Abstract
In foreign language classrooms, students and teachers together create practices for language use; they talk the institutional context and institutional identities into being. According to Lave and Wenger, learning can be seen as a deepening process of participation in a community of practice. Learning a language in a classroom can be regarded as gradual changes in language use that are accomplished through participation in collaborative work. In that case, the study of variation in language alternation, code-switching and the activities the students and the teachers are engaged in can contribute to understand language learning in a foreign language classroom context. The purpose of this article is to describe and analyse how code-switching as a local practice is used in teaching and learning Swedish as a foreign language. The context of the study is classes in Swedish as a foreign language in upper secondary schools in Finland. The material consists of video recordings of one lesson in second grade where the students have been studying Swedish for 5 years. The article focuses especially on the practices of language choice in repair. Students tend to keep up a division of labour between FL used for institutional work and L1 as the language reserved for interaction. The article discusses the conditions of learning repair practices in an FL classroom.
Keywords
Introduction
In a foreign language (FL) classroom, both the FL and the first language (L1) are present, and used, in different activities and to different extents, by both teachers and students. 1 From a socioconstructionist point of view, an FL classroom is an institutional context where students and teachers construct their language use together and create practices for the use of L1 and FL (cf. Auer, 1998; Cromdal, 2005). In order to understand the premises for learning an FL in a school context, one starting point is to study how the students use their language(s) in their interactions in institutional encounters. In the FL classrooms, students with the same L1 often start conversations in the new language with very little previous knowledge of the language taught. Constructing an orderly bilingual interaction, they establish a pattern for language use (cf. Grosjean, 1982). With language learning seen as taking place during participation and as embedded in the structures of social activities, it is relevant to determine in which activities and with what methods the students construct practices for alternating between languages (cf. Firth & Wagner, 2007; Kohonen, 2006; Lantolf & Pavlenko, 2000: Lave & Wenger, 1991). Using an FL, a learner can, in interaction, gradually start to construct and shape an identity as a user of FL or L2.
Studies with a conversation analytical orientation have shown that second language or FL speakers, regardless of their proficiency in the L2, are interactionally competent and draw on the resources on hand in different dialogical situations (Firth & Wagner, 2007; Gardner & Wagner, 2004; Seedhouse, 2004, p. 98). The Common European Framework of Reference for Languages (Council of Europe, 2001, p. 4) introduces to FL education the concept of a plurilingual speaker. This concept emphasizes the fact that individual speakers build up ‘a communicative competence to which all knowledge of and experience of language contributes and in which languages interrelate and interact’. The FL classroom represents one type of institutional context with certain goals and activities defined locally and globally in alignment with the institutional aims (Seedhouse, 2004).
The purpose of this article is to analyse repair sequences where students use their L1 (Finnish) to solve problems and to create understanding in interaction during different tasks in FL (Swedish) classrooms. The context of the study is upper secondary schools in Finland, and the focus is on peer interaction both in plenary talk and during group work. Taking a conversational analytic framework as its base, the analysis will show how the students, by switching code, talk into being the institutionality of the FL classroom. They frequently create a division of labour between L1 and FL, switching code into L1 systematically during repair or similar work. The questions explored in this article are why and how the plurilingual students code-switch in problem-solving situations. What does this phenomenon tell us about the students’ competence, the linguistic resources available to them, and, above all, their understanding of the interaction in the FL classroom?
In the FL classrooms studied here, students use code-switching especially when doing other-initiated self-repair or candidate understandings (for definition, see Code-switching in candidate understandings). The problems to which they orient can be divided roughly into the following two categories. First, repair is used, as in all kinds of interaction, as a way to deal with problems of hearing or understanding in order to continue the conversation. In an FL classroom context, this means getting on with the task according to instructions. Second, students also focus on problems probably not found in everyday conversations but typical of the institutional interaction in the classroom, such as keeping up with the current agenda, or problems related to the texts on which the interaction is based. The analysis presented in this article strives to participate in the current discussion on the relationship between language use and learning by showing what implications the systematic use of code-switching in repair can have on learning of a language in an FL classroom (cf. Firth & Wagner, 2007). 2 (See Appendix 1 for the transcription key).
Bilingualism in Finnish education
For practical purposes, the term FL will be used in this article for Swedish as a subject taught in Finnish schools. As Finland is a bilingual country with Finnish as a majority language and Swedish as a minority language (cf. McRae, 1997, chap. 4G), Swedish is a compulsory subject in all Finnish secondary and upper secondary schools. In most Finnish schools, the classrooms where Swedish is taught to Finnish students can be defined as institutional settings for bilingual interaction where both students and teachers share the same L1, namely Finnish. With the exception of immersion education in Swedish, mainstream secondary schools in Finland teach Swedish not as a second language but as an FL (Keski-Raasakka, 1998, p. 2). The minimal length of study is 3 years, but students going through upper secondary schools usually study Swedish between 6 and 9 years. As of today, we know that in Finnish schools, students use both L1 and FL in classrooms where Swedish is taught, but there is little knowledge of in which situations and in which functions the languages are used.
