Multilingual students’ use of their linguistic repertoires while writing in L2 English

+ Models

LINGUA-2684; No. of Pages 17

Available online at www.sciencedirect.com

ScienceDirect Lingua xxx (2018) xxx--xxx www.elsevier.com/locate/lingua

Multilingual students’ use of their linguistic repertoires while writing in L2 English Tina Gunnarsson Educational Sciences, Lund University, Box 117, 221 00 Lund, Sweden Received 2 August 2018; received in revised form 25 March 2019; accepted 25 March 2019

Abstract This article examines six multilingual students’ composing processes and language use while writing in L2 English. Four participants have two L1s: Swedish and either Bosnian (N = 2) or Macedonian (N = 2). The remaining two participants have Swedish as their L1. Building on an L2 composing model (Wang and Wen, 2002) and the theory of Language Mode (Grosjean, 2008), the study uses thinkaloud data to examine participants’ use of their language repertoires while writing an essay in L2 English. Think-aloud data revealed that Swedish served as the base language of thought for four of the participants. The other two participants used mainly English and for a greater range of composing activities. Bosnian, a heritage language for two of the participants, was used by one participant for contextspecific idea-generation. The results suggest that multilinguals rely on their L1s vis-a-vis the target language for specific purposes when working individually on a complex task such as L2 writing. An elaboration of the Wang and Wen (2002) model revealed that English was used mainly to read the text being produced, whereas Swedish was used to comment on the writing process, to solve problems and to have an inner dialogue. © 2019 Elsevier B.V. All rights reserved.

Keywords: Multilingualism; L2 writing; Language Mode; Think-aloud protocols

1. Introduction The role of previously learned languages in the processes of learning and teaching English as a non-native language is ´ Cain, 2009; Velasco a major current concern (DiCamilla and Anto´n, 2012; Hall and Cook, 2012; Turnbull and Dailey-O and García, 2014; Zhang, 2018). As of yet, there is no evidence to support better learning outcomes by employing what has become to be known as the English Only method, in which the teacher aims to strictly adhere to the target language while teaching (Macaro, 2009). However, research on translanguaging suggests that the incorporation of students’ entire language repertoires can increase student confidence and motivation in class (Garcìa and Wei, 2014). Previous research has shown the following positive effects of translanguaging in the classroom: (i) translanguaging facilitates comprehension and content may thus be processed more deeply (Baker, 2011; Creese and Blackledge, 2010; Cheng, 2013; Garcìa et al., 2012), (ii) students can more easily make themselves understood (Arthur and Martin, 2006; Lin and Martin, 2005), (iii) the integration of speakers of different proficiency levels in the target language can be more easily achieved (Baker, 2011), (iv) translanguaging increases student motivation (Lin, 1999) and (v) students become more prone to participate in class discussions (Källkvist, 2013a,b). Students of different ages have been shown to

E-mail address: [email protected]. https://doi.org/10.1016/j.lingua.2019.03.007 0024-3841/© 2019 Elsevier B.V. All rights reserved.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

2

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

translanguage, for example by employing their literacy skills in their L1 to support their writing in L2 (Edelsky, 1986; Lanauze and Snow, 1989). Also, the implementation of a multilingual pedagogy specifically for writing has been shown to facilitate the writing process (Lay, 1988) and also to improve the text (Cummins, 2006). Recent research reveals that the use of the L1 in L2 teaching often varies depending on student proficiency, student age, student needs (Tsagari and Diakou, 2015), and also teacher attitudes (Copland and Neokleous, 2010; McMillan and Rivers, 2011; Tsagari and Diakou, 2015). There also seem to be differences in the amount of L1 used in teachers’ L2 practice, with two studies showing teachers underreporting their use of the L1 (Copland and Neokleous, 2010; Tsagari and Diakou, 2015). Copland and Neokleous suggest that there is a certain stigma attached to the use of the L1 and ‘‘to do so would be to admit incompetence, and, perhaps more damningly, would challenge their personal philosophies of learning and teaching’’ (2010:278). Students, on the other hand, seem to hold a positive view of the incorporation of the L1, stating that it has a facilitative effect and increases confidence (Tsagari and Diakou, 2015). Numerous studies have shown that the majority of individuals naturally use their L1 as a language of thought (Cenoz and Gorter, 2011; Murphy and Roca de Larios, 2010; Tullock and Fernández-Villanueva, 2013; van Weijen et al., 2009; Wang and Wen, 2002), and others have shown beneficial effects of using the L1 as a resource when engaging in L2 writing tasks (DiCamilla and Anto´n, 2012; Velasco and García, 2014). Even though recent research has gone beyond bilinguals and focused also on L3 and L4 writing (Cenoz and Gorter, 2011; Tullock and Fernández-Villanueva, 2013), there is a notable lack of studies of school-age individuals with a migrant background, typically bilinguals of the majority language spoken in their new home country and a heritage language,1 usually spoken in the home rather than at school. Additionally, very little is known about the L2 writing process among L1-Swedish-speaking students and whether and how they employ Swedish when writing in an L2. These two types of students, those with L1 Swedish and those who are simultaneous bilinguals of Swedish and a heritage language, make up the existing student demographics in Swedish schools. It is important to represent students with both of these two language profiles as we can then begin to understand student groups as a whole. The present study was therefore designed to explore whether and for what purposes 15--16-year-old students in Sweden draw on their complete language repertoires, including the heritage language, when working individually on a writing task in English, their L2. The present article will divulge how the results of the study led to an elaboration of the existing Wang and Wen (2002) model and how the elaborated model makes finer distinctions for the text-generating activity, revealing a variety of language functions present while generating text. I begin by describing the context researched in the present study: English as a school subject in Sweden.

2. English in the Swedish school context and beyond English has been a mandatory subject since 1962, but was before then introduced as a required subject for the natural science track in 1845 (Hult, 2012). English has a high status in Swedish society, used to a great extent in elite professions and in academic settings (Berg et al., 2001). English is further used extensively in Swedish media, television, music, computer games, and on the internet. Research shows that many students engage in spare time activities involving the use of English (Henry, 2014; Sundqvist, 2009). Most children start compulsory school at the age of seven. English is introduced as a subject either in year 1, 2, 3 or 4 depending on decisions taken in the local municipality. Throughout compulsory school, students are entitled to 480 hours of English (Swedish National Agency for Education, 2018a). The current English curriculum was implemented in 2011 and is aligned with the The Common European Framework of Reference for Languages (CEFR), adopting its aim of the European citizen as a competent plurilingual language user (Hult, 2017). At the end of the ninth school year students should have achieved a B 1.1-level (independent user on threshold level) according to the CEFR-scale (Swedish National Agency for Education, 2017) provided they have obtained at least a pass grade (E). The Swedish National Agency for Education promotes English Only as the medium of instruction (Swedish National Agency for Education, 2011) and Swedish teachers of English tend to use more English in their instruction than do teachers of English in other European countries (Lundahl, 2012). While promoting English Only, The National Agency for Education has opened up for other languages to be used stating that ‘‘It requires professional judgement about what best helps a student approach the objectives and about whether there are situations when the use of a language other than English can occasionally aid a student's understanding or improve their knowledge development’’ (Swedish National

1 By heritage language I refer to the definition of Benmamoun et al. which states that ‘‘a heritage speaker is an early bilingual who grew up hearing (and speaking) the heritage language (L1) and the majority language (L2) either simultaneously or sequentially in early childhood" (Benmamoun et al., 2013:133).

