 when producing the sentences. Surprisingly, it turned out that the adult heritage speakers were still quite accurate when producing SRCs, but scored at chance in ORCs, this was not influenced by word order. It also turned out that the loss of use of RCs was not due to influence of English, but rather as a result of attrition. As mentioned earlier, if language interference had been on the foreground, the adults should have shown a different perception of SRCs and ORCs (Polinsky 2008: 351). The experiment of Polinsky focused on processing of both types of RCs. She mentioned nothing about production of these sentences by bilingual speakers of Russian.

2.3.3 Dutch

According to Van Kampen (2007: 112), the Dutch relative pronouns can be picked from the set of d-pronouns (die, dat 'that') and sometimes from the set of w-pronouns (wie 'who', wat 'what', waar 'where'). When looking at daily Dutch, one can imagine that there are multiple options and possibilities for choosing between the d-pronouns and w-pronouns.

In the sentence Het meisje (a. dat b. wat c. die) ik heb gezien ‘The girl that I have seen’12, the only correct answer in standard Dutch is a. dat. The options b. wat and c. die are more informal

11_{The Russian examples 15 to 18 are extracted from Polinsky (2008a).} 12_{The Dutch example is extracted from Van Kampen (2007: 112).}

(20)

(Algemene Nederlandse Spraakkunst (ANS) 1997: par. 5.8.3.2, 5.8.5.513). In contemporary Dutch, the usage of wat over the correct use of dat seems to increase as the preferred usage. According to Bennis (2001), the more informal relative w-pronouns will in the long run obstruct the relative d-pronouns. The ANS states the following when it comes to relative pronoun selection: “If it is possible to express gender agreement between the antecedent and the relative, select the d-pronoun. Select the w-pronoun if such gender agreement cannot be expressed.”

Language acquisition is a gradual process where the earlier steps of acquisition are used as a basis for the later steps (Van Kampen 2007: 112). For Dutch, this means that the acquisition of relatives will take place when the earlier acquisition of d-pronouns and question w-pronouns in root sentences has been acquired. However, this system is not yet in place before the acquisition of the V2nd rule (Van Kampen 2007: 112) 14_.

According to Van Kampen (2007: 113), Dutch distinguishes six main pronouns in root sentences, meaning three from the w-set and three from the d-set. This is summed up in table 1:

d-set <±neuter> referent w-set <±animate> referent Strucural die <-neuter> wat <-animate>

dat <+neuter> wie <+animate>

Oblique [daar] ...(op) [waar] ...(op)

Table 1. Relative pronouns in Dutch, extracted from Van Kampen (2007)

According to Van Kampen (2007: 114), as is explained in table 1, the d-set is connected to the grammatical <±neuter> and in turn, the w-set is connected to the semantic <±animate>. The d-pronouns are attached to a discourse antecedent; they consist of an identified referentiality and are able to express the grammatical gender of the antecedent in the Determiner Phrase (DP). In contrast to topic d-pronouns and relative pronouns, interrogative w-pronouns maintain a reference that is still to be identified (Van Kampen, 2007: 114). The oblique form of the d-set, daar, is not sensitive for gender distinction. This also concerns the oblique form of the w-set, waar, which is not sensitive towards this distinction as well.

13_{http://ans.ruhosting.nl/e-ans/14/05/03/08/03/body.html (visited 08-06-2014, 20:40)}

(21)

The relative pronouns that are <+animate> allow the variant [preposition + wie] next to [waar + preposition] with regard to [+inanimate] (Van Kampen 2007: 114). For example15:

19) Daar staat {Jan, iemand, een meisje} te kijken met wie we moeten spreken. there stands {Jan, someone, a girl} looking with who we must talk

‘There {Jan, someone, a girl} stands and looks around whom we need to talk to.’ 20) Daar ligt een boek, waarover ik jou vertelde.

there lies a book, about.which I you told ‘There is a book, that I told you about.’

According to Van Kampen (2007: 120), because the d-pronouns {die, dat} lack the natural <+animate>, oblique topic pronouns have to be expressed by the pronominal adverb or by the (stressed) personal pronoun. These forms cannot express grammatical gender in Dutch, which is why the oblique relatives switch to the w-system. Monolingual children process the oblique case forms when they are almost four years old. In the research of Van Kampen (2007: 120), the child Sarah first used the words at the age of 3;10. The child, however, only used the w-set, instead of the d-set, which was also available in her input. This will later be explained in the section concerning the processing of relative pronouns in Dutch.

