Children’s beliefs about bilingualism and language use as expressed in child-adult conversations

Abstract The aim of this article is to describe young children’s beliefs about language and bilingualism as they are expressed in verbal utterances. The data is from Swedish-medium preschool units in three different sites in Finland. It was generated through ethnographic observations and recordings of the author’s interactions with the children. The meaning constructions in the interactions were analyzed mainly by looking closely at the participants’ turn taking and conversational roles. The results show that children’s beliefs of bilingualism are that you should use one language when speaking to one person; that languages are learnt through using them; and that the advantage of knowing more than one language is being able to talk to (other) people. The results also show that this knowledge of languages is no different from other knowledge within their world. This will probably change over time as the children enter school, and it is something in which our presence as language researchers will have played a part.


Introduction
Bilingualism from the perspective of preschool children themselves has not been much researched (however see Crump 2014). This probably reflects the fact that young children do not have the cognitive skills to answer hypothetical questions (Curtin 2001). Asking young children about their views or opinions is also problematic because children develop socially, cognitively, emotionally, and in other ways through interactions with adults (Gauvain and Perez 2006) and tend to trust adults to help them along in a conversation. Talking about language is also an awareness-raising experience that enhances knowledge of language use (Denham and Lobeck 2010), which means that talking to children about language can be seen as socializing (manipulating?) them into thinking about something they had not thought about before.
Nevertheless, by recognizing the impact of the research methodology and the researcher's presence and participation in children's preschool life, we can let children's voices be heard, or at least show the socialization processes that are taking place. In searching for adequate ways of hearing children's beliefs about bilingualism, the question of what it means to have a voice is addressed. Socialization into interactive agency ( van Nijnatten 2013) and certain beliefs, such as the value of multilingualism, are at the core of the research presented. The objective of this article is to answer the call of Spyrou (2011) to bring the interactional context into analysis of children's voices and beliefs.
This study was conducted within an ethnographically informed research project, Child2ling, 1 which aimed to reach a deeper understanding of the concepts and discourses of language and bilingualism in Swedish-medium preschools in Finland. One crucial aspect of projects on discourses about language use is that informants' verbal statements about language use are not necessarily the same as their praxis. In this article it is claimed that, even though the praxis shows that code-switching exists among preschool children (see Bergroth and Palviainen in this issue), their verbal formulation about language use is normative in the sense that languages should be used one at a time and that you use one language with one person (cf. Crump 2014).

Theoretical background
The theoretical foundation of this study is the concept of historical body, which comes from the methodological approach called nexus analysis (Scollon and Scollon 2004). Historical body refers to the experiences and ways of thinking that are our habitus, our cultural heritage. According to this concept, people have certain beliefs and ways of behaving that they themselves are (more or less) blind to, as those beliefs and ways of behaving seem to be completely natural to them. This perspective was adopted because the historical bodies of people influence their interaction with others. Experiences of being a child or an adult are also part of the historical body and as such influence the interaction between children and adults.
The relationship between child and adult does not hold reciprocal responsibilities or aims; adults are, for instance, valued as caregivers to children and thus have power and responsibilities that children do not have toward adults.
Adults also have more advanced cognitive abilities than children as well as bigger bodies, both of which are resources that affect child-adult interactions. Children and adults are (to a different extent) aware of these aspects and thus act "as a child" or "as an adult", expecting the other party to the interaction also to play their part (cf. Kampmann 2003).
The different roles of children and adults have given rise to the conceptualization of a process referred to as socialization (Parsons 1951;Speier 1970), i. e. the process whereby children are transformed into fully fledged members of society through their social environment. From this perspective, socialization is a process through which cultural norms and ways, or praxis, are transferred from one generation to the next. This way of looking at childhood and children's (lack of) participation in constructing their own lives has been problematized and questioned (see for instance Corsaro 1985;Jenks 2005). Rogoff (1995) stresses the fact that development is not a one-way affair taught by adults to children, but a phenomenon that goes both ways, with implications for all the participants and therefore also for the cultural communities in which they live. Society is regenerational, that is, we adapt to the structures, institutions and values of previous generations, but also continuously recreate it as something new.