Analysing interaction in the FL classrooms as institutional discourse
Although conversations in an FL classroom are carried out in an institutional context, for example, with specific pedagogic goals and turn-taking organizations, they are also situated in a reality where social meaning is created locally, there-and-then in the microcosms of different classrooms (Rampton, 2006, p. 142; Tainio, 2005, 2007). In FL classrooms, local, national and global practices are met. As FLs are learnt in classrooms all over the world, it is important to determine what sort of interactional practices they share, regardless of location, and what kind of practices are the product of, for example, national educational traditions. Several studies have shown that the implicit rules that are learned and maintained in the classroom can be crucial for educational outcome (Mehan, 1998, p. 249). When phenomena in FL and L2 classrooms are studied, it is important to pay attention to the different conditions that have an impact on these phenomena and to the contexts and activities in which the interaction occurs (Martin-Jones, 1995; Ohta & Nakaone, 2004; Swain & Lapkin, 2000). Rogoff (1990) addresses the question of classroom contexts as sites of local cultures of knowledge production. Macbeth (2004, p. 714) also points out that in school contexts, there are probably differences depending on the settings, cultures and grade in question.
Typical traits of classroom conversations that are shared with other institutional interactions are specific goals, specific constraints on participation and inferential frameworks and procedures (Drew & Heritage, 2001, p. 22; Seedhouse, 1999, 2004, pp. 206–207). Drew and Heritage (2001, 28 ff.) describe how these traits, the fingerprints of institutionality, can be realized in interaction and talked into being by the participants through orientation to them (Heritage, 1984b, p. 283, 1997, p. 161). In the FL classrooms, two institutional dimensions, interaction based on texts and the agenda, are relevant for the analysis of repair and code-switching reported in this article.
FL classroom interaction characteristically evolves round texts (cf. Heritage, 1997, p. 161). Instructions and materials for tasks aimed at stimulating discussion are presented in writing and recontextualized in different interactive situations. Depending on the theoretical framework, researchers have described this typical feature of classroom interaction in different terms. According to Gustavsson (1988, p. 221), language is treated as a decontextualized object of discussion during lessons, but the participants also recontextualize linguistic items by taking them as a ‘point of departure for creating a new and wider context around’ them. 3 Barton (1994, p. 181) defines the ways that students learn to make meaning at schools as literacy events consisting of talk around texts. Describing different ways in which students give text an oral form in an EFL (English as a foreign language) classroom, Pitkänen-Huhta (2003, p. 11) shows how literacy practices are ‘established, maintained and contested locally in particular literacy events’. She does not, however, analyse the division between students’ use of FL and L1 in their discursive practices.
In many FL classrooms worldwide, students share the same L1 with which they build up their personal relationships in conversations. The use of FL in interaction in class is closely tied up with the purpose of teaching it, whereas students use L1 in everyday conversations outside the classroom and in off-task talk in the classroom (Seedhouse, 2004, 70 f.). Creating mutual understanding on a turn-by-turn basis seems to be regarded by FL students as interactional work belonging to everyday interaction and they choose L1 to deal with the creation of this understanding. Institutional FL interaction seems to be interpreted by the students as talk based on texts. As the analysis presented in this article will show, problems occurring during tasks done in FL are dealt with in L1 as problems of everyday interaction, whether they are problems of hearing and understanding each other or difficulties related to the institutional context. The same goes for more specific institutional problems, such as understanding the texts the class is working with, which are tied up with the multimodality of the institutional context of the classroom, and problems having to do with the agenda of the lessons.
Important for the organization of the classroom is the overall agenda of the teacher concerning the order in which the different activities are realized and the length of time reserved for them. The activities or tasks also follow a local agenda based, for example, on written or spoken instructions. In the classroom activities, all the participants constantly seem to orient towards focusing on listening to others and saying or doing the right thing at the right place in a sequence (Linell, 1990, p. 26). 4 Some of the students’ problems illustrated in this article have to do with keeping up with what the teacher or the other participants are saying or doing.
Code-switching in conversation
An important factor motivating the study of code-switching in interaction in an FL classroom is studies of bilingual language use in general. Although the context is institutional, language choices in a classroom are likely to be made on the same premises as in other bilingual settings where a pattern for language use is established between the participants from the start and is then followed later on (Grosjean, 1982).
It has been shown in the literature of code-switching that the choice of code in bilingual interaction is often ‘open to local processes of language negotiation and code selection’ (Auer, 1998, p. 3). Following Gumperz’ notion of contextualization cue, Auer (1995, 1998, p. 1) has established (a theory of) an analytic framework for studying language alternation as a sequentially embedded verbal action. Code-switching can be seen as a ‘contextualization cue, which effectively contributes to the interpretation of utterances in context’ (Alfonzetti, 1998, p. 207; Cromdal, 2005). In analysis of code-switching, it is necessary to take into account the speakers’ linguistic preferences, their competences and the conversational context (Alfonzetti, 1998, p. 182; Auer, 1998, p. 8). In institutional contexts, such as classrooms, it is important to understand the interactional projects in which the code-switching occurs (Cromdal, 2003, p. 747).