When we look at the rules in the ANS, we can see that the relation between the d-set and w-set is blocked. According to Van Kampen (2007: 114), the d-w-set blocks the w-w-set by prevailing the grammatical gender over the more universal distinctions, for example semantic animacy, which means that the d-set is used where the w-set would be triggered because gender is of a higher order. Two examples from the d-set in standard Dutch are:

21) Het boek <+neuter> dat <+neuter> ik leuk vind. the book-n.sg.nom that-rel.n.sg.nom I nice find ‘The book that I like.’

22) De man <-neuter> die <-neuter> ik leuk vind. the man-m.sg.nom that-rel.m.sg.nom I nice find ‘The man that I like.’

(22)

When there is no separate antecedent left the w-relative is selected: 23) Wat overblijft is niet noemenswaardig.

what-rel.sg.nom remains is not appreciable ‘(That) What remains, is not appreciable.’

In sentences like (24) the antecedent and relative are ‘fused’ by the usage of a w-set for an <+animate> referent, this is because there is no obvious difference in antecedents:

24) Wie zoet is krijgt lekkers16. who-rel.sg.nom sweet is gets sweets ‘(he) Who is sweet, gets sweets.’

The selection of a d-relative is not just based on gender agreement. Besides gender agreement, the d-set is also heavily influenced by semantic animacy (Van Kampen 2007: 115). This is shown in table 2:

a. type w-set * Jan wie ‘John who’ * hij wie ‘he who’ * iedereen wie ‘everybody who’ * iemand wie ‘somebody who’ b. type d-set Jan die

‘John that’ hij die ‘he that’ iedereen die ‘everybody that’ iemand die ‘somebody that’

Table 2. Relative pronoun selection of the d-set, taken from Van Kampen (2007: 115).

Van Kampen (2007: 115) states, “that <+neuter> equals ‘unspecified for gender’. When

<+animate> and <+gender> are sole features, the topic pronoun die could be selected by antecedents that grammatically react to gender and/or animacy, whereas dat is selected by antecedents that are grammatically unspecified for gender”.

Van Kampen (2007: 119) states that the root pronouns are acquired before the relative pronouns. This has to do with relative clauses being a dependent clause; main sentences are acquired earlier than dependent clauses. Van Kampen’s research shows that before the age of three, children are able to use root topic d-pronouns and root question w-pronouns. Die is used as a topic pronoun for all <+animate> antecedents. Research produced 34 cases of overt non-cliticized root w-pronouns (wat, waar, wie) recorded between the age of 2;6 to 3;0. The child also produced 33 root d-pronouns (mainly die en dat) with a clear discourse antecedent.

(23)

According to Van Kampen (2007: 120), RCs seem not to appear up to the age of three, when the child has fully acquired the pronouns for root questions and root topics. By that time, gender information is well acquired, the default wat takes a solid position in the relative system, and after that will remain as a first option. This is also shown by informal use of Dutch relatives by adults. The best learnable features of relative agreement in Dutch turn out to be those where the antecedent is marked for gender and animacy. Relative pronouns where the d-antecedent is ‘visible’, never seem to be the topic of switching to the w-system, they all result in die. This turns out to be the strongest part of the relative paradigm. When gender is unspecified, the usage of w-pronouns occurs. This would explain why children would start with using d-pronouns, especially when this is also used by the parents or caretakers, but will move to w-pronouns when they start to reason for themselves. Besides, it also turns out that the d-pronouns are acquired fairly late in comparison to the w-pronouns, which seem to be more default (Van Kampen 2007: 123).

2.4 Wh-questions Russian vs. Dutch

In this section, I will address the properties of Wh-questions. In order to do so properly, I have developed a scheme which is applicable to both Russian and Dutch, and which sheds light on the similarities in both languages and, of course, the differences between the two languages.

In table 3 an overview of the interrogatives in Russian and Dutch is given.