Interactive agency and voice
One crucial aspect that differentiates children and adults in a conversation is the different status of their interactive agency, i. e. the individual's competence to make his or her voice heard ( van Nijnatten 2013). Interactive agency is developed during early childhood mainly through the child's interactions with significant others, but also with others in their immediate environment. It is thus not only semantic meaning that gets constructed in a conversation, but interpersonal meaning too. The qualities of the interactions between children and adults are important for sound agency to develop ( van Nijnatten 2013); in other words, the child's inherent dialogical nature (Trevarthen 1998) develops in different ways according to what kind of interactions are met during the first years (Laible and Thompson 2006).
Individual agency is the result of a dialogical process that makes the individual experience him-or herself as a person, being one and the same over time, having reached a functional "internal locus of control" ( van Nijnatten 2013: 31). When the term agency is used to describe an infant's innate, action-based urge to communicate (Murray and Trevarthen 1985), it describes a capacity that will dry up and maybe die if not answered by adult caregivers. Put simply, individual agency means that you have a voice and interactive agency means that you have the skills to make this voice heard. Interactive agency is the result of dialogue; learning to get a balance between oneself and others is a process that is culturally anchored (cf. Hasan 1996 on semantic variation and individuation).
In this article we take a dialogical and developmental approach to agency, using the term interactive agency when referring to making oneself heard in interactions. "'Interactive agency' refers to a performative capacity to act with a certain degree of autonomy and to take position in relation to other people" ( van Nijnatten 2013: 33). This perspective is important if we are to make statements about children's own beliefs, 2 and not about beliefs that, even though verbalized by children, only mirror those of the current adult conversational partner.
Research into children and childhood raises the question of whether anyone who has reached adulthood can ever find out what a child experiences and thus understand their perspective. In order to get to hear children's opinions or feelings, researchers have tried to find ways to reduce the differences between themselves and their subjects by acting more as a friend and putting adult responsibilities aside during their fieldwork. This is sometimes referred to as the "least adult" role (Spyrou 2011;Johansson 2012). Experimenting with roles like this in research interactions does not always work out as well as one might have hoped. For instance, Curtin (2001) comments that children can run wild, and Johansson (2012) reports that when the researchers tried to act out-of-adult roles, taking less responsibility than usual, the children reacted by saying that the adults had been too kind and had not kept order. In such cases we need to reflect carefully on what we are researching; it might just as well be children's tolerance of degrees of "strangeness" as anything else.
In order to meet the demands of The Convention of the Rights of the Child that children's voices be heard and that children be regarded as full members of society, researchers have tried to find ways around the problems associated with traditional interviews and conversations. For example, it has proven fruitful to talk to young children about a specific subject of research by getting them to draw pictures; the conversations that take place during this activity hold information about the children's thoughts and ways of thinking (see for example Harris and Barnes 2015 discussing four-year-olds' views on teachers and gender).

Beliefs
The phenomenon of beliefs is elusive both when defining what belief means and when establishing credibility for a description of an informant's belief. The theoretical standpoint in this article is in line with the Contextual Approach (Kalaja and Barcelos 2006). Beliefs are considered to be constructed in everyday interactions, and are emergent (i. e. not pre-existing), while still working as lenses through which the individual's experiences are framed. Beliefs about language and bilingualism can be detected when language is the topic of conversation or in the conversational partners' praxis. In this article we take beliefs to be someone's ideas (in this case, about language) as verbalized in conversations. The definition used in this article is "beliefs indicate that individuals accept something as true" (Aro 2009).