In the literature on code-switching in everyday conversations, there are some indications of code-switching used to initiate and complete repair sequences, although the subject has not been studied systematically. Auer (1998, p. 4) emphasizes the fact that code-switching is more likely to occur in initiative turns rather than in responsive turns. Alfonzetti (1998, pp. 180, 191) shows how bilingual speakers of Italian and an Italian dialect switch code in order to structure their talk and mark the beginning and end of a repair sequence. Milroy and Wei (1995, p. 151), in their data consisting of conversations between bilingual speakers in a Chinese community in England, mention that the combination of code-switching and repair sequences is common; but they do not include examples.
The use of code-switching in bilingual and L2 classrooms has been studied with different aims and in different theoretical frameworks (for an overview see Martin-Jones, 1995). In conversation analysis (CA)-oriented studies of bilingual students engaged in group work, Cromdal (2003, 2005) has shown how the interactants switch code in order to accomplish different interactional goals. In an article relevant for the present analysis, Cromdal (2005) shows how students who are bilingual in Swedish and English code-switch when writing an English text together. During the task, the students create locally a division for language use in which the sequences concerning the task, that is, producing the text, are in English; and other interaction is in Swedish. Although Cromdal’s analysis is not concerned with repair, and although his informants are L2 speakers of English, the language choices he has fleshed out show certain similarities to my data. In Cromdal’s analysis, English as the institutional language is reserved for the interaction concerning the creation of a text, the institutional activity, whereas the L1 (Swedish) is used for other interactional purposes.
Other studies of code-switching in school contexts have been concerned with the impact of code-switching on aspects of learning. In a conversational analytic study of how bilingual students orient towards a monolingual code in a Swedish primary school in Finland, Slotte-Lüttge (2005) claims that code-switching enables the students to take the floor more quickly, to participate in the classroom interaction in a meaningful manner and to build up their knowledge. There is, however, at the same time, a price to pay for participating as a bilingual in a monolingual discourse at school, as working with two languages takes up more time and does not always result in accurate use of language. Sociocultural studies of the use of L1 in L2 and FL classes also deal with positive and critical aspects (Nikula, 2005, 2008). For instance, Swain and Lapkin (2000, p. 268) suggest that the use of L1 in immersion contexts supports L2 learning. They see the use of L1 as a helpful tool that serves students in performing tasks. The variation in the use of L2 can be explained partly by what is required in a task, the activity in which the students are engaged, and partly with differences in the students’ achievements. According to Swain and Lapkin (2000), weaker students need to use L1 more.
In the following sections, I will present my data and the previous literature on repair. The analysis will focus on code-switching sequences where students switch from FL (Swedish) to L1 (Finnish) during repair when solving problems in interaction. Sequences in L1 other than problem-solving sequences are, however, not discussed in this article even though in the data the students also code-switch frequently for other purposes. Non-institutional off-task talk is, for example, conducted entirely in L1 (Finnish) (cf. Seedhouse, 2004).
Data
The analyses in this article are based on data that consist of a 90-minute Swedish lesson taking place in the second-year class of a Finnish upper secondary school located in a town in southern Finland. The students are attending a so-called B-course, which means that they have been studying Swedish for 5 years. The class consists of 17 students and a male teacher who uses mainly Swedish during the curricular activities. The theme of the lesson is becoming dependent on something. The structure of the lesson is typical of language lessons where plenary talk alternates with tasks done in small groups (cf. Sahlström, 1999, p. 64). In this article, the focus is on two activities, in which five student pairs are engaged: (a) Liisa and Maija, (b) Virpi and Niina, (c) Seija and Katri, (d) Anni and Krista and (e) Matti and Kari. The lesson was video filmed with two cameras, and five minidiscs were placed on the desks in order to record the desk talk. The cameras only manage to give an overview of the whole class, so the analysis is mostly based on the audio recordings.
Two curricular activities
During the first activity (i), which lasted for 10′12′, starting 22′30 into the lesson, the teacher asks the class to give the Swedish equivalents of certain Finnish words which occur in a Swedish text that the students have just read. At the beginning of the activity, the teacher instructs the students not to answer to him but to the student sitting next to them. The students’ quiet answers to each other are followed by a turn by the teacher, who repeats the correct answer to the whole class. The turn-taking pattern does not follow the typical initiation, response and follow-up (IRF) cycle in a plenary classroom, where the teacher has control of the whole sequence and questions posed by the teacher are answered by the students and commented on by the teacher (Seedhouse, 2004, p. 102). One consequence of the turn-taking pattern is that the students can treat each other as experts and teachers, not only by answering each other, but also by asking questions or checking their own understanding of the teacher’s turn.
During activity (i), the students have problems keeping up with the teacher’s agenda. The problematic sequences resulting in code-switching and repair are heard only in the minidisc recordings. The teacher sets the pace for the activity by delivering his questions, but he does not listen to or take part in the desk talk. In order to do the institutional work, the students have to try to understand the teacher’s turns and keep up with the tempo set by him. The analyses in this article focus on the students’ desk talk taking place immediately after the teacher has asked a question.
