Research on three L2 writing conditions: Students’ perceptions and use of background languages when writing in English

Abstract This paper gives an account of Norwegian upper secondary school students’ self-reported use of linguistic resources while composing a text in English (L2) under three different writing conditions, i.e. English-only, translation, and translanguaging. After writing a text in English, 200 students answered a questionnaire about their use of background languages as well as their perceptions of the assigned writing condition. A combination of quantitative and qualitative approaches to the analysis of the questionnaire data was employed to capture how the students use their background languages and what they consider to be relative advantages and disadvantages of the assigned writing condition. The results indicate: (i) a strong presence of English as a metacognitive language of choice in all three writing conditions, (ii) students’ strategic use of L1 to generate ideas and structure information, and (iii) students’ willingness to experiment with languages to enhance certain aspects of their writing. By integrating translation and translanguaging into the drafting stage of writing in a target language, the present study contributes to the empirical research that embraces bi- and multilingual approach to English writing instruction in modern language classrooms as they become more linguistically and culturally diverse.


PUBLIC INTEREST STATEMENT
This paper examines Norwegian upper secondary school students' language use while composing a text in English (L2) under three different writing conditions: English-only, translation, and translanguaging. Two hundred students answered a questionnaire about their use of background languages and their perceptions of the assigned writing condition. A combination of quantitative and qualitative approaches to the analysis of the questionnaire data was employed to capture how the students use their background languages and what they consider to be relative advantages and disadvantages of the assigned writing condition. The results indicate a strong presence of English in all three writing conditions, students' use of L1 to organize ideas, and students' willingness to experiment with languages to enhance certain aspects of writing. The present study contributes to the empirical research that embraces a multilingual approach to English writing instruction in modern language classrooms as they become more linguistically and culturally diverse.

Introduction
This study is placed in Norway and examines Norwegian upper secondary school students' perceptions of three writing conditions as well as their use of background languages when composing essays in English, their L2. 1 Norwegians have a generally high level of English proficiency, which is reflected in EF English Proficiency Index (Education First, 2019) that ranks the country third out of 100 countries and regions around the world. However, Norwegian users of English do not perform well in all areas. Research in Norway focusing on the English writing and reading skills of tertiary level students described Norwegian students' literacy skills as inadequate. Hellekjaer (2009Hellekjaer ( , 2010) provides evidence that students have difficulty in reading academic literature, following lectures, and writing academic texts. Furthermore, several studies of both L1 and English writing skills of Norwegian lower and upper secondary school students (Berge et al., 2005;Hundahl, 2010;Nygaard, 2010) show that students experience problems with creating coherence and structure in their texts. The fact that Norwegian students struggle with written English is further supported by English exam results in Norwegian lower secondary school from the previous five years, which bear out that students score higher on oral exams than on written (The Norwegian Directorate for Education and Training, 2019). This might be an indicator of Norwegian learners' need for more proactive and innovative teaching and learning practices targeting writing skills in English specifically, since literacy lags behind oral and listening skills. The current investigation of the students' language use and perceptions may offer some new insights with regard to assisting Norwegian learners in developing their literacy skills in English in close proximity to other languages they know.
The questionnaire data for the present investigation were collected during an experiment that involved 200 first-year upper secondary school students (age 15-16) from two mainstream schools in Northern Norway. The students were divided randomly (per class) into three groups. Each group was assigned one of three writing conditions: English only, translation from L1 into English, and translanguaging (García, 2009(García, , 2012García & Wei, 2014;Lewis et al., 2012;Williams, 1996). The students received the same task, that is, to write a fantasy narrative (see Appendix B: Writing Prompts). The English-only group wrote their texts directly in English. The translation group wrote their drafts in Norwegian and then translated them into English. The translanguaging group could choose any language or a mix of languages to write a draft, which they then used to write a final essay in English.
In the context of this study, translanguaging is understood in its classroom application, where languages are used in a dynamic and functional manner in order to further literacy in all languages involved (Lewis et al., 2012). In the translanguaging condition in our experiment we prompted the students to mix languages in their drafts before writing the final essay in English. This is consistent with the weak version of translanguaging (García & Lin, 2017;Turnbull, 2019;Williams, 1996), which pertains to the softening of the borders between the languages to tap into the students' background knowledge and linguistic repertoire, while acknowledging the existence of a target language (and the goal of developing it) as opposed to other available languages. By contrast, the strong version of translanguaging calls for the elimination of the social construct of "named languages" and argues for providing the students with the opportunity to " […] fully use their entire language repertoire, without regard to the socially and politically defined boundaries of named languages […]" (García & Lin, 2017, p. 10).
As explained earlier, in the course of the experiment, we collected questionnaire data on students' language use employed in the English-only, translation, and translanguaging writing conditions. Participants in each group also provided their feedback on the assigned writing condition. We adopted a combination of quantitative and qualitative approaches to the analysis of the questionnaire data to provide an account of the students' language use and feedback. In this study we aim to gain insights into the students' writing process, specifically, the way they utilize their linguistic resources at the drafting and final stages of text production under three different writing conditions. This study seeks to answer the following research questions: (1) What language(s) is/are employed during the drafting and final stages of the writing process in different conditions?
(2) What are the students' perceptions of the English-only, translation, and translanguaging writing conditions?