Preschool children's beliefs about language and bilingualism
It has been found that older school children (six to ten years old) often talk about their linguistic and ethnic identity (Sahlström et al. 2012); however, few comparable studies have been conducted with younger children. Research with children under the age of six mainly focuses on children's language awareness from a purely linguistic perspective (see for example Goetz 2003; on theory of mind, or Fedman and Shen 1971;Ianco-Worrall 1972; on the arbitrary relation between word and object). Crump (2014) is one of the few studies with young children that focuses on the children's view of multilingualism. Crump talked to children aged four to six years while playing games with them, some of which were tied to aspects of identity and language, thus introducing language use as a theme into the conversation. The children in her study were very aware of their multilingualism, but they did not quite understand why she wanted to know about their languages. They seemed to take them for granted. Crump reports that the children associated languages with people and places, and that they had a normative stance (a monolingual bias) toward when to speak to whom in which language. She also shows that the children's choice of language was affected by a person's appearance.
In research with older (school)children, the informants' language learning beliefs have been the main realm of interest (see Barcelos 2003 for an overview). The informants' role is thus that of learner rather than language user. However, Martin (2012) investigated how students experienced their linguistic identity, using a multimodal research approach in order to study the impact of bilingual education on identity. The results of this (partly) non-verbal approach reveal an increase in linguistic self-awareness and language awareness in the older schoolchildren. Students whose languages were not supported within the education program did not include their languages in the same abstract way as those whose languages were supported, suggesting a strong connection between informants' experience of their linguistic identity and the kind of language socialization taking place in the educational setting. A similar conclusion is drawn by Kolb (2007), whose study focuses on the students' language learning beliefs. In her study, Kolb shows that eight-to nine-year-old students' beliefs can be divided into two strands: a communicative view on language learning (holistic-creative approach), or a view that focuses on form and the language system (analytic-reproductive approach). Although this is in line with the beliefs of adult learners, she proposes that language learning beliefs come into being to some extent through cultural transmission and that, during their primary years, children learn the analytic-reproductive approach.

Method
The aim of this study is to investigate children's beliefs about bilingualism and whether there are differences in beliefs that could be traced to their attending Swedish-medium preschools in different linguistic environments. In order to obtain a corpus of data from which comparisons could be made, we conducted similar ethnographic studies and played the same eliciting games at three different sites.

Research sites and informants
The study presented in this article is part of the research project Child2ling, which included children attending three different Swedish-medium preschools in Finland (for a more detailed description of Child2ling, see Bergroth and Palviainen this issue). Finland has two national languages and parents have to register their child as having one of those languages as their mother tongue. They also have to choose whether the child will attend a Swedish-medium or a Finnish-medium school. Within Child2ling, nine children were selected to be the focus of the study. The children selected lived in families where, according to the parents' own report, one parent spoke mainly Finnish and the other mainly Swedish. The three research sites were chosen because they were situated in areas with different linguistic demographics in terms of the two national languages. According to the Official Statistics of Finland (2013), Site I has 81 % inhabitants registered as having Finnish as their mother tongue and 6 % as having Swedish; at Site II the ratios were 96 % Finnish and 0 % Swedish; and at Site III the split was 70 % Finnish and 23 % Swedish.
In this article, however, the longer examples used are all from Site III. This is not a mere coincidence, but reflects the fact that the children at this site were particularly eager to join in the research games, and it was clear that participating in activities in which they were asked to contribute in accordance with the adult's expectations was within their repertoire of practice (Rogoff et al. 2007). The children at this facility were all more or less fluent in Swedish. All informants in this research project were between three and six years old.
The longer transcribed excerpts show examples of what Ester (5.1 years old), Eva (6 years old) and Ella (4.6 years old) said about language during the recorded activities (eliciting games), but the analysis also includes interactions observed between children and the researcher at all three sites. Ester, Eva and Ella mostly spoke Swedish during the visits of the researcher to the preschool. The author overheard Ester switch to Finnish on a few occasions when speaking to certain friends, and occasionally she codeswitched during her interactions with the author. In contrast, the author never heard Eva or Ella codeswitch, other than mentioning Finnish words in the context of talking about the different languages.