The second activity (ii) studied in this article is a so-called AB task done in pairs, the aim of which is to construct an oral task where the students are supposed to pose questions to each other and answer them in the FL (Swedish). This activity starts 1 h and 13 min into the lesson and ends 7 min later. The contents of both questions and answers are related to a tape that the class has just been listening to and to a text in their textbook (Kajander, Määttä, Nordgren, Repo, & Saariaho, 2005, p. 82). The theme of the text is the true story of a Swedish young adult, Helena, who became a drug addict as a teenager and who now struggles to get started on a drug-free life. The questions in the task are given in L1 (Finnish) on a piece of paper and the students’ task is to translate or interpret them into Swedish and find the answers in the textbooks lying open in front of them. The students are also expected to alternate in asking and answering the questions according to the following sequence: Student A asks question 1 Student B answers question 1 and asks question 2. Student A answers question 2 and asks question 3.
Typical of the institutional type of activities (i) and (ii) is that the peer interaction is strictly organized, even though the criteria differ.
Repair sequences and code-switching in peer conversation
In this article, the focus is on sequences of other-initiated self-repair and candidate understandings produced in L1 during FL conversations. According to Macbeth (2004, p. 722), as the students engage in these activities to create mutual understanding and solve local problems in their conversation, the activities can be defined as repair rather than correction. Although the organization and use of repair in plenary talk in classrooms have been analysed in numerous studies, for example, by McHoul (1990) and Macbeth (2004), repair in peer talk has received less attention (cf. Ohta & Nakaone, 2004; Swain & Lapkin, 2000). The repair sequences in peer interaction are not part of the instructive talk, tied up with asymmetrical institutional roles with initiation–response–evaluation (IRE) sequences controlled by the teacher, but they still take place in an institutional task and reflect the students’ orientation to different dimensions of institutionality.
Repair organization, as one of the central practices in conversation with which speakers deal with local problems in speaking, understanding and hearing, is important for our understanding of code-switching during students’ interaction (Schegloff, 2000; Schegloff, Jefferson, & Sacks, 1977). With repair, the interactants can show that there is a problem or a breakdown in understanding. As long as intersubjectivity, or mutual understanding, is maintained, it is demonstrated by ‘the succession of appropriately conjoined turns’ (Kurhila, 2006, 19 f.; Seedhouse, 2004, p. 21). The interlocutors meet disruptions in interaction caused by some kind of problems by doing repair. Repair is sequentially organized, consisting of initiation and outcome (Schegloff et al., 1977, pp. 376–377), with self-initiated self-repair and other-initiated self-repair preferred to other-repair in everyday conversation.
In everyday conversations and institutional conversations outside the classroom, the source of the problem is often a previous turn (Kurhila, 2006; Schegloff, 1992; Schegloff et al., 1977). In my data, this is often the case, but the source can also be the Swedish text present in the context (activity (ii)). In the talk during activity (i), the problems occur during the teacher’s previous turn in Swedish; in activity (ii), they occur during the other student’s turn in Swedish. The language of the problem source is always the FL (Swedish).
The repair sequences in the data show a similar, coherent structure of two to three turns: in the minimal version, the repair is initiated by student A in L1 (Finnish) and followed by the self-repair of student B in L1 in the next turn, after which the interactants switch code back to Swedish, the institutional language in which they are doing the task. Sometimes, the sequence is ended by a third turn where student A accepts the repair in L1. The repair sequences show a preference for the same language talk, a trait typical of code-switching with a tendency to accommodate to the other interactant’s choice of code during highly cohesive turns (Auer, 1995, p. 130). After the side sequence, the students go on with their task, switching back to Swedish.
In my data, the repair sequences are produced as side sequences, which temporarily suspend the working with the task on hand, and the students proceed only after the problem has been solved satisfactorily (Jefferson, 1972). Couper-Kuhlen (1992, p. 345) points out that repair sequences are a special kind of side sequences that the speakers want to resolve as quickly as possible.
Code-switching in other-initiated self-repair of problems in understanding
In my data, problems of understanding or hearing are rarely initiated by the other speaker with unfocused problem markers consisting only of a question word, even though this is a common practice in everyday conversations. At least in English conversations, simple questions words (who, where) or unfocused problem markers such as sorry? what? are used frequently (Drew, 1997; Schegloff et al., 1977, pp. 367–368). On the other hand, initiations are more often made with longer, specific question clauses than with a single question word (examples 3 and 4). The formulation of repair initiations during a task probably depends on the institutional constraints placed on the interaction. During activity (ii), for example, the students pre-orient towards hearing a certain question they are supposed to answer.
However, some repairs are made by code-switching with unfocused question markers. During activity (ii), Krista and Anni are engaged in asking each other questions according to the instructions and answering them on the basis of the textbook or what they remember from the tape (example 1). In line 1, Krista asks Anni what kind of rules there are in the house where the main person of the story, Helena, lives. Observe that, although Krista asks her question in Swedish, she utters the noun säännöt (‘rules’) in Finnish. In the examples, utterances or words in Swedish are given in roman font, whereas Finnish is in italics.