Russian Dutch Translation Specification

kto čto kakoj kogda kak počemu, začem gde kuda wie wat welk(e) wanneer hoe waarom, waarvoor waar waarheen who what which when how

why, what for where where to human non-human feature time manner reason place place + direction

(24)

The words in table 3 can have different functions in the sentences, which can be seen in table 4:

Function Russian Dutch English

Subject kto čto wie wat who what

Object čto (Acc.)

kogo (Acc.) komu (Dat.) kem (Instr.) o kom (Loc.) wat wie (aan) wie (met) wie (over) wie what whom (to )whom (with) whom (about) whom Other functions kakoj

kogda kak počemu začem gde kuda welk(e) wanneer hoe waarom waarvoor waar waarheen which when how why what for where (place) where (direction)

Table 4. Functions, based on Rojina (2005: 2)

In languages such as German, English and Dutch, there is no free word order, which is characteristic for Russian. In the formation of Wh-questions, the fronting of Wh-elements is involved which is more than often compulsory. For example17:

25) Waarom ging je daarheen? why-inter went you there ‘Why did you go there?’ 26) *Ging je daarheen waarom?

went you there why *‘Went you there why?’

Sentence (26) could be considered as faultive because the Wh-element is not at the front of the sentence, but instead at the at the end of the sentence, which is perceived as incorrect. In Dutch, the

(25)

only possible way of Wh-question formation is to put the Wh-element in the beginning of the sentence. In Russian this seems not to be the case. This will be shown by the distinction between multiple types of Wh-questions.

In the following sections we will concentrate on object and subject Wh-questions because they are relevant for our study. Other types of Wh-questions are beyond the scope of this thesis. There is no specific research conducted on interrogative adjectival pronouns in combination with nouns in any language involved in this study. This is the reason why the interrogative pronouns, such as who and what, are discussed. However, the observations are applicable to NPs with interrogative adjectival pronouns as well.

2.4.1 Subject Wh-questions

In subject Wh-questions, the interrogative pronouns who or what fulfil the function of a subject and are put in a fronted position in a sentence. For Russian, SRCs are not problematic since Russian does not have auxiliary verbs, which are available in Dutch and English (Rojina 2005: 71). For the languages involved, subject-verb (SV) is the default word order.

27) Kto prišel /kakaja devuška prišla na večerinku?18 who-inter.m.sg.nom came /which-inter.f.sg.nom girl came on party ‘Who/which girl came to the party?

2.4.2 Object Wh-questions

In these kinds of sentences, the Wh-word is in the object position and usually triggers the accusative case. In Russian, object Wh-words can occupy different positions in the sentence (Rojina 2005: 70). For example:

28) Kogo /kakuju devočku ty videl? who-inter.m.sg.acc/which-inter.f.sg.acc girl you saw ‘Whom/Which girl did you see?’

29) Ty kogo /kakuju devočku videl? you whom-inter.m.sg.acc/which-inter.f.sg.acc girl saw ‘Whom/Which girl did you see?’

18

(26)

30) Ty videl kogo /kakuju devočku? You saw which-inter.m.sg.acc/which-inter.f.sg.acc girl ‘Whom/Which girl did you see?’

In example (27) the structure is quite obvious, the Wh-word moves from within the Verb Phrase (VP) to the beginning of the sentence. In example (28), the Wh-word occupies a lower position than the subject. The subject ty ‘you’ moves higher and now occupies the Topic position; which is now above Complementizer Phrase (CP) (Rojina 2005:70). In example (29) the Wh-word remains unmoved as in an indicative sentence, which is incorrect. In English and Dutch, Wh-in-situ19 is only used to create echo-questions, which, according to Rojina (2005: 71), are used for the expression of surprise or amazement, or to ask a speaker to repeat a particular word in the sentence. This is not the case in Russian, for example:

31) *Ty razgovarival s kem?

you talked with who-inter.sg.m.instr? ‘With whom did you talk?’

2.4.3 Multiple Wh-questions

The forms mentioned in the previous sections, are examples of single questions (only one element present in the sentence). Russian also allows multiple questions (two or more Wh-elements are present in the sentence), which is the same for Dutch, but it will only be perceived as grammatical when one of the Wh-elements remains in-situ (Adger 2003).

32) Wie toonde wat? who-inter showed what-inter ‘Who showed what?

33) *Wie wat toonde? *who-inter what-inter showed *‘Who what showed?’