Research procedure
Two different field methods were used to collect data for this study: ethnographic observation and eliciting games. Some of the eliciting games were recorded. Methodological decisions concerning which interactions with the children to record were made on the basis of a study in which the focus lay on getting to know the children and their everyday preschool routines (see Almér 2015). The research games used to generate the data that is analyzed here were an attempt to follow up what the children themselves had initiated during the researcher's earlier visits to their preschools. In this way the researcher tried to connect with the children closer to their everyday lives rather than bring them closer to the research agenda; however, the children did not participate in the actual planning of the games.
The researcher presented herself as someone who came from Sweden and she told the children that she did not know any Finnish at all. The preschool teachers answered the children's questions about why the researcher was visiting with statements such as "hon är intresserad av språk" (She is interested in language). The children seemed to conceptualize the researcher as a "language person", spontaneously telling her which languages they spoke. They also spontaneously told her the language situation within their own families when they sat down to paint or play games around a table.
The numbered examples in this article are excerpts from transcriptions of the conversations that took place during three different games that were created to elicit talk about language. One game (Story) is a narrative situation based on pictures that the children had not seen beforehand, but around which the children and the researcher constructed an oral story together. The second game (Interview) is an interview situation in which the children chose pieces of paper from a pile and talked with the researcher about the question written on the paper. In the third game (Seven Friends), the researcher presented seven dolls that she claimed had gone around a town in Finland to explore and get certain things done without using any verbal language. For example, the researcher showed how one can find the bus station by looking at the signs picturing buses, or how a picture of a bicycle helmet can be shown if you would like to ask someone where to buy one.
Research games were recorded on 14 occasions with two to five children present each time. The children did not talk about language in all of the recordings: altogether the recorded material included 33 episodes about language or language use, and in nearly every case it was the researcher who initiated these episodes. The situation was similar in the non-recorded research games and in other interactions during the ethnographic study, although there were occasions when children brought up the subject of language use spontaneously. These were the situations that the researcher tried to recreate in the eliciting games; however, the attempts yielded no results.

Methodological considerations
Researchers have written a great deal about their experiences of doing research with young children and they have offered others much good advice (Curtin 2001;Christensen and James 2008;Albon and Rosen 2014). It is somewhat ironic that two of the most important things one learns when doing research in this field are firstly that one needs to get to know the particular children who are going to be the informants in that particular study (Curtin 2001;Barley and Barth 2014); and secondly that one needs to remain flexible in one's intentions about how to proceed with the research (Crump and Phipps 2013;Crump 2014). This means that to some extent, knowledge gained earlier about methodological aspects of the research process are of no help when embarking on new projects. This problem emerged in the project Child2ling when moving between the different research sites, making it hard to collect comparable data.
The process of developing methods to encourage conversations in which the children would talk about language use and bilingualism led the researcher to question the quest itself. This came about mainly because of the variables that could not be changed at the three different sites, but also because of my historical body.
For example, the intention of doing research with the aim of creating situations where the subject of bilingualism came up in conversation spontaneously, without me posing any obvious questions, was never truly realized except for the children telling me which languages they spoke and which languages their family members spoke. During the observation, I more than once asked the obvious questions I wanted to avoid, and by so doing limited the possibilities of the range and kind of thoughts that could be formulated in interaction with the children. This behavior also laid the foundation for the relationship between the children and me, as I came to be "the language person". This (so far as I can see) influenced our communication on several occasions. The relationship between us had implications for what kind of utterances emerged, such as verbally mediated beliefs about language use. Even if these beliefs were un-elicited, they were most likely an effect of my being there as a non-Finnish speaking "language person".
The fact that I met the children at the Swedish-medium preschool they attended also played a crucial role in the interactions. The older children were aware that the two languages, Finnish and Swedish, were supposed to be separated in certain situations. Many children, for instance, told me with whom in their close environment they spoke Finnish or Swedish, as if they never used more than one language in their interaction with those people. Prout (2011) points out the research advantages of giving up dichotomies such as child/adult, being/becoming, nature/culture, or agency/structure and, instead, of adopting a policy of fluidity and relationality. Also Spyrou (2011) discusses the importance of analyzing the whole interactional context, not just the utterances, when studying young children's voices. In this article the results of the recordings show that the role of being a "traditional adult" is not easily shaken off or overcome. They also show why the researcher analyzing children's thoughts about language and bilingualism in child-adult conversations still needs to take the categories of child and adult into account. I use these categories, while also responding to the call made by Prout (2011) andSpyrou (2011) for fluidity, relationality and context to be brought into the analyses.