(1) Ggr4250106
It is possible that the code-switching in line 1 makes Krista’s utterance more difficult to process. Anni’s repair initiation in line 2 consists not only of the Finnish question word mitä (‘what’), but also of two Finnish discourse particles nii(n) and ett(ä) (Hakulinen et al., 2004, p. 820). The particles are usually used to draw a conclusion about the previous turn, but nii also has an intensifying meaning (Hakulinen et al., 2004, p. 1106; Raevaara, 1996, p. 4). By starting her question with the particles, Anni emphasizes that she is unable to answer Krista’s question, leaving open the question whether the problem concerns understanding or hearing. In line 3, Krista self-repairs by giving a translation of line 1 in Finnish. The word order, however, represents idiomatic Finnish and is different from line 1. Krista’s repair is followed by Anni’s answer in Swedish (lines 4 and 5) to Krista’s questions in lines 1 and 3. Anni’s answer restores the institutional order of Swedish as the language of interaction after the code-switching to Finnish in the repair sequence.
As pointed out earlier, FL is usually used in interaction during tasks and L1 in repair sequences, but in the data there are some exceptions to this order. Cromdal (2005) showed that when writing a text in English, bilingual children doing tasks followed a strict division of labour between Swedish (L1) and English (L2). He also analysed examples where this established language order was suspended in dispreferred and oppositional actions (Cromdal, 2005, pp. 344–349). In my data, the reasons for not switching code seem to be related to competence in Swedish rather than to the actions of the speakers (cf. Auer, 1998, p. 8). Significantly, in the few cases where unfocused other-initiations of repair are made in Swedish without code-switching to Finnish, all the repair sequences are initiated with the single Swedish question word va (‘what’), as in example (2). Here, the boys Kari and Matti are engaged in the same activity as in example (1): (2) Ggr5250106
There is a lot of noise in the classroom during the group work, so Matti’s initiation of repair in Swedish (line 3) can be either because he does not hear Kari’s question or that he does not understand it. As in example (1), the repair is realized by repetition of the question, but in this case, the repair is made in Swedish. The code-switching during the turn initiating the repair seems to define the language used by the speaker doing the repair, the realization of the repair signifying the sequential closeness between the initiation and repair turns (Auer, 1995).
Language choices during repair seem, to a certain extent, to reflect the students’ linguistic competence. The other-initiations during the institutional activities are often realized in a more complex manner than with simple unfocused question words. Several other examples include both more specific question clauses and varied contextualization cues (examples 3 and 4). Competence as such cannot, however, be regarded as the only reason for the orderly use of L1 during repair, but rather as one dimension of the students’ orientation towards repair as a phenomenon of the L1 interaction, also used for solving problems during institutional FL talk.
In example (3), an example is shown of a typical repair sequence realized by code-switching into L1, where the repair initiation consists of a whole clause focusing on a problem of understanding.
(3) Ggr5250106
The long pause in line 2 shows that Matti has problems answering Kari’s question in line 1. Unfortunately, Matti’s and Kari’s dialogue cannot be seen on a video, but during the long pause, there is a lot of distracting noise in the classroom with pairs sitting on both sides of Matti and Kari asking the same questions. Matti’s question in Finnish in line 3, initiating a repair, reveals that he probably did not hear the question in line 1. Starting with the consecutive particle siis (‘so’), his formulation of the repair initiation reveals that the students are involved in an institutional task where they are supposed to ask questions and answer them. By referring to Kari’s utterance as the question with the definite article se, Matti shows that he is oriented towards answering a specific question on the agenda which he did not hear or understand. In example (3), as in example (1), the repair is also in Finnish (line 5). In example (1), the students’ code-switched back to FL (Swedish) after the problem was solved, but in example (3), Matti does not know the answer to the question and continues in Finnish in line 7 with an account of why he cannot accomplish the institutional task even though he understood Kari’s question.
Code-switching in candidate understandings
Problems in conversation can also be met in other ways than through repair, namely as checking whether one’s interpretation of the previous turn is correct. In my data, students sometimes code-switch to L1 and produce candidate understandings when confronted with a problem of understanding. Similar results have been obtained by Kurhila (2006, p. 154) in whose data, consisting of Finnish native–non-native institutional encounters outside classrooms, the non-native speakers seldom initiate repair with unfocused markers. Instead, they meet problems by offering candidate understandings. Kurhila (2006, p.155) defines a candidate understanding as a turn produced by a speaker ‘that explicitly articulates the recipient’s interpretation of the trouble turn’ (see also Heritage, 1984a, pp. 31–32). Based on her data, Kurhila (2006, pp. 157–161) came up with the following properties of candidate understandings: With candidate understanding, the speakers engage in the activity of checking their understanding of a prior turn, which they design to elicit a confirmation. Candidate understandings typically initiate a side sequence with a minimal structure of two, sometimes three, turns. The candidate understanding is followed by a confirmation, a preferred response, or a disconfirmation (Sorjonen, 1997, p. 112). The different types of candidate understandings in Kurhila’s (2006) data show several kinds of variation, depending, for example, on how the speakers formulate their tentativeness. Usually, these candidate understandings consist of a discourse particle, such as a new receipt token ai or aha (‘oh’) or a question and a paraphrase or elaboration of the problematic turn.