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

3

Agency for Education, 2011:2 as cited in Hult, 2017). It is therefore up to the teachers themselves to decide when and how to use different languages while teaching English. A report released by the Swedish Schools Inspectorate in 2011 revealed that students with a migrant background, i.e. students born abroad or students born to parents of a non-Swedish origin, on average have lower grades in English than their Swedish-born peers. The Swedish Schools Inspectorate hypothesized that this result was due to teachers and school staff not taking the students’ background into account (2011). According to the National Agency for Education, these students comprised roughly 27.1% of the total student population for the school year 2017/2018 (Swedish National Agency for Education, 2018b). It has also been found that teachers of English tend to make comparisons to the Swedish language and culture when teaching English and that this presents a disadvantage to students of a non-Swedish background (Tholin, 2014).

3. Swedish and mother tongue instruction in Sweden According to Josephson (2004), Swedish and English have the same status in Swedish society, positioned at the top of the linguistic hierarchy. Swedish as a subject is introduced in year 1 and is allocated 1490 hours throughout compulsory school (Swedish National Agency for Education, 2018a). Mother tongue instruction is offered to all students who speak a language other than Swedish at home provided that the municipality is able to find a teacher of said language and provided there are at least five students in the group. The student can only be offered mother tongue instruction in one language, even if the student speaks more than one mother tongue, which is not Swedish, in the home. If the student speaks one of the five minority languages protected by the Language Act established in 2009, they have a right to mother tongue instruction regardless of the number of students in the group (SFS, 2009, paragraph 7). These languages are Finish, Yiddish, Meänkieli, Romani Chib or Sami. The Language Act further states that those with a mother tongue other than Swedish should be given opportunity to develop and use their first language (SFS, 2009, paragraph 14). According to law, mother tongue instruction can be offered as a subject between year 1 and year 9 for either a total of 600 hours, 320 hours or 382 hours (SFS, 2011:185, chapter 5, paragraph 8). A fourth alternative is to offer the subject outside the normal curriculum, in which case the principal is obligated to provide lessons during 7 years for an unspecified number of hours. Even then, the principal is required to continue lessons in the mother tongue if the student has a special need for such instruction (SFS, 2011:185, chapter 5 paragraph 11). Briefly, the minimum requirement in order for the students to pass the course is that they are able to read and understand texts in the mother tongue of different length and genres, are able to summarize what they have read, are able to write different types of texts that are understandable to the reader, are able to speak about and discuss a variety of subjects in a simple manner, asking questions, stating opinions, showing reasoning skills, comparing arguments and holding smaller structured talks that are adjusted to a specific audience (Swedish National Agency for Education, 2016). While mother tongue instruction is considered important, there is no pass requirement in order for the students to move on to upper secondary school. In his analysis of the languages represented in Swedish educational policy documents, Hult concludes that while Swedish and English have very high status in school and in the country as a whole, immigrant mother tongues remain ‘‘on the margin of Swedish society’’ (Hult, 2012:246). 4. L1 use in L2, L3 and L4 writing Previous research has shown that the L1 is used in different tasks and for different purposes in the L2 classroom (see Corcoll, 2011 for an overview). Such activities include the use of the L1, and previously learnt languages, in translation ´ Cain, 2005), for reading tasks exercises (Corcoll, 2011; Källkvist, 2013a,b), in oral communication (Liebscher and Dailey O to understand new vocabulary and complicated concepts (Bhooth et al., 2014; Smidfelt, 2015), in collaboration tasks where students need to interact in order to complete a task (DiCamilla and Anto´n, 2012; Zhang, 2018) and in L2 writing (Murphy and Roca de Larios, 2010; Plata-Ramírez, 2016; van Weijen et al., 2009; Wang and Wen, 2002). Studies focusing on L1 versus L2 use in L2 writing when thinking aloud have shown that the L1 is used to some extent regardless of proficiency or text type (Murphy and Roca de Larios, 2010; Tullock and Fernández-Villanueva, 2013; Wang, 2003; Wang and Wen, 2002). Previous studies have found that while some participants used the L1 extensively when thinking aloud, other participants employed the L1 considerably less (Murphy and Roca de Larios, 2010; Wang and Wen, 2002), sometimes even as little as 0.5% of the entire think-aloud protocol (henceforth TAP) (Tullock and FernándezVillanueva, 2013). Students use the L1 to generate ideas or pretext for the content and to plan, organize and evaluate or revise the text being produced (Tullock and Fernández-Villanueva, 2013; Wang and Wen, 2002). For the 20 first-year English majors in Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

4

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

van Weijen et al.’s (2009) study, the L1 (Dutch) was used specifically when engaging in self-instruction (on average 45%) and metacomments2 (on average 43%) but less so for structuring (19%) and generating ideas (14%). Studies have further shown that the complete language repertoire is employed for lexical searches (Cenoz and Gorter, 2011; Jessner, 2006; Tullock and Fernández-Villanueva, 2013) and that this holds true regardless of whether the participants are writing in an L2, L3 or L4 (Cenoz and Gorter, 2011; Jessner, 2006; Tullock and Fernández-Villanueva, 2013). Tullock and Fernández-Villanueva (2013) found that the number of lexical searches made was relative to proficiency, i.e. the more proficient the participant was in English the fewer lexical searches were made. In the study by Wang and Wen (2002), 16 Chinese university students of English (aged 18--22) at four different levels in Chinese higher education (4 freshmen, 4 sophomores, 4 juniors and 4 seniors who each had 8 years of classroom exposure to English prior to entering university) used both languages (L1 Chinese and L2 English) in 31 out of 32 TAPs (2 TAPs per participant). Results showed that the participants tended to use more L1 as they were generating ideas for content and for organizing these ideas as well as controlling their writing process (e.g. controlling time and word limit etc.), while the L2 was used mainly to examine the task (i.e. the writing prompt, which was in the L2) and when generating text in English, which had to be in the L2. Studies of L3 and L4 writing revealed that the multilingual's languages are activated simultaneously (Jessner, 2006) for different purposes such as lexical retrieval, evaluating the text and applying grammatical structures from the L2 or L1 (Cenoz and Gorter, 2011) and that switching languages appeared to be done rather effortlessly (Tullock and FernándezVillanueva, 2013). Tullock and Fernández-Villanueva (2013) found that the majority of their participants (8 out of 10) resorted to thinking aloud in their L1 (either Spanish, German or Catalan) while writing in their L4 (English), but that all ten participants also used the school language, German (which was either the L1, L2 or L3 for the participants), when generating ideas for the content. Tullock and Fernández-Villanueva suggest that daily contact with a language may be a good indication as to whether or not the specific language is activated during the writing process in English. To summarize, previous research has shown that the L1 is used when students write in L2, L3 and L4. There are several purposes for using the previously learnt languages, such as lexical gaps and text evaluation. These results seem to hold irrespective of whether the L1 and L2 (L3, etc.) are typologically close or not (e.g. L1 Dutch and L2 English in van Weijen et al., 2009; L1 Chinese and L2 English in Wang and Wen, 2002). Although several of the studies reviewed above have used think-aloud data (Murphy and Roca de Larios, 2010; Tullock and Fernández-Villanueva, 2013; van Weijen et al., 2009; Wang and Wen, 2002), the focus of the studies has differed and the manner in which the think-aloud data has been analyzed has therefore also differed. While van Weijen et al. (2009) counted each segment3 containing L1 use and based their coding on the Flower and Hayes model (1981), Wang and Wen (2002) counted the number of words uttered in each language. In these studies, L2 proficiency was found to be positively linked to either the text quality (van Weijen et al., 2009) or the amount of L2 use during the TAP (Wang and Wen, 2002). On the other hand, such a relationship was not found in the studies by Murphy and Roca de Larios (2010) and by Tullock and Fernández-Villanueva (2013). In these studies, the TAPs were first coded for planning, formulation and revision. Secondly, the segments that included planning and revision were taken out, and the authors proceeded to count the lexical searches made in the formulation segments. The conflicting results may thus be attributed to the authors using different coding schemes for the segmentation of their TAPs, while focusing on different parts of the writing process, L1 and L2 use versus lexical searches. Furthermore, all the studies mentioned above except one (Cenoz and Gorter, 2011) used participants that were older than 15--16, attending upper secondary school (Tullock and Fernández-Villanueva, 2013), university (van Weijen et al., 2009; Wang and Wen, 2002) or post-graduate education (Murphy and Roca de Larios, 2010) and writing was constrained either by a word or time limit. In the present study, no such constraints are posed; the participants are younger (15 years); four have acquired two languages simultaneously from birth, and all were attending their final year of compulsory school. Results will be analyzed using the Wang and Wen (2002) model and the theory of Language Mode, to which I now turn. 5. Language Mode Language Mode refers to ‘‘the state of activation of the bilingual's languages and language processing mechanisms at a given point in time’’ (Grosjean, 2008:39). According to Grosjean, bilingual individuals position themselves on a Language Mode continuum ranging from a monolingual mode to a bilingual mode depending on the interlocutor and the purpose of communication. When communicating, a multilingual individual activates one language as the base language,