Methodological approach of analysis
When analyzing the field notes from the ethnographic study it became obvious that both the researcher's relationship with the children and aspects that can be ascribed to their historical body (Scollon and Scollon 2004) were mentioned in the entry for every visit. Rather than try to exclude them, it was clear that we needed to include these aspects in the analysis. The advantage of this kind of research is that it shows the boundaries of what we cannot know, rather than constructing new knowledge.
To analyze the conversations we use an approach influenced by interactional linguistics (Lindström 2008;Norrby 2014) and conversational analysis (Heritage 1984). Our primary focus is on explaining how interactional and sequential aspects of the conversations provide the framework for understanding the meanings that are constructed.
The analysis is also informed by field notes, and by interactions among the children, as well as by their interactions with the researcher outside the recorded games. Such interactions brought an extra dimension to understanding the sense making that took place in the recorded conversations.

Results
In this section the children's beliefs about language and bilingualism are presented. This is done through four transcribed excerpts from the eliciting games and by reference to the ethnographic fieldwork. The excerpts were chosen as they exemplified the same beliefs on language and bilingualism that emerged in the non-elicited interactions.
We spoke Swedish in all the conversations from which the following examples are taken. The Finnish words are citations of hypothetical persons speaking Finnish. In the transcriptions, words spoken in Finnish are marked in bold text, as are their equivalents in the translation. Some of the communication between us was carried out by non-verbal means such as glances, pointing, body movement or the use of props, but in the transcriptions the utterances are presented one after the other, with only a few comments on body movement. During the interactions we all sat on a soft carpet on the floor of a library. Although the children were moving around, stretching and changing positions, we all paid attention during the conversations.
We found that children's beliefs were centered on three themes, and that there were no differences between the three sites concerning those themes or concerning other aspects of the children talking about language. The first theme was that specific interactions should be conducted in one language, the second that language does not stand out from other kinds of knowledge, and the third that using a language equals learning the language.
A particular example of the difference between praxis and talking about language use is when an informant says that she cannot speak a certain language in the very language she claims not to be able to speak. This is captured during one of the recorded interview games (Example 1), when Eva and Ester were choosing pieces of paper with questions to answer. Ester used her necklace and headband as a hearing aid throughout this particular day, although in fact she has no hearing loss.
(9) If you are down to:wn or x as someone then you then you see someone so that one can say (2)  At the end of the excerpt Ester says that one should speak Finnish to other people and justifies it by claimingin Swedishthat she knows no other language. Later on the same day she tells me that she does know Swedish, which shows the inconsistencies people verbalize. This phenomenon of claiming not to know a language while speaking that very language is not something that is met with only in young children. The author makes no further claim as to whether this constitutes a language belief or not, but will focus rather on other interactional aspects in the example. Still, inconsistencies like this should be kept in mind in order to avoid jumping to conclusions when analyzing utterances about languages. Also, sometimes utterances stating which languages one cannot speak should be acknowledged as the informant's way of voicing that they do not want to use this particular language in the research situation, or that they do not identify themselves with it at all (cf. Maguire 2005).