In example (4) from activity (ii), there is a long side sequence consisting of six lines (5–11) where Anni and Krista switch from Swedish to Finnish in order to deal with a problem of understanding each other. The problematic turn is in lines 1 and 2 where Krista poses a question in Swedish to which Anni is supposed to reply but cannot answer immediately. Anni’s answer to Krista’s question finally comes in line 15. The side sequence starts with a candidate understanding produced by Anni in line 5. The difference between the repair sequences discussed above and the sequence in example (4) is that Anni, in asking her questions, shows how she has understood the previous turn, but is not able to explicate the problem in her understanding. In lines 1–4 before the long side sequence, the girls solve the Swedish equivalent of the Finnish katkaisuhoito (‘detoxication’) through a word search, a self-initiated other-repair.
(4) Ggr4250106
Lines 5–11 show Anni’s difficulties, the crucial problem being how she understands the future tense ska gå (‘will go’) used by Krista in her question. In line 5, Anni presents a candidate understanding in Finnish of Krista’s question in Swedish. The structure of the turn in line 5 resembles the cases Kurhila (2006, p. 161) calls paraphrasing candidate understandings, which include a discourse particle and rephrasing of the previous turn. It is marked with the news receipt token ai ‘oh’ followed by reformulation of Krista’s question in Finnish (Hakulinen et al., 2004, p. 161). However, Anni uses the past tense meni ‘went’ instead of future when she repeats Krista’s question (ska gå ‘will go’). Krista does not realize the problem, but confirms Anni’s understanding in line 6 with nii ‘yes’. In line 8, Anni starts answering Krista’s question (lines 1 and 2) in Swedish but cuts off her answer. The aborted turn is followed by a long pause. By this time, it is apparent that Anni is still finding it problematic to answer Krista’s question. In line 12, Anni reformulates her candidate understanding but does not change the tense in her question. By using two Finnish discourse markers, the initial conclusive siis ‘y’mean’ and the final vai ‘or’ Anni upgrades the uncertainty shown in her earlier rephrasing. Vai ‘or’ is in Finnish used as a turn final marker to give the turn over to the following speaker (Hakulinen et al., 2004, p. 988). Anni also explicates the deictic adverb sinne ‘there’ in line 5 by explicitly asking whether Krista is asking why Helena went for detoxication. Anni’s turn in line 12 resembles structurally the uncertainty-marked candidate understandings found by Kurhila (2006, p. 195) in L2 conversations in Finnish with an utterance framed by an initial conclusive discourse marker and the final marker vai ‘or’. In line 13, Krista solves the problem and repairs Anni’s candidate understanding by reformulating her question in Finnish and changing the tense from past to future (on menossa ‘is going’). Krista starts her repair with the discourse marker tai ‘or’ commonly used to show an alternative solution (cf. Sorjonen & Laakso, 2005). We can see that Anni shows acceptance and understanding in a third turn by a long drawn a:: (‘oh’) already before Krista has finished her turn (line 14). Anni’s acceptance marks the end of the side sequence that has been dealing with the problem. Anni returns to the institutional work by switching to Swedish in her answer to Krista’s question in line 15.
The work done by Krista and Anni during the side sequence to solve their problem is not only realized by code-switching but also through several discourse particles functioning as contextualization cues. All the earlier examples of repair sequences done in L1 show several examples of discourse particles used to initiate, to frame or to end a turn in the repair sequences. Researchers of code-switching and repair (Auer, 1995; Cromdal, 2003, p. 756) have pointed out that, in order to accomplish their actions, speakers employ several contextualization cues together with code-switching. It has been shown that typologically different languages organize self-repair differently and use different types of lexical or non-lexical devices, for example, to initiate self-repair (cf. Fox, Hayashi, & Jasperson, 1996; Green-Vänttinen & Lehti-Eklund, 2007; Kärkkäinen, Sorjonen, & Helasvuo, 2007; Sorjonen & Laakso, 2005). First-year students in Scandinavian languages at university level do not use code-switching when making other-initiated repair sequences in Swedish interaction in classroom tasks, but they still sometimes initiate self-repair with certain Finnish particles instead of the typical non-lexical Swedish devices (Green-Vänttinen & Lehti-Eklund, 2007). Therefore, it is to be expected that the cues with which other types of repair can be initiated or candidate understanding can be shown form an equally complex and language-specific set of markers, for example, in the case of discourse particles.
Throughout the long and complex sequence when they are working together towards an understanding of Krista’s original question, both participants maintain a systematic order between the languages in which the sequences starting with the candidate understandings are in L1 (lines 5, 6, and 10–13), whereas the institutional work (lines 8, 14–18) is realized in Swedish.