2 Metacomments are referred to as ‘‘[r]eflections on the writing process as a whole or comments on the assignment and sources’’ (van Weijen et al., 2009:240). 3 By segment I refer to van Weijen et al.’s study which states that "If a writer switched from one activity to another (e.g., from Planning to Generating ideas), then the segment containing the first activity ended and the new activity was placed in the next segment" (2009:240).

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

5

which is ‘‘the main language being produced or perceived at a particular point in time’’ and ‘‘the language that governs language processing’’ (Grosjean, 2008:40), while other languages fall into the periphery. A background language cannot be fully deactivated but remains easily accessible in case our communicative needs shift and thus require its use. Language Mode can thus explain why a specific language is employed in a specific context. While engaged in a communicative task, bilinguals find themselves in three different modes depending on the comparative activation of the two languages employed: a monolingual mode, where the main language employed is the base language and the second language (henceforth the guest language) is activated to a lesser extent; in an intermediate mode, employing one language as base language and activating the guest language more; in a bilingual mode activating both the base language and the guest language to an almost equal extent. However, according to Grosjean (2008), one language will always be activated more for processing and this language remains the base language unless the bilingual decides to switch base language, which; depending on topic and communicative situation, is a possibility. The theory of Language Mode can be applied to subjects employing more than two languages in communication, who will similarly find themselves in three different modes ranging from a monolingual to a trilingual (or quadrilingual, etc.) mode. This applies to the present study, in which four participants are regular users of three languages. The language employed mainly in school is the majority language, Swedish, whereas the language in the home of the participants with two L1s tends to be the heritage language (either Bosnian or Macedonian). The language in the English classrooms in Sweden is usually either English or Swedish, as the latter is frequently used for contrastive purposes (Swedish Schools Inspectorate, 2011; Tholin, 2014). However, for the participants in this study all essays were expected to contain English only as they were to be graded by an English teacher. In what follows, the Wang and Wen (2002) model, which was used to distinguish and code the different writing activities, will be described. 6. A model of the L2 writing process The Wang and Wen model of the L2 writing process builds on the Flower and Hayes model of L1 writing (1981), which was modified as it was considered too linear (Wang and Wen, 2002). It is precisely the recursiveness of the Wang and Wen model that made it a suitable tool for the analysis of the data for this study, as the participants similarly went back and forth between the different writing activities during the writing process. Wang and Wen (2002) distinguished the following five composing activities in their think-aloud data from Chinese university students who were writing two essays while thinking aloud in the language(s) of their choice: (1) task-examining, (2) idea-generating, (3) idea-organizing, (4) processcontrolling and (5) text-generating. Task-examining refers to the stage at which the participant processes the writing prompt for the task. Idea-generating pertains to the conceptualization of ideas for the content; idea-organizing to the order in which to use these ideas, while process-controlling refers to structuring the text, i.e. paragraphing, punctuation and word limit etc. Finally, text-generating refers to the stage where the actual writing commences. The results in the Wang and Wen (2002) study showed the task-examining and text-generating activities to be ‘L2 dominant’, which means that more L2 words (English) than L1 words (Chinese) were used by participants when thinking aloud. The remaining three activities, idea-generating, idea-organizing and process-controlling, were ‘L1 dominant’, meaning that a greater number of L1 words than L2 words were used. Among the five writing activities, text-generating was the predominant activity engaged in by the participants, ranging between 63.5 and 68.5% of the think-aloud data collected. The present study builds on the Wang and Wen L2 composing model while using younger participants and different language pairs. Although the Wang and Wen (2002) study was conducted in a different context (a university in China), the model is suitable for analysis of L2 writing in a Swedish context. The Wang and Wen (2002) study further represents an opportunity to replicate a study focusing on students’ L1 use in L2 writing, an aspect of L2 writing that has not been studied among school-age students who have two L1s. 7. The study 7.1. Aim and research questions The aim of this study is to find out the extent to which, and for what purposes, six multilingual 15--16-year-old students use their L1(s) and their L2 as languages of thought while writing an essay in their L2 (English). To reach this aim the following research questions are addressed: (1) Which of their languages do the six participants use when thinking aloud while writing a narrative in English under exam-like conditions? (2) Are the languages used for specific composing activities identified by previous research? If so, which languages and for which composing strategies? Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

6

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

7.2. Method 7.2.1. Participants Participants were six volunteers -- all students in school year 9. They were approached and asked to take part in the study on the basis of their language background. Four have a migrant background and are simultaneous bilinguals of a heritage language (two of Bosnian and two of Macedonian) and Swedish, whereas two have a Swedish background and are native speakers of Swedish. They were drawn from a larger sample of participants used to initiate a larger project (Gunnarsson, 2015). Each participant wrote one narrative essay on a topic from the national test in English for year-9 students. The essay was written under think-aloud conditions, where the participants were asked to think aloud using languages of their own choice. For the purpose of anonymity, the participants have been given fictitious names. The first letter of their names corresponds with the first letter of one of their L1s. Belma and Benjamin are simultaneous bilinguals of Bosnian and Swedish; Maja and Marko are simultaneous bilinguals of Macedonian and Swedish; Sara and Sofie are the Swedishspeaking participants. All six participants started school at the age of 7. All had instruction in Swedish as a subject from their first year and instruction in English as a subject from their second year. Four of the participants in this study had mother tongue instruction twice a week for 40 minutes per lesson starting in their first school year. Table 1 profiles the participants in terms of gender, language repertoire, grades in English, mother tongue and Swedish, and grades awarded on the essay written (on a scale from A to F, where A is the top grade). 7.2.2. Data The think aloud method was used as it is the only method that can capture real-time data (Cohen, 1996). As the thinkaloud method has been questioned (Smagorinsky, 1998), a pilot study was initially carried out with a year-9 student -- a native speaker of Italian. As this student showed no signs of difficulties in understanding or using the method while composing an essay, the think-aloud method was adopted for this study. As is customary in think-aloud studies (Wang and Wen, 2002), a short trial session was held individually prior to the first TAP. In this session, the method was first explained to the participant and he/she then moved on to try the method by thinking aloud while formulating a text about his/her hobbies. Special care was taken to stop them if they started to explain their thoughts, as this would become an extra burden during the actual writing process (Ericsson and Simon, 2010). Once the participants were familiar with thinking aloud, they chose a signal (in all six cases a popping sound) to serve as a reminder whenever there was a longer pause in the TAP. Care was taken to create a non-threatening environment, where each participant was seated in a small room in their school alone with paper, pen and essay prompt. The author remained