Monolingual bias concerning interactions
In this section the interactional aspects of the episode in Example 1 will be described. The interactional aspects show how a monolingual bias is mediated and how each of us uses our interactive agency. Eva made her voice heard by taking the initiative to read and in that way positioning herself as a reader, to be counted in the group of readerssomething that came up with the five-year-old children. I gave her permission to read because I wanted to take a positive approach to the children's contributions to the interactions. However, by so doing I also created a gap between Eva and me on the one side and Ester on the other, since Ester was not yet able to read.
In Utterance 1 Eva reads the question Hur vet man vilket språk man ska prata med någon? 'How do you know which language you should speak to someone?'. I had written this question and it confirms that I carry with me the belief that one uses one language when talking to someone (vilket språk 'which language' is singular). I noticed on several occasions during my fieldwork that I had this monolingual basis for my beliefs about language use. This belief is not in line with my knowledge of the actual state of affairs: I know that people mix different languages in conversations with one and the same person. It should be noted, though, that this same normative stance (monolingual bias) was present in all of the three preschools even though no one objected to the children using the two different languages in the same conversation. For example, the children at Site II were asked in which language they wanted to write their Fathers' Day card; mixing languages was not presented as an option. The monolingual idea could also be found when, in different contexts, the children referred to specific languages as, for example, pappas språk 'daddy's language' or Elsas och Annas språk 'Elsa's and Anna's language'. The latter example (captured at Site I) comes from a conversation about an animated movie in which the sisters Anna and Elsa are princesses. The children were referring then to English, but in one conversation it was said that the princesses spoke Swedish (one of them had apparently seen a translated version of the movie). While one of the children actually came to the conclusion that the princesses knew both languages, the other children firmly maintained that the princesses' language was English. They also considered English to be a beautiful language because the princesses spoke it, just as some of the children in the study considered Swedish or Finnish to be beautiful because it was "mother's language" or "father's language".
My historical body also shines through in Example 1, Utterance 2, when I assume the right to assess Eva's reading skills. My assessment in this context is most likely tied to the fact that I believe reinforcement motivates and keeps children eager to learn. I comment on it here to illustrate how the role of being an adult (a norm-carrying caregiver) sneaks into the conversation. In this particular case, Eva does not reply verbally to my assessment but answers the question she has read, which I had by then repeated. Her answer mediates the thought that one person speaks one language. In Example 1, Utterance 3, Eva says that one should talk to a person in the same language as the person addresses one in: så kan den säga på finska heippa så måst man så kan, så måst man så måst man också säga tillbaks 'so that one can say in Finnish hello then you have to then you can you have to have to reply'). Here she is using her interactive agency to clarify that the answer to my question is not a specific language, but rather that it could be any language as long as one answers in the same language as the one in which one is addressed.
I then turn to Ester to hear what she has to say on the matter, but she claims that she has not heard as her hearing aid was out of place. I play along and wait until she has put on her "hearing aid" before I ask again. Having been excluded during the actual reading part, she uses her interactive agency in this way to reclaim her place within our group of three. I could not have ignored her make believe and at the same time kept her attention. Her answer is, as already mentioned, nå för att (.) jag kan int nå andra språk 'Well because (.) I do not know any other languages' (Example 1, Utterance 9).
In Example 2, Utterance 1, I ask which language(s) 3 the two people in the picture we are looking at are speaking while they are baking. The episode is from the game to create a story, and my questions are angled to explore what we can imagine happening within the story world.
Example 2 Participants: The children Eva 6 years and Ester 5.1 years, and the researcher. hela tiden så dom blir helt råddiga i huvudet.
All the time then they will be all mixed up in their head. ((giggles)) Even though I started this episode (Example 2) by openly asking about which language(s) the couple in the picture are speaking, it seems that neither I nor the children can let go of a monolingual belief; it is part of our historical bodies. Eva starts by answering svenska 'Swedish' (Utterance 2) and later on in the conversation I ask which language, i. e. singular, they speak (Utterance 6). At this stage Ester joins in the conversation, saying that she does not know, but in Utterance 9 she proposes finska eller svenska 'Finnish or Swedish'. When I ask for clarification in Utterance 10 by saying ha vilket som? 'ah, which one?', she changes her word choice and says finska och svenska 'Finnish and Swedish' (Utterance 13) which, in the ongoing situation, I interpret as her suggesting that they use both languages while baking. I therefore again ask for clarification, wanting Ester to tell me exactly how they use both languages, but when I ask växl[ande]? 'switch[ing]?' (Utterance 14) Eva, not Ester, again verbally mediates the belief that two people talking to each other should be using only one language. She says it would be really strange if the woman said 'good morning' in Swedish (god morgon) and the man answered in Finnish, Huomenta 'good morning' (Utterance 15). She concludes that switching between two languages hela tiden 'all the time' would make the speakers helt råddiga i huvudet 'mixed up in their head' (Utterance 19). 4