CS in repair of problems produced by keeping up with the agenda
Many of the repair sequences in the data are associated with problems having to do with the agenda that have to be resolved before the work can continue. Joutseno (2007), who has studied problem sequences in plenary talk in a class doing tasks, points out that many of the students’ problems are procedural and are related to the instructions of or carrying out of activities. The problem sequences she has studied are realized as insertion sequences. In my data, the students orient to problems in following both the teacher’s agenda during activity (i) and the written guidance during activity (ii). In example (5) from activity (ii), Krista initiates a repair because she (line 3) has a problem with where the students are on the worksheet.
(5) Ggr4250106
After Anni has cut herself off in the middle of the question she is posing (line 1), Krista switches to Finnish and initiates a repair by asking Anni which question she is trying to formulate (line 3). In order to be able to answer each other’s questions, the students have to keep track of the numbers of the questions. In line 4, Anni starts her repair in Finnish by answering Krista with the first syllable of the word for number two kaksi in Finnish (kak-), but she cuts herself off and does a self-repair in the same turn. She marks the self-repair lexically with the most common lexical marker in Finnish eiku (literally no but) and solves the problem by giving the turn over to Krista (lines 5 and 6, cf. Sorjonen & Laakso, 2005). Krista starts a new question in Swedish according to her worksheet. As in previous examples, the other-initiated self-repair in lines 3 and 4 is done as a sequence in code-switching to L1. Only after the problem has been solved can the students continue with their work in Swedish.
An example of students having problems in following the teacher’s agenda is found in example (6). During activity (i), the teacher is asking the class to give the Swedish equivalent of the Finnish perfect participle kohonnut (‘risen’). The teacher’s turn in lines 2–5 is problematic for Liisa, who cannot find the Swedish equivalent to the verb form in the text (line 6). As pointed out in section ‘Two curricular activities’, the teacher’s turns are directed to the whole class during this activity, but the students talk to each other quietly.
(6) Ggr1250106
In line 2, the teacher gives the infinitive form of the verb raise in Swedish (höja), and in line 3, when the students are supposed to have answered, the correct answer höjt ‘risen’. In line 6, Liisa initiates as an overlap with the teacher’s plenary talk a repair in Finnish by asking Maija where the perfect participle form kohonnut occurs in the text in front of them. In her repair in line 7, Maija introduces the Swedish context, the nominal phrase höjt bloodtryck (risen ‘high’ blood pressure) with the Finnish deictic toi (‘that’) and probably points at the verb form in the text (this unfortunately cannot be seen on the recording). Maija utters her repair as an overlap with the teacher’s turn. In line 10, Liisa accepts Maija’s repair in Finnish. In a ‘normal’ IRF sequence in plenary talk where the students are supposed to answer the teacher’s questions in front of the whole class, Liisa probably would direct her repair initiation to the teacher. Because of the specific turn-taking order, Liisa gives Maija the role of expert and directs her repair initiation in line 6 to Maija, who accepts the role and makes the repair.
In activity (ii), all the repair sequences have been made as other-initiated self-repair. As Liisa in example (6) initiates the repair, she is, according to the instructions, not supposed to turn to the ‘self’ who has produced the problem, the teacher, but to Maija sitting next to her. Therefore, the repair sequence in example (6) is actually produced as other-initiated other-repair, where the two others are not represented by the same person. In desk talk that runs parallel to the teacher’s plenary talk, the students have the possibility to give each other the role of expert, one who can deal with problems caused by the teacher’s talk (cf. Sahlström, 1999).
Problems evoked by the texts in the surroundings
In section ‘Analysing interaction in the FL classrooms as institutional discourse’, the relevance of the texts for the interaction in FL classrooms was outlined. Texts function either as a source for conversations or as the outcome of conversations, or both (e.g. Cromdal, 2003, 2005; Nikula, 2005, p. 35). The interdependence of interaction and texts can be actualized in different ways. In the lesson studied here, the questions during activity (ii) are based on paper slips given to the students, who may use their textbooks as a source for their answers. In everyday conversations, other-initiated self-repair often is oriented to problems manifested in a previous turn (Schegloff, 1992, p. 1302; Schegloff et al., 1977, p. 367), whereas repair sequences in classrooms can also be oriented to problems raised by the texts used in the interaction. In example (7), Krista is posing a question in Swedish to Anni in lines 1–3, and Anni answers in lines 5 and 6. Instead of accepting the answer or giving Anni the following turn, Krista initiates a repair in line 7: Example (7) Ggr4250106
The girls are sitting with their textbooks open in front of them formulating their answers on the basis of the texts. Anni’s answer in lines 5 and 6 is almost a direct quote from the first of the following sentences in the textbook: När hon [Helena] var 14 blev hon kär i en kille som kunde fylla lite av det tomrummet. Han var missbrukare. (‘When she [Helena] was 14 she fell in love with a guy who could fill a little of the empty space. He was a drug addict’.) (Kajander et al., 2005, p. 82). Krista is probably checking the answer with the text in the book (unfortunately, this cannot be seen on the video recording). In the other groups, the students combine the two sentences in the text in their answers to the same question: she fell in love with a guy who was a drug addict (hon blev kär i en kille som var missbrukare). On the basis of the audio recording, Krista’s repair initiation in line 7 seems to be caused by the fact that she does not understand the Swedish noun missbrukare (‘drug addict’) in the textbook, which occurs in the text after the sentence used by Anni in her answer.