Table 1 The participants’ language profiles, grade on essay, grade in English, grade in Swedish, and for four participants the grade in the subject of the mother tongue. Participant

Gender

Languages

Grade on essay

Grade in the subject of English

Grade in the subject of Swedish

Grade in the subject of the mother tongue

Belma

Female

E

C

D

A

Benjamin

Male

E+

A

B

A

Maja

Female

D+

A

C

A

Marko

Male

C+

A

C

B

Sara

Female

C

B

B

-

Sofie

Female

Bosnian -- from birth Swedish -- from birth English -- from age 8 Bosnian -- from birth Swedish -- from birth English -- from age 8 Macedonian -- from birth Swedish -- from birth English -- from age 8 Macedonian -- from birth Swedish -- from birth English -- from age 8 Swedish -- from birth English -- from age 8 Swedish -- from birth English -- from age 8

E+

C

B

-

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

7

on the other side of a wall, ready to help or use the reminder signal if needed. Before each participant started writing their first essay they were able to ask questions if any part of the task was unclear. They then proceeded by writing the first essay while verbalizing their thoughts in the languages of their choice. There was no need to use the reminder signal, as all participants were able to think aloud while writing without longer pauses.

7.2.3. Writing task The writing task was the essay part of the national test for English from 2009 entitled Crossroads. The prompt is intended to elicit narrative text relating to personal experiences. The prompt begins with a description of the topic, which reads ‘‘Sometimes you have to choose direction and make up your mind about which way to go or what to do. Write about a choice that you have made recently or that you will be making in the near future or the years to come’’. It goes on to describe different possibilities in a bullet list for what to write about stating ‘‘It could be about what to study and work with, where to live, family and friends, spare time activities, politics, religion, the environment’’. After this bullet list the instruction poses three questions to think about while writing. They are: ‘‘What/who inspires or influences you? What alternatives are there? What are the consequences?’’. Finally, the writer is instructed to ‘‘Describe and explain to make your reader understand!’’. Three of the participants (Benjamin, Marko and Sofie) chose to write about ‘‘what to study and work with’’. The remaining three chose to write about different topics. Belma chose to write about her family, Maja about where she wanted to live in the future and Sara about a choice she had made recently regarding her friends. Since the task involved thinking aloud as well as writing the essay, participants were permitted to take as much time as needed (the normal time limit is 80 minutes). Two English teachers with several years experience from marking and grading the national test, graded the essays independently of each other (see Table 1). The grading criteria that accompany the national test of English were used. The grading scale ranges from A to F, giving equal weight to content and language use. Whenever there was a lack of agreement as to the grade, the essay in question was discussed using the grading criteria for support until agreement was reached.

7.3. Analysis Each think-aloud/writing session was audio recorded using a small Dictaphone. The data was then transcribed verbatim following the transcription conventions of Wang and Wen (2002). Segmentation was carried out by listening closely to the recordings and coding the writing activities on the basis of the Wang and Wen model. Once identified, segments were coded from top to bottom of the transcriptions for each of Wang and Wen's (2002) five composing activities and for language (Bosnian, Macedonian, English, Swedish). To establish intercoder-reliability, an experienced linguist and myself independently coded the same two TAPs, yielding agreement for 89% of the segments. The 11%, which were coded differently, were discussed until agreement was reached. I then proceeded to code the remaining data. While other studies have calculated switches between languages (Wang, 2003), or number of segments containing L1 (van Weijen et al., 2009), this study calculated the number of words to concur with Wang and Wen's (2002) L2 composing model. Once all TAPs had been coded, the total number of words in each language and in each composing activity was calculated. The results of this analysis will be the focus of the next section. 8. Results 8.1. Research question 1: Which of their languages do the participants use while thinking aloud? Fig. 1 shows how much of the TAP was produced in the heritage L1 (i.e. Bosnian or Macedonian), in the majority L1 (Swedish) and in English L2 in terms of percentages of the total number of words produced during the think-aloud session. As shown, the use of English ranges from 31.4% (Sofie) to 99.4% (Marko) and the use of Swedish ranges from 0.6% (Marko) to 68.6% (Sofie), while the heritage L1 (Bosnian) is used by one of the four heritage-language speakers (Belma) to the extent of 8.5%. As can be seen in Fig. 1, in terms of the sheer number of words, five participants used more English than Swedish while thinking aloud, whereas one (Sofie) used more Swedish. Sofie was the participant who used the least amount of English in her TAP (31.4%) while the other Swedish-English bilingual, Sara, used English to the extent of 80%. Sara, instead, used Swedish as her language of thought for only 20%, making the TAPs of the two L1-Swedish-speaking participants almost mirror images of each other. This reveals two ‘high’ users of English: Marko (a simultaneous bilingual of Macedonian and Swedish) and Sara (whose L1 is Swedish), two ‘intermediate’ users of English (Benjamin and Maja), one user of three languages (Belma) and one ‘low’ user of English (Sofie). Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

8

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

Fig. 1. Percentages of words in different languages used for thinking in TAP.

8.2. Research question 2: Are the languages used for specific composing activities identified by previous research? Previous research has shown that use of a specific language may be linked to a specific composing activity (van Weijen et al., 2009; Wang and Wen, 2002). Therefore, as a preliminary, Fig. 2 illustrates the extent to which the participants engaged in the five different writing activities while thinking aloud. As shown in Fig. 2, all participants mainly engaged in text-generating. Once again, Marko used the greatest proportion of words for this activity (85%), while Benjamin used the least with 58% of his words being dedicated to text-generating. The activity of free speech was added as a sixth activity and was engaged in only by Maja.4 Figs. 3--5 show the languages employed in terms of percentages of words used by each participant for three of the composing activities (task-examining, idea-generating and text-generating). The two remaining composing activities are not reported using figures as the participants engaged very little in them (0--1% for idea-organizing and 0--5% for processcontrolling). Marko and Benjamin did not engage in idea-organizing and the activity represents less than 1% of the entire TAPs of the remaining four participants. The languages employed for this activity are Bosnian for Belma, Swedish for Sofie and Maja and mostly Swedish for Sara. When they do engage in this activity, it is to elaborate on the order of their ideas as in nu ändrar jag lite ordning (‘now I’m changing the order a little’) by Sofie. Four out of the six participants engaged in process-controlling. While Marko and Belma did not engage in processcontrolling, the rest used between 2% (Sara) and 5% (Benjamin) out of their protocol to control processes. Processcontrolling was done in Swedish to make comments on practical matters, such as rewriting things on a clean sheet of paper. Results for task-examining are provided in Fig. 3, showing the percentages of the entire TAP, thereby illustrating the proportion of each activity in relation to the whole TAP. Whereas Swedish is used to comment on the task, the English words verbalized for this activity mainly consist of participants reading the writing prompt. Only Marko and Sara, the ‘high’ English users, verbalize in English using their own words rather than reading. Benjamin engages in task-examining the most of all the participants (21% of the total number of words in the TAP). While reading the prompt in English he comments on what he has left to do in Swedish and sometimes transitions into idea-generating, also in Swedish. This is exemplified below, where Benjamin starts by reading part of the prompt in English (Where to live) and then responds by generating an idea in Swedish. Benjamin's exact utterance is provided (a mixture of Swedish and English) with my translation underneath. (1) Where to live. . . jag har inte tänkt så mycket var jag ska bo och så hära men tror jag flyttar nånstans. . . Where to live. . . ‘I haven’t given it much thought where I am going to live and such but I think I will move somewhere. . .’