Using and learning languages
When I asked the children which languages were good to know, their reply was Finnish, Swedish and English; some of them said that English was the best language to know. Spanish and German were also mentioned. The only reason for learning a language that the children verbalized when I was present was being able to speak with people who spoke another language. In Example 3, another episode from the Interview game, both Ester and Eva say that one needs to know languages in order to speak to other people. Ester says that one needs to speak another language om det kommer nån olika 'in case someone different comes' (Utterance 4) and Eva explains that one needs to know English om … nån säger på ett typ på engelska hej 'if … someone says in kind of English hello' (Utterance 6).
Then you can answer first why you think you should learn more than one language. 2 Ester: man ska lära alla språken finsk, svenska och alla såna möjliga. You should learn all languages, Finnish, Swedish all that are possible. 3 Researcher: för att?
Because? 4 Ester: för att att man lär sig att prata om det kommer nån olika så måste man om man så måst man riktig prata en annan språk och därför måst man lära sig språk. 'Cause you learn to speak in case someone different comes then you have to if you then you have to really speak another language and that is why you have to learn languages. 5 Researcher: mm (2) jag förstår vad tycker du varför ska man lära sig fler än ett språk? Mm.
Well if (2) you for example sometimes if you are downtown and someone says in a kind of English hi you have to know (.)in English. 7 Researcher: Svara Answer 8 Eva: ((nods)) 9 Researcher: mm just det bra.
In this excerpt the children initiate the idea that one needs to speak several languages so that one can speak to other people. They do not refer to their own bilingualism or their friends' bilingualism, but to people who come around or who happen to approach one and address one in another language in town. Both Ester and Eva are bilingual, and they claim that learning more than one language is important in order to be able to communicate when meeting other people. This is a theme that I recognized in the speech of many of the other children at the three different sites and that was sometimes voiced when children were talking about their experience of traveling abroad. Both Ester and Eva answer the question why they think they should learn more than one language, and I confirm and end the episode by affirming their answers. As already stated, it seems like the children's own bilingualism is "invisible" to them.
The children had similar answers to the question of how one learns a languagethey all verbalized variants of the belief that "using a language is learning it". An example from another research game than the ones transcribed to a longer extent in this paper is Ester saying Man lär en språk om man bara lär sig, enkelt. Man lär sig språk om man försöker bara, nånting mera? 'You learn a language if you learn it, simple. You learn language if you just try, anything else you would like to ask?'. In Example 4, Utterance 7, Ella concludes that she would probably learn more English if her father started to learn more English (implying that if he did so, he would also read in English). This episode comes from the game Seven Friends and was triggered by our looking at a postcard depicting a cat sitting down to read.
My mother re-only once reads with me all other times only my father reads. 2 Researcher: alltså det är pappa som är storläsaren hemma?
Oh so it is your father that is the big reader at home? 3 Ella: Jå.
In whi-in which language does he read then? Yes that is true if you read in English too. 9 Ella: mm men (2) det är speciellt alltid så går vi bara nära mitt dagis till bibban för att den är ju ganska nära nånstans här.
Mm but (2) that is always special we only go somewhere near my preschool to the library because it is quite near somewhere here. 10 Researcher: Aha Aha In the conversations around the games I made an effort to follow up what the children said in order to boost their interactive agency. In Example 4, Utterance 4, I asked which language Ella's father read to her in, anticipating that the answer would be which language he most often read to her in, but Ella proved to have another view on what I had asked. In Utterance 5 she tells me that he läser på finska fall det är finska, läser på svenska fall det är på svenska 'reads in Finnish if it is Finnish, reads in Swedish if it is in Swedish'. It was quite common that my questions were understood in other ways than I had intended them to be. In Example 4, as on several other occasions, the children led the conversation in an interesting direction, giving me an insight into the fact that they had a functional interactive agency to make their voices heard. Sometimes, however, this meant that there was no mention at all of language in the conversation, or the conversation touched upon language only with an animal theme, for example when they were teaching me that a Swedish dog says vof vof vof but a Finnish one says hau hau hau. It appears that none of the dogs was bilingual, or at least not practicing code-switching.
Ella's hypothesis that if her father learned more English she might learn more English shows this capacity to use her interactive agency and make her voice heard. She does not answer my question, but continues with her own line of thought and manages to get my attention. This is a strong indication of her own belief rather than an example of her copying mine. While this is a belief that she has probably been socialized into at home and in preschool, nevertheless that does not diminish the fact that she stood up for her own belief in the conversation analyzed here. The fact that she can use it for her own purpose shows that it has been appropriated.