(27)

Since Russian has free word order, both Wh-words can move and occupy different positions in the sentence (absorption structure) and there the meaning of the sentence remains the same. Due to the Wh-restriction in Dutch, moving the Wh-elements to different positions causes a change in the meaning (Rojina 2005: 72). When looking at the following sentences in Russian and Dutch:

34) Kto čto našel? who-inter.nom what-inter.acc found? ‘Who found what?’

35) Čto kto našel? what-inter.acc who-inter.nom found ‘Who found what?’

36) Wie vond wat?

who-inter.nom found what-inter.acc ‘Who found what?’

37) Wat vond wie? what-inter found who-inter ‘What found who?’

One can see that, in Russian, no change in the meaning of the sentence is observed. On the contrary, in Dutch, both sentences can be perceived as grammatically correct; only the meaning of the sentences is different. In example (35), the who can have found multiple whats. In the second example (36), the what can be found by multiple whos. However, the Wh-element remains in-situ.

Accoring to Rojina (2005: 73), there also seems to be a Superiority Effect. If both elements (both subject and object) are questioned, the closer the constituent to the left is, the more of an upper position it takes in the syntactic structure (Rudin 1986). This is present in various languages, for example, in English (Pesetsky 2002: 22). In this situation, “the closest (the leftmost) Wh-element will move and the second one „must have „crammed in” underneath the first phrase” (Pesetsky 2000: 22). Russian is not sensitive to this effect due to its free word order; the Wh-word can occupy any position. Due to the word order restrictions in Dutch, Dutch cannot move words around as freely as the Russian language can and is as sensitive to this effect as English (Rojina 2005: 73). This has already been demonstrated in examples (36) and (37).

(28)

2.4.4 The acquisition of Wh-questions by children

According to Rojina (2005: 76), monolingual children first start using Wh-questions at the age of 18;5 months. In her test, the one-word sentences were removed, since this seemed to be no basis for true Wh-questions. This gradually increases over the months until it eventually peaks at 22;5 months. From now on the child has acquired the basic elements needed to construct Wh-questions and is now practicing to formulate Wh-questions. After this point, the use declines.

Research shows that the children start fronting the Wh-element in the sentence from the moment they are speaking in so called two-word utterances. Different studies also show that Wh-complement questions are acquired first, and then later on, subject Wh-questions are produced (Radford 1995, Klima & Bellugi 1966, Bowerman 1973, Plunkett 1992). The participant in Rojina (2005: 78) shows that a Wh-word as a predicate is used the most, followed by a Wh-object, and finally at the age of 1;8 years a Wh-subject occurs. Since Russian is a language with free word order, the children seem to switch the words in the sentence, which never have a "default" position.

According to Van Kampen (2009: 128), the acquisition of long Wh-questions will commence when the earlier acquisition of Relative pronouns is complete. It must be sufficiently acquired before the connection with Wh-questions appears. According to earlier research, this has to be the w-set, because the d-set is only firmly acquired at the age of 6;0. This means that at that age both the w- and d-set have been fully acquired. Wh-questions can then be seen as an accompanying step in the acquisitional process. Due to their knowledge of the Relative paradigm, Dutch children are already aware of the use of the structures needed to produce Wh-questions, which simplifies this step.

(29)

3 Study on the use of Wh-questions with an interrogative pronoun

and Relative clauses in Russian and Dutch by bilingual children

3.1 Motivation and the main goal of the study

The main goal of this thesis is to determine whether bilingual children in the age of 4 to 6 are able to comprehend and to repeat Wh-questions and RCs correctly. Next to that, I would like to discover which specific type of Wh-questions and/or RCs is potentially repeatable, and in which language. The reason for choosing for Wh-questions and RCs can be explained by the following factors:

1. Wh-questions belong to the basic features of both Russian and Dutch and are of great importance to the spoken language.

2. In their main features, Wh-questions and RCs are unambiguous and transparent.

3. The use of Wh-questions and RCs differ in Russian and Dutch, which has been shown in the theoretical framework of the thesis, although the most basic meanings of these features match in both languages.

4. Both structures are acquired at early stages of language development, although limited input can be of influence on the knowledge of these structures in bilingual children.

5. The influence of both languages onto each other can be significant and visible.

3.2 Hypotheses

As expected, bilingual children need more time acquiring the basic features of their languages than monolingual children. The Wh-questions and RCs belong to the basic features of both Russian and Dutch, and, as research shows (Rojina 2005 and Van Kampen 2007, 2009) are acquired at the earlier phases of language acquisition (from the age of 1;6 years up to 6;0 years). It does not mean that these features are acquired free of errors. Moreover, when asked to reproduce a sentence with a specific type of RC or Wh-question, children can respond using other variations that are not particularly wrong, but not asked to repeat.