4

Free speech involves activities that were not related to the writing task, such as singing or Maja stating that she was hungry.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

9

Fig. 2. Percentages of composing activities during TAP.

Fig. 3. Percentages of words used in different languages for thinking when task-examining.

The data exemplify the different writing activities for which Swedish was used (to comment on the task using the participant's own words) and English (to decode/read the prompt, i.e. someone else's words). The English used here may thus be considered ‘reading’ aloud rather than ‘thinking’ aloud. Fig. 4 provides results for the idea-generating activity, showing that five of the six participants use mainly their L1(s). The sixth participant, Marko, used Swedish only to the extent of 0.2%, consisting of the comment jag vet inte (‘I don’t know’). Marko is the ‘high’ English user and, unlike the other participants, he verbalizes his own ideas in English when thinking aloud. A typical example of Marko idea-generating in English is illustrated in (2):

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

10

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

Fig. 4. Percentages of words used in different languages for thinking while idea-generating.

Fig. 5. Percentages of words used for thinking in different languages for text-generating.

(2) I’m planning on maybe starting my own business. . . the business could be about uhm. . . maybe. . . I will say cars or something. . . cars. . . and parts. . . and designing them. . . Sofie, who uses English the least, generates ideas entirely in Swedish, as can be seen in (3): (3) en konsekvens med linjen jag valt. . . vad kan det vara. . . jag kanske inte trivs. . . ‘a consequence of the program I have chosen. . . what could that be. . . I might not feel at home. . .’ Benjamin and Maja used more Swedish (16 and 18.5% respectively of the total number of words) than English when generating ideas, making this activity exclusively Swedish for Benjamin and almost so for Maja (with only 0.3% being spoken in English). Sara, on the other hand, generates ideas either entirely in Swedish or entirely in English. In her TAP she has 5 segments that display idea-generating. While two of the segments are entirely in Swedish, three are entirely in English. Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

11

Fig. 6. An elaboration of the text-generating activity.

Fig. 5 provides the results for text-generating, showing five out of six participants using a greater proportion of words in English than in Swedish. For Marko (the high English user), this is an activity carried out entirely in English apart from one word (nej meaning ‘no’), while the other participants all employ Swedish more. However, the text-generating activity in the TAPs showed the participants code-switching while engaging in writing activities that are not distinguished by the Wang and Wen (2002) model. On the basis of this data, I created a more elaborated model for the text-generating activity. Fig. 6 provides a breakdown of the different sub-activities engaged in by the six participants while text-generating. In Fig. 6, Wang and Wen's (2002) writing activity ‘text-generating’ is divided into three sub-activities: decoding, encoding and problem-solving. This allowed a more nuanced description of the text-generating activity. ‘Decoding’ refers to participants reading out loud what they have written. ‘Encoding’ breaks down into ‘writing’ (participants are simultaneously verbalizing what they are writing on paper) versus ‘formulating’ (participants saying out loud what they are about to write). ‘Formulating’ is further broken down into ‘back-translating’ (participants translate word for word, often back and forth between Swedish and English) and ‘rehearsing’ (different words or phrases are tried out before the participants choose which to write down). Finally, ‘problem-solving’ refers to moments when they make metacomments on their text, solve lexical problems and use metamarkers. The metacomments concern the stylistic or grammatical features of the emerging text and refer to participants verbalizing about their spelling, word-choice and whether or not a word is translated correctly from Swedish. When using a metamarker, the participant signals a need for something to be altered or doublechecked by using words such as vänta (‘wait’) or nej (‘no’). This is in line with the study by Nyroos, Sandlund and Sundqvist (Nyroos et al., 2017) in which the Swedish conjunction eller (’or’) is used to initiate self-repair in paired oral proficiency tests. Lexical gaps is another issue attended to when problem-solving, in which the participants specifically state that they do not know a word in English. Although the elaborated model of the text-generating activity in Fig. 6 reveals different sub-activities, it should be noted that these sometimes overlap, for instance ‘rehearsing’ as a means to fill a lexical gap. However, the order in which the different activities are engaged in is beyond the scope of this study and should be the focus of future research. Further, the elaboration of the text-generating activity only reveals the sub-activities present in the essays of the six participants who took part in this particular study. The elaborated model may therefore be expanded to allow for additional sub-activities that were not observed in these data, such as syntactical and morphological errors. Table 2 shows the elaborated model of the text-generating activity divided into encoding, decoding and problemsolving sub-activities explained above, including the frequency, i.e. the number of segments in the TAP in which the participants engaged in back-translating, rehearsing, metacommenting and using metamarkers as well as the number of lexical gaps the participants explicitly stated they had during the TAP. The elaborated model reveals frequent use of Swedish when metacommenting and using metamarkers. Swedish is also the language used to handle lexical gaps,5 which is present in two out of the six TAPs. A typical example of a lexical gap and the use of metamarkers can be seen in (4), where Maja uses a meta-marker (vänta lite meaning ‘wait a little’) when realizing that she doesn’t know how to say ‘the nutcracker’ in English.

5 It should be noted that only lexical gaps that were explicitly stated as such are included and that more lexical gaps may be included in the substantial amount of back-translating that was done by the participants.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

12

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

Table 2 Text-generating in the think-aloud protocol.

(4) when I. . . for the first time. . . saw. . . vänta lite nötknäpparen hur säger man det på engelska when I. . . for the first time. . . saw. . . ‘wait a little the nutcracker how do you say that in English’ 85% of the metacomments are in Swedish, the rest in English. This can be seen in the example by Sara below, where she starts by reading part of a phrase she has produced (my best friend), writes the word loves and then realizes that this does not work. (5) my best friend. . . uhm. . . loves. . . nej. . . så kan jag inte skriva. . . my best friend. . . uhm. . . loves. . . ‘no. . . I can’t write that. . . These results show that English is used to formulate text, to read the participant's own text and for rehearsing. An example of rehearsing can be seen below, where Marko starts by reading back a few words that he has written (‘‘and I can’’) and then goes on to rehearse in order to select the most appropriate wording to follow. (6) and I can achieve. . . I can. . . I can uhm. . . make it. . . I can uhm. . . I can succeed Table 2 reveals certain individual differences, such as the number of words spent reading by Maja (954 words) and Sara (734 words) compared to the other four participants (ranging from 40 to 237 words). Back-translations are present in all TAPs except for Marko's. Instead, he engages in rehearsing in English to a greater extent than the others. Rehearsing is not engaged in by Belma and Sofie, the two participants with the lowest grades on their essays. Being the ‘high’ Swedish user, Sofie employs Swedish more when formulating her text. She is the participant who employs back-translating the most with 22% (196 words) of her entire TAP being dedicated to this sub-activity in Swedish alone. More than half (53%) of her text in English is therefore generated through the use of Swedish. 9. Discussion and concluding remarks All five composing activities distinguished in the Wang and Wen model were discernible in the TAPs. The results show that certain activities were tied to using specific languages, i.e. Swedish, English or Bosnian. The findings are consistent with Wang and Wen's (2002) in that the activity of task-examining was L2 dominant. The transcription data revealed that this is due mainly to the participants reading the prompt in English. Although not many words were spent Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