Language and other knowledge
Apart from talking about their own and their friends' language knowledge or pointing out the shortcomings in mine, the children never mentioned language proficiency, language use or bilingualism without meor some other adultasking about it (as in Examples 1-4). This does not mean that they never talk about language or mention it, but it does indicate that it is not a matter of everyday concern. In this section we have no recorded interaction from which to take any examples, only the ethnographic observations. On the occasions when I or another adult asked about language knowledge or language use, the questions were answered in such a way that it became clear that, as far as the children were concerned, there was nothing special about language: they did not distinguish language knowledge from other knowledge. They said several things that could easily be turned into anecdotes about "cute kids", but doing this diminishes children: one is not then taking the children's perspective, but forcing an adult perspective on their statements. For example, one little girl mentioned which languages her friends spoke in the same sentence as she told me that she had pinworms, and a little boy saidgiving equal importance to both skillsthat that he could speak three languages and spin around really fast. This is an important point, because it focuses on the socialization process that takes place during the preschool years. When the children leave kindergarten and start school they will (most likely) know that language knowledge is generally considered more important than spinning round fast, but at this point the children did not show that kind of awareness. The children's primary attention is elsewhere; highlighting their thoughts about bilingualism is therefore rather false, and adultcentric, i. e. interpreting what children say in the light of what adults think they should say (van Ausdale and Feagin 2001: 3). Adults' questions and our presence as researchers will guide them little by little to give their attention to questions about language and language use.
The process is not, however, unidimensional. The children's use of their different languages gives them experiences that in turn can modify or change their beliefs. The children are active, and it cannot be claimed that this comes about only because of adults' questions and guidance. Some of the children take action to help out, by meta-communicating about language or telling someone about others' language (in)competence. When I tried to say something in Finnish they were always happy to laugh at me, or they tried to get me to repeat something until I got it right, so there were aspects of language that captured their attention from time to time. See for example Bergroth and Palviainen (this issue) and Boyd et al. (this issue) for other examples of children joking with one another about the names of different colors.

Conclusions
The aim of this study was to investigate children's beliefs about bilingualism and to explore if the beliefs they held differed because of their living in different language areas of Finland and attending Swedish-medium preschools. The primary result presented in this article is that there were no site-connected differences between the children's verbal utterances about language use. In fact, the children in this study did not mention or talk about language knowledge or bilingualism unless it was either directly relevant due to (perceived or actual) language problems in the communication situation or was prompted by their perceiving me as "a language person". This result shows that children who have been brought up bilingually take bilingualism for granted. As this article furthermore demonstrates, such children are also being socialized into seeing bilingualism as a concept that is worth focusing on in conversations. In addition, when the children answered questions concerning their language use or language knowledge, they did not distinguish this knowledge from other kinds of knowledge or information. As such, it is possible that our presence as language researchers will contribute to the establishment of language knowledge as something they regard as important. Crump (2014) concludes that the children in her study were aware and proud of their bilingualism in relation to relatives who did not have the same competence. I was told by the teachers and by my research colleagues that I was unusual in preschools in Finland in that I spoke Swedish but did not speak Finnish. They said that most, if not all, of the children had never met anyone who knew Swedish but did not also speak Finnish. This means that to many of the children, their bilingualism was not exceptional: the Swedish speaking people around them were all bilingual. This was not the case, however, in all three sites. At Site II, which was situated in a mostly Finnish speaking area, there were children who knew people who did not speak Swedish. This means that even if the children did not know anyone who spoke Swedish without also knowing Finnish, it was not the case that everyone in their environment was bilingual.
The children verbalized the belief that one person spoke, or should speak, one language at a time with another person (cf. normative stance, Crump 2014). It could be that my role as an adult leading an activity affected their answers. The sequential mode of a conversation is also crucial: an utterance is built on the one that goes before it, and we tend to answer questions without questioning the question; when I asked the children which languageusing the singular formit was closer to the question to answer by mentioning one language rather than two or more. Nevertheless, the analysis shows that on several occasions the children voiced specific formulations that were not in line with my questions, reinforcing my argument that the children were expressing their own beliefs. Also, they voiced this monolingual bias even when I did not specifically ask about it.
The children described learning a language as a process that takes place by hearing or using the language. They were not yet acquainted with the idea of formal language education (see Kolb 2007 for primary school children's beliefs about language learning). They also claimed that the usefulness of knowing more than one language was being able to talk to people who speak other languages. I close this article with some words of Ester's on this subject (Example 3, Utterance 4), acknowledging the welcoming air these particular words have: '… in case someone different comes then you have to […] really speak another language and that is why you have to learn languages'.