In this thesis, I have researched whether the bilingual children were able to repeat a sentence/grammatical structure, when it came to Wh-questions and RCs. First, I expected the children to have some difficulty with repeating the target structure if it had not been acquired yet. Second, I expected the group in the age of 5;6 to 6;5 to perform better than the group in the age of 4;6 to 5;6, since their knowledge of the target structure could be more solid than in the younger children who could give an erroneous response or even omit the target structure.

(30)

In the light of this, the following hypotheses are formulated. Hypothesis 1

The children acquire Wh-questions faster and more successfully than RCs in both languages. Hypothesis 2

The children acquire questions in Dutch faster and more successfully, since Russian Wh-questions are grammatically more complex.

Hypothesis 3

In Dutch, there is no difference in the acquisition of Wh-words, neither in type nor in acquisitional sequence.

Hypothesis 4

In Russian, the children acquire Wh-questions without preposition faster and more successfully than prepositional Wh-questions.

Hypothesis 5

In both languages, Subject Relative clauses are acquired faster and more successfully than Object Relative clauses.

3.3 Method

3.3.1 Participants

The participants of the experiment were selected via a parental questionnaire that was used as part of the PhD research by Janssen (2015, in progress). The children were recruited from Russian Saturday and Sunday schools in the Netherlands. The questionnaire stated that the children needed to be active bilinguals. This is because that these children have acquired both languages to a certain degree, which makes them suitable to this research. The children were selected on the basis of the following criteria:

- status of the parents: at least one of the parents needed to be high(er) educated (higher vocational or university degree);

- parental strategies: who spoke which language with the child, and how many hours per day these children were addressed in Russian;

(31)

- language disorders: whether the children were suffering from any kind of impairment or deficiency.

After the questionnaires had been fully analysed, two groups of participants were formed; a group consisting of ten younger bilinguals in the age of 4;2 to 5;2 with a mean age of 4;6, and another group, consisting of ten older bilinguals in the age of 5;3 to 6;5 with a mean age of 5;7. The reason for splitting these bilingual children in two age-groups was the fact that monolingual children at the age of 6;0 show a ceiling effect in their performance on the grammatical structures involved in this research (Janssen 2015: in progress). It was interesting to see whether bilingual children were comparable to monolingual children or to what degree they lagged behind with respect to acquisition of the target structures in one or both of their languages.

In summary, all of the children’s parents belonged to the middle class, lived in the Netherlands, were highly educated, and came from an urban area. The children had no measurable language deficiencies. The children had been addressed in both languages from birth and attended classes at their Russian schools in Amersfoort and Eindhoven.

3.1.1 Sentence Repetition Task (SRT)

According to Marinis and Armon-Lotem (2015), a Sentence Repetition Task (SRT) is of great value when measuring language proficiency. The SRT measures the language proficiency indirectly: not acquired structures would not be repeated. An important feature of SRTs is that the length of the sentences has to be adjusted to the memory span of the participants. SRTs are widely used, for example in populations with Specific Language Impairment (SLI). Janssen based her SRT on a longer version designed by Armon-Lotem & Meir for Russian-Hebrew children (Marinis and Armon-Lotem, 2015). Because the participants in this research were younger than the children in Israel, it has been decided to shorten the original test from 70 to 48 sentences and make the sentences containing fewer syllables (9-14 instead of 10-17). This was done in order to prevent a fatigue effect from becoming a decisive factor, and to adjust the length of utterances to a memory span of the younger children. The version used in this thesis was piloted by Janssen in Israel, with five years old Russian-Hebrew children (in cooperation with Dr. Armon-Lotem at the Bar Ilan University, Ramat Gan).

A pilot version of the current Dutch SRT was tested by Luuk van de Scheur (2012) (under supervision of Jan de Jong). Van de Scheur’s version was based on the English SRT by Theodoris Marinis. However, the participants of Van de Scheur’s pilot were 6;0 to 8;0 years old.

On the use of Russian and Dutch WH-Questions and Relative Clauses by Bilingual Dutch-Russian Children