13

process-controlling and idea-organizing, these activities were mainly carried out in Swedish (the L1), which is consistent with Wang and Wen's results. Idea-generating was L1 dominant for four of the participants, whereas one (Sara) generated ideas either in the L1 (Swedish) or the L2 (English) and one (Marko) solely in the L2 (English). Two of the participants in this study (Sara and Marko) therefore differ from the participants of Wang and Wen, as they are not L1 dominant for idea-generating. Finally, the text-generating activity, which is shown to be L2 dominant in the model of Wang and Wen, can also be said to be L2 dominant in this study if one looks at the activity as a whole. However, as the elaborated model of the text-generating activity (Fig. 6) shows, it is Swedish, which is the majority language and school language, that is used to metacomment or solve problems in the text for all participants except for Marko. This concurs with the results of van Weijen et al. (2009), who found the L1 used specifically to metacomment and self-instruct. It is through this decomposition of text-generating that we can finally see why this activity is L2 dominant, namely because the participants spend a large part of their thinking aloud reading back what they have previously written. One participant in particular, Marko, stands out as being a ‘high’ user of English with only 0.6% of the total number of words in the TAP spoken in one of his L1s, namely Swedish. Sara uses Swedish for 20% of the protocol, while Maja and Benjamin employ Swedish more (33.3 and 47%, respectively). Sofie is at the other end of the spectrum, as a high user of Swedish, with 68.6% of her protocol consisting of Swedish. Similar individual differences have been found both in the study by Wang and Wen (2002) and by Murphy and Roca de Larios (2010), where one participant never used the L1, two used the L1 sporadically and two were frequent users of the L1. The elaborated model (Fig. 6) reveals the majority language and school language Swedish to be used to double check and comment on their writing through metacommenting, to formulate sentences through back-translating and also, for two of the participants, to solve lexical gaps in English. The use of the L1 for lexical gaps when writing has been found in previous studies (Cenoz and Gorter, 2011; Jessner, 2006; Murphy and Roca de Larios, 2010; Tullock and FernándezVillanueva, 2013; Wang, 2003). The present study goes beyond previous research by revealing the extent to which a heritage L1 was used. The heritage L1 (Bosnian) serves two purposes for one out of four participants (Belma): to idea-generate and to help her formulate sentences before writing them down on paper when text-generating. Bosnian is used for specific contexts linked to memories obtained through Bosnian, which is consistent with the findings of Friedlander (1990) and Lay (1988). The two participants who idea-generate using English (Marko and Sara) have the highest grades on the two essays (C + and C, respectively), suggesting that they are more proficient in English than the other four. This finding agrees with previous research: the higher the L2 proficiency, the more L2 is used as a language of thought (Wang and Wen, 2002). However, it should be noted that the link between L2 proficiency and L2 use was not found in the studies of Murphy and Roca de Larios (2010) or Tullock and Fernández-Villanueva (2013). Then again, these studies were not focused on the amount of L1 or L2 use but on lexical searches, which were counted after the formulation segments in the TAPs had been isolated. In Grosjean's theory of Language Mode, the language referred to as the base language ‘‘governs language processing’’ (2008:40). The participants in the present study activate and use English when reading the prompt (which was in English) and when writing their text (which has to be in English). When engaging in these activities, there is a nonpresent interlocutor who requires them to use English (the writer of the essay prompt, and the teacher who will read and grade their essay). Four of the six participants (Benjamin, Belma, Maja and Sofie) use Swedish as the base language whenever possible to comment on their writing, to signal when something needs correction or confirmation or to solve lexical problems. As long as the essay has not been submitted, they can have a dialogue and reason with themselves for which their L1s are possible base languages. Drawing on Language Mode, these participants can be said to be in an intermediate mode where the base language that governs all the processing is Swedish, and the guest language, the language they use when reading the prompt and formulating their text, is English. Marko, the ‘high’ English user, on the other hand, uses English as the base language throughout his TAP as he verbalizes only a few single words in Swedish. In Grosjean's (2008) terminology, he is in a monolingual ‘English’ mode, barely using the guest language, Swedish, at all. Sara switches base language as she idea-generates sometimes in English and sometimes in Swedish. Sara can therefore be said to be in a bilingual mode, activating both languages to an almost equal extent and changing base language depending on the writing activity she is engaging in. The higher grades given to Marko's and Sara's essays by the teachers may reflect their higher proficiency levels compared to the other four, who use Swedish as the base language. Language Mode (2008, p. 63) can explain this as: ‘‘bilinguals who are highly dominant in one language may simply not be able to control language mode in the same way as less dominant or balanced bilinguals’’ and ‘‘[the weaker language] will simply not be developed enough or active enough to allow them to stay in a monolingual mode’’ (Grosjean 2008:63). This is further confirmed by Kobayashi and Rinnert who claim that ‘‘L2 knowledge mainly remains separate from L1 knowledge in the early stages of acquisition of writing knowledge’’ (2012:126) and ‘‘when writers acquire more experience, some L1/L2 writing knowledge is combined into a single, merged component of their repertoire, available regardless of the language being used, while some knowledge appears to remain associated with one specific language’’ Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

14

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

(2012:129). Kobayashi and Rinnert drew on the perspective of multicompetence (Cook, 1992) to propose that the languages we know overlap or merge in terms of how they are stored in our minds and the degree to which the languages overlap when writing increases with increased levels of literacy in the different languages (2012). The reason behind Sofie's substantial use of Swedish may therefore be that she is less proficient in English, whereas Marko and Sara are more proficient in English, enabling them to use English as the base language. Finally, Belma, the only participant to use both her L1s as well as English, finds herself in a trilingual mode throughout the think-aloud task. According to Grosjean (2008:46), ‘‘the base language usually governs the language production process (it is the ‘host’ or ‘matrix’ language) and hence it is used much more than the other’’. While writing she uses either Swedish or Bosnian as base languages to govern her process and produces text in English, which in this context becomes a guest language as it does not govern the process. Belma never uses English to structure or comment on the task. For most of her TAP she uses Swedish for the aforementioned activities, but when thinking of experiences gained through the use of Bosnian she begins to verbalize her thoughts using Bosnian. In Grosjean's (2008) words, the L1 that is not used as a base language is left in the periphery ready to be used if her communicative needs shift, which they do throughout her writing process as she thinks back on her memories from Bosnia. The present study has shown that the heritage language was used by only one out of four participants and was mainly used to generate context-specific ideas through memories gained through the use of Bosnian. Five out of six participants rely heavily on Swedish, which in this particular setting is an L1 as well as the majority language and the school language. This finding agrees with the study by Tullock and Fernández-Villanueva (2013) whose participants all used the school language regardless of their L1s. This study shows that students aged 15--16 naturally draw on their entire language repertoire while writing a narrative in L2 English. A heritage L1 was drawn on only to a limited extent, however. The results suggest that students may benefit by being encouraged to use their entire language repertoires when drafting and revising an essay in an L2. Acknowledgements I gratefully acknowledge financial support from the Swedish Research Council and the Crafoord Foundation. The Swedish Research Council and the Crafoord Foundation did not take part in the design of the study, the collection of data, the analysis and interpretation of the data, in the writing of this report or the decision to submit this manuscript. To the teachers that helped me asses and grade the essays, I would like to extend a warm thank you. To the six students who agreed to write six essays each in their spare-time, I am ever so grateful. I would like to express my very great appreciation to my supervisors Marie Källkvist and Alex Housen. I especially like to thank Marie Källkvist for her tireless and neverending support in finishing this manuscript, for taking me under her wing and for encouraging me to pursue an academic career. Appendix A This section includes additional examples of the different composing activities taken from the TAPs. Benjamin task-examining at the end of his TAP, making sure he has followed the instructions can be seen in (7). The text in italics displays Benjamin reading part of the instruction. As in the main text above, Benjamin's exact utterance is provided with my translation underneath. (7) what who inspires or influence you