The repair sequence in lines 7–10 differs somewhat from the other examples in this article as it is only partly done as code-switching. In this extract, the text is both the source of and the solution to the problem, which has an impact on the use of code-switching in the repair. The repair (line 7) is initiated in L1 as in the other-repair sequences, and so is the completion at the end of line 9 where Anni code-switches into Finnish with a discourse particle indicating conclusion (siis, eng. ‘that is’) and giving the Finnish counterpart of missbrukare, namely, huumeidenkäyttäjä (‘drug addict’). Krista’s acceptance in line 10, however, is in Swedish. In between, there are also two Swedish turns, Krista’s in line 8 and the beginning of Anni’s answer in line 9, both direct quotes are from the textbook. In his analysis of students construing a written text, Cromdal (2005) has shown how one code, L2, is reserved for writing the text, whereas the interaction is in L1. Similarly, embedding a text written in the FL in an interaction can evoke code-switching. In example (7), the voice of the text is used in the repair done in L1 through a reference to the text in the FL. In Bakthinian terms, using other voices in discourse has been called double-voicing (cf. crossing in Rampton, 1998; Bakhtin, 1986, p. 110). The voice of the text influences the language choice in Krista’s acceptance in line 10.
The analysis of example (7) has to some extent revealed the reflexive relationship that exists between the texts and the interaction between the students in an FL classroom. Through reading, talking about texts and writing, the texts can have an impact on students’ interactions and vice versa. Repair initiated in conversation because of problems in interpreting texts shows that students interact not only with each other when doing tasks but also with the texts around them. Problems in an interaction, which is shaped in the multimodal surroundings consisting, for example, of texts, pictures on blackboard, videos, textbooks or worksheets, can be caused not only by previous turns, as in everyday conversation but also by texts that students in their interaction rely on (cf. Schegloff et al., 1977). When used in conversation, the language of the texts in FL can cause code-switching.
Concluding discussion
The analysis in this article has focused on code-switching to L1 in repair during different activities in an FL classroom. The main findings, the most important of which concern students’ understanding of the interaction in the FL classroom, are going to be discussed below. Linguistic resources and repair in a classroom and some implications for language learning will also be touched upon.
This analysis supports earlier findings, from different school contexts, of a general tendency among students to keep up a division of labour between FL used for institutional work and L1 as the language reserved for the interaction (Cromdal, 2005; Nikula, 2005). The systematic construction of repair sequences in L1 reflects the students’ interpretation of FL classroom interaction. At least in these data, the discursive practices of the classroom rely on and are shaped by the use of different texts. By code-switching to L1 when creating mutual understanding, irrespective of which kind of problems they are dealing with in their institutional interaction during Swedish lessons, the students can be said to show two aspects in their orientation. On the one hand, they seem to regard the strictly instructed classroom interaction as talk about texts, talk which in Gustavsson’s (1988) terms can to some extent be seen as taking a decontextualized FL as its object. On the other hand, the students seem to signal that repair is a central practice of the interaction between them, created both outside and inside classrooms in L1. In their socially constructed reality, problems in mutual understanding are not problems of classroom talk, but rather problems of interaction.
However, as shown in one of the examples, in repair sequences where problems in the texts are discussed, these repairs can be partly done in the FL, as the students recontextualize parts of the FL texts in the making of meaning. Repair caused not by previous turns in conversation, but by the surrounding texts used in interaction, can probably be found also in other kinds of institutional talk where documents play a central role (cf. Heritage, 1997, p. 161).
The examples illustrate that repair is a complex interactional phenomenon in which different kinds of contextualization cues are combined. When switching to L1, the students can use formulations, question patterns and discourse particles that enable them to come quickly and effectively to a mutual understanding and resume the institutional work (cf. Slotte-Lüttge, 2005). As repair is such an important practice in conversation, it is, however, vital to create opportunities in an FL classroom to master this necessary and seemingly difficult practice.
The institutional realities of an FL classroom, for example, the strong reliance on texts, seem to imply that learning central practices in interaction can be a problem. Since repair is often constructed in L1, the question arises whether repair practices in an FL can be taught. Research on how to repair in a certain language and which devices to use to initiate or complete a repair sequence has generally not found its way into textbooks on FLs. The analyses made above indicate that knowledge and explicit instruction of, for example, the practices used to be able to repair ought to be included in FL teaching (cf. Carroll, 2005, p. 211). Since doing repair seems to be a demanding activity in which the student ought to be able to command and combine the use of several kinds of contextualization cues, repairing in an FL could perhaps, through explicit instruction, be made into an institutional activity that can be learnt in the classroom.
Footnotes
Appendix 1
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.