ja det var ju att jag ville ju bli polis


eller ja också att det var en bred linje


what who inspires of influence you. . . yes well it was the fact that I wanted to become a police officer. . . or yes that it was a broad program too. . . Similarly, Sara can be seen task-examining at the end of her TAP in (8) below, also double checking that she has followed the instructions. (8) and what are the consequences


uhm that you get picked on


by the others


and I have that


ok


An example of Sofie idea-generating in Swedish can be seen in (9): (9) jag har ju även funderat på att bli typ domare


det verkar vara väldigt kul


jag tror jag skulle gilla det


Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

15

I’ve also thought about becoming a judge. . . it seems like a lot of fun. . . I think I would like that. . . Maja idea-generating in Swedish can be seen in (10): (10) när börja jag


när ville jag flytta härifrån


ja jag gillar Sverige visst men


jag vill inte leva här resten av mitt liv


jag var typ


hur gammal var jag


when did I start. . . when did I start to want to move away from here. . . yes I like Sweden sure but. . . I don’t want to live here for the rest of my life. . . I was like. . . how old was I. . . Benjamin solving a lexical gap can be seen in (11): (11) hur säger man ungefär nu


what really give


that could kind of


kind of


now how do you say kind of. . . what really give. . . that could kind of. . . kind of. . . In example (12) below, Sara can be seen text-generating in the middle of her TAP. In this example she is verbalizing what to write as she is writing, including short pauses in order to keep up with her hand movements on the paper. (12) uhm


now


when


when


no one


says


any


anything


about it anymore


anymore


I anymore


I say


that


it was worth


worth it


I can say that it was worth it


ok


In the next example (13), Sara is once again text-generating, however now she is at the end of her TAP and is reading through her text. The pauses are therefore no longer present as she is reading rather than writing. To concur with the Wang and Wen (2002) transcription conventions, Sara's utterance is displayed within quotation marks to show that she is engaged in reading her previously written text. (13) ‘‘I recently started to hang out with new friends


I did not have much in common with my old ones in fact I felt like I did not fit in the group my best friend from kindergarten loves to hang out with them so my choice made me leave her with them’’


Example (14) shows Belma text-generating through the use of backtranslations. (14) ibland kollar jag på min mamma


sometimes I look


at


my. . . mother


ibland kollar jag på min mamma


uhm och tänker


uhm


and


think


vilken underbar mamma


what a


wonderful


mother


I have


I am a proud uhm


daughter


sometimes I look at my mother. . . sometimes I look at my mother. . . sometimes I look at my mother. . . uhm and think. . . uhm. . . and. . . think. . . what a wonderful mother. . . what a. . . wonderful. . . mother. . . I have. . . I am a proud uhm. . . daughter. . . In example (15), Maja can be seen rehearsing while text-generating, trying to decide on the best way to express her intended meaning. (15) ok all of the movies


were


were placed


all of the movies


uhm


all of the movies


placed


all of the movies happened


happened in


ugh


all of the movies


were placed in

References Arthur, J., Martin, P., 2006. Accomplishing lessons in post-colonial classrooms: comparative perspectives from Botswana and Brunei Darussalam. Comp. Educ. 42, 177--202. Baker, C., 2011. Foundations of Bilingual Education and Bilingualism. Multilingual Matters, Clevedon.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

16

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

Benmamoun, E., Montrul, S., Polinsky, M., 2013. Heritage languages and their speakers: opportunities and challenges for linguistics. Theoretical Linguistics 39 (3--4), 129--181. Berg, E., Hult, F., King, K., 2001. Shaping the climate for language shift? English in Sweden's elite domains. World Englishes 20 (3), 305--319. Bhooth, A., Azman, H., Ismail, K., 2014. The role of the L1 as a scaffolding tool in the EFL reading classroom. Proc. Soc. Behav. Sci. 118, 76--84. Cenoz, J., Gorter, D., 2011. Focus on multilingualism: a study of trilingual writing. Mod. Lang. J. (3), 356--369. Cheng, T.-P., 2013. Codeswitching and participant orientations in a Chinese as a foreign language classroom. Mod. Lang. J. 97 (4), 869--886. Cohen, A., 1996. Verbal reports as a source of insight into second language learner strategies. Appl. Lang. Learn. 7 (1), 5--24. Cook, V., 1992. Evidence for multi-competence. Lang. Learn. 42 (4), 557--591. Copland, F., Neokleous, G., 2010. L1 to teach L2: complexities and contraditions. ELT J. 65 (3), 270--280. Corcoll, C., 2011. Developing children's language awareness: switching codes in the language classroom. Int. J. Multiling. 10 (1), 27--45. Creese, A., Blackledge, A., 2010. Translanguaging in the bilingual classroom: a pedagogy for learning and teaching? Mod. Lang. J. 94 (1), 103-115. Cummins, J., 2006. Identity texts: the imaginative construction of self through multiliteracies pedagogy. In: García, O., Skutnabb-Kangas, T., TorresGuzmán, M. (Eds.), Imagining Multilingual Schools: Languages in Education and Glocalization. Multilingual Matters, Clevedon, pp. 51-68. DiCamilla, F.J., Anto´n, M., 2012. Functions of L1 in the collaborative interaction of beginning and advanced second language learners. Int. J. Appl. Linguist. (22), 160--188. Edelsky, C.K., 1986. Writing in a Bilingual Program. Había una vez. Ablex Publishing, Norwood, NJ. Ericsson, K.A., Simon, H.A., 2010. How to study thinking in everyday life: contrasting think-aloud protocols with descriptions and explanations of thinking. Mind Cult. Act. 5 (3), 178--186. Flower, L., Hayes, J.R., 1981. A cognitive process theory of writing. Coll. Compos. Commun. (32), 365--387. Friedlander, A., 1990. Composing in English: effects of a first language on writing in English as a second language. In: Kroll, B. (Ed.), Second Language Writing: Research Insights for the Classroom. Cambridge University Press, Cambridge, pp. 109--125. Garcìa, O., Wei, L., 2014. Translanguaging. Language, Bilingualism and Education. Palgrave, London. Garcìa, O., Homonoff Woodley, H., Flores, N., Chu, H., 2012. Latino emergent bilingual youth in high schools: transcaring strategies for academic success. Urban Educ. 48, 798--827. Grosjean, F., 2008. Studying Bilinguals. Oxford University Press, Oxford. Gunnarsson, T., 2015. Multilingual Students’ Writing in English: The Role of Their L1(s). Centre for Languages and Literature, Lund University. Hall, G., Cook, G., 2012. Own-language use in language teaching and learning. Lang. Teach. (45), 271--308. Henry, A., 2014. Swedish students’ beliefs about learning English in and outside of school. In: Lasagabaster, D., Doiz, A., Sierra, J.-M. (Eds.), Motivation and Foreign Language Learning: From Theory to Practice. John Benjamins, Amsterdam, pp. 93--116. Hult, F., 2012. English as a transcultural language in Swedish policy and practice. TESOL Quart. 46 (2), 230--257. Hult, F., 2017. More than a lingua franca: functions of English in a globalised educational language policy. Lang. Cult. Curric. 30 (3), 265--282. Jessner, U., 2006. Linguistic Awareness in Multilinguals: English as a Third Language. Edinburgh University Press Ltd., Croydon. Josephson, O., 2004. ‘‘Ju’’: Ifrågasatta självklarheter om svenskan, engelskan, och alla andra språk i Sverige (‘‘Ju’’: Questioning Assumptions About Swedish, English, and All Other Languages in Sweden). Norstedts Ordbok, Stockholm, Sweden. Källkvist, M., 2013a. Languaging in translation tasks used in a university setting: particular potential for student agency? Mod. Lang. J. 97 (1), 217-238. Källkvist, M., 2013b. The engaging nature of translation: a nexus analysis of student--teacher interaction. In: Tsagari, D., Flores, G. (Eds.), Translation in Language Teaching and Assessment. Cambridge Scholars, Newcastle upon Tyne, pp. 115--134. Kobayashi, H., Rinnert, C., 2012. Understanding L2 writing development from a multicompetence perspective: dynamic repertoires of knowledge and text construction. In: Mancho´n, R. (Ed.), L2 Writing Development: Multiple Perspectives. De Gruyter Mouton, Berlin, pp. 101--134. Lanauze, M., Snow, C.E., 1989. The relation between first and second language writing skills: evidence from Puerto Rican elementary school children in bilingual programs. Linguist. Educ. 1, 323--339. Lay, N., 1988. The comforts of the first language in learning to write. Kaleidoscope 4, 15--18. ´ Cain, J., 2005. Learner code-switching in the content-based foreign language classroom. Mod. Lang. J. 89 (2), 234--247. Liebscher, G., Dailey O Lin, A.M.Y., 1999. Doing-English-lessons in the reproduction or transformation of social worlds? TESOL Quart. 33, 393--412. Lin, A.M.Y., Martin, P. (Eds.), 2005. Decolonisation, Globalisation: Language in Education Policy and Practice. Multilingual Matters, Clevedon, UK. Lundahl, B., 2012. Engelsk språkdidaktik: Texter, kommunikation, språkutveckling (English Language Education: Texts, Communication, Language Development). Studentlitteratur, Lund. ´ Cain, J. Macaro, E., 2009. Teacher use of codeswitching in the second language classroom: exploring ‘optimal’ use. In: Turnbull, M., Dailey O (Eds.), First Language Use in Second and Foreign Language Learning. Multilingual Matters, Bristol, pp. 35--49. McMillan, B., Rivers, D., 2011. The practice of policy: teacher attitudes towards ‘‘English only’’. System 39, 251--263. Murphy, L., Roca de Larios, J., 2010. Searching for words: one strategic use of the mother tongue by advanced Spanish EFL writers. J. Second Lang. Writ. (19), 61--81. Nyroos, L., Sandlund, E., Sundqvist, P., 2017. Code-switched repair initiation: the case of Swedish eller in L2 English test interaction. J. Pragmat. (120), 1--16. Plata-Ramírez, J.M., 2016. Language switching: exploring writers’ perceptions on the use of their L1s in the L2 writing process. Revista Internacional de Lenguas Extranjeras 5, 47--77. SFS, 2009:600. The Language Act in English. Available at: https://www.regeringen.se/49bb9d/contentassets/ 9e56b0c78cb5447b968a29dd14a68358/spraklag-pa-engelska (retrieved 22.11.18). SFS, 2011:185. Skolförordningen (Elementary and secondary reform). Available at: https://www.riksdagen.se/sv/dokument-lagar/dokument/ svensk-forfattningssamling/skolforordning-2011185_sfs-2011-185 (retrieved 05.11.18). Smagorinsky, P., 1998. Thinking and speech and protocol analysis. Mind Cult. Act. 5, 157--177.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007

+ Models

LINGUA-2684; No. of Pages 17

T. Gunnarsson / Lingua xxx (2018) xxx--xxx

17

Smidfelt, L., 2015. Il processo delle inferenze lessicali in italiano L3: Il ruolo delle lingue apprese in precedenza e altre strategie di comprensione. Études Romanes de Lund 100. Lund University, Lund. Sundqvist, P., 2009. Extramural English Matters: Out-of-school English and Its Impact on Swedish Ninth Graders’ Oral Proficiency and Vocabulary. Karlstad University Studies, Karlstad. Swedish National Agency for Education, 2011. Alla kommentarer. (All commentary). Available at: https://www.skolverket.se/download/18. 6011fe501629fd150a28916/1536831518394/Kommentarmaterial_gymnasieskolan_engelska.pdf (retrieved 17.10.18) Swedish National Agency for Education, 2016. Kommentarmaterial till kursplanen i modersmål utom nationella minoritetsspråk. (Commentary to the mother tongue other than the national minority languages curriculum). Available at: https://www.skolverket.se/sitevision/proxy/ publikationer/svid12_5dfee44715d35a5cdfa2899/55935574/wtpub/ws/skolbok/wpubext/trycksak/Blob/pdf2591.pdf?k=2591 (retrieved 05.11.18) Swedish National Agency for Education, 2017. Kommentarmaterial till kursplanen i moderna språk. (Commentary to the modern languages curriculum). Available at: https://www.skolverket.se/sitevision/proxy/publikationer/svid12_5dfee44715d35a5cdfa2899/55935574/wtpub/ws/ skolbok/wpubext/trycksak/Blob/pdf3866.pdf?k=3866 (retrieved 01.11.18) Swedish National Agency for Education, 2018a. Timplan för grundskolan. Available at: https://www.skolverket.se/undervisning/grundskolan/ laroplan-och-kursplaner-for-grundskolan/timplan-for-grundskolan (retrieved 09.10.18) Swedish National Agency for Education, 2018b. Grundskolan-Elever-Riksnivå. Sveriges Officiella Statistik. Available at: https://www.skolverket. se/skolutveckling/statistik/sok-statistik-om-forskola-skola-och-vuxenutbildning (retrieved 12.10.18) Swedish Schools Inspectorate, 2011. Engelska i grundskolans årskurser 6-9. Rapport 2011:7. Skolinspektionen, Stockholm. Tholin, J., 2014. ‘‘Swedishness’’ as a norm for learners of English in Swedish Schools: a study of national and local objectives and criteria in compulsory schools. Scand. J. Educ. Res. 58 (3), 253--268. Tsagari, D., Diakou, C., 2015. Students’ and teachers’ attitudes towards the use of the first language in the EFL state school classrooms. Res. Pap. Lang. Teach. Learn. 6 (1), 86--108. Tullock, B., Fernández-Villanueva, M., 2013. The role of previously learned languages in the thought processes of multilingual writers at the Deutsche Schule Barcelona. Res. Teach. Engl. (47), 420--441. ´ Cain, J., 2009. First Language Use in Second and Foreign Language Learning. Multilingual Matters, Bristol. Turnbull, M., Dailey-O van Weijen, D., van den Bergh, H., Rijlaarsdam, G., Sanders, T., 2009. L1 use during L2 writing: an empirical study of a complex phenomenon. J. Second Lang. Writ. (18), 235--250. Velasco, P., García, O., 2014. Translanguaging and the writing of bilingual learners. Biling. Res. J. (3), 6--23. Wang, L., 2003. Switching to first language among writers with differing second language proficiency. J. Second Lang. Writ. (12), 347--375. Wang, W., Wen, Q., 2002. L1 use in the L2 composing process: an exploratory study of 16 Chinese EFL writers. J. Second Lang. Writ. (11), 225-246. Zhang, M., 2018. Collaborative writing in the EFL classroom: the effects of L1 and L2 use. System (76), 1--12.

Please cite this article in press as: Gunnarsson, T., Multilingual students’ use of their linguistic repertoires while writing in L2 English. Lingua (2019), https://doi.org/10.1016/j.lingua.2019.03.007