Use of background languages in the L2 writing process
In Wang and Wen (2002) study of L1 use in the L2 composing process of 16 Chinese EFL writers, the authors conclude that "[…] the L2 writing process is a bilingual event: L2 writers have two languages (i.e., L1 and L2) at their disposal when they are composing in L2" (p. 239). The analysis of the think-aloud protocols showed that the L1 accounted for around 30% of the data and was used predominantly in process-controlling, idea-generating, and idea-organizing activities. Another important conclusion that the authors draw from their analysis is that the proportion of L1 usage decreases as students engage in the types of activities that are closely related to textual output, while the processes leading to text-generating are L1 dominant.
Similar findings are reported in several studies that focused on the interaction of languages in the mind of L2 writers (Manchón et al., 2007;Woodall, 2002). The upshot of research on the role of L1 in L2 writing is that purposeful use of L1 occurs in different stages of the composing process and appears to be an integral part of composing in L2 (see Van Weijen et al., 2009). However, L2 writers employ this strategy to achieve different goals as their L2 proficiency develops. More proficient writers resort to their L1 when handling tasks of higher complexity (Manchón et al., 2009;Van Weijen et al., 2009), whereas the amount of L1 use to solve linguistic problems correlates negatively with the increase in the L2 proficiency level. The only exception is linked to the use of L1 for monitoring function, that is, when L1 facilitates the writer's ability to cope with the cognitive overload and working-memory constraints, and manage the writing process by means of self-instruction and meta-comments (Cohen & Brooks-Carson, 2001;Cumming, 1989;Woodall, 2002). In this case, writers tend to utilize their L1 in a similar way and to a similar degree regardless of the level of L2 proficiency.
Studies that target multilingual writers' use of background languages are particularly relevant to our research since in our study we incorporate translanguaging as a multilingual approach to English writing instruction. Cenoz & Gorter (2011) report that the participants in their study, 165 Basque-Spanish bilinguals and secondary school students, activated both of their background languages while writing in English. The quantitative and qualitative analysis of the writing samples indicated that the students engaged in multidirectional cross-linguistic transfer and employed similar general writing strategies across the languages. Another study of multilingual students' writing strategies (Tullock & Fernández-Villanueva, 2013) supports that multilinguals, in this case 16 and 17-year-old Spanish-Catalan-German trilinguals composing essays in English, employed all three background languages to solve lexical problems. However, the participants' L1 was activated to a greater degree than other languages.
A study of multilinguals' use of background languages in the Swedish context (Gunnarsson et al., 2015) includes participants of a similar age group (15-16) as our own study. The participants were divided into three groups according to their L1, i.e. Swedish L1, simultaneous L1s group (exposed to Swedish and another L1 from early age), and Other L1 group (with L1 other than Swedish). The authors analyzed the survey data on participants' language background and their use of different languages as they composed an essay in English (L2 or L3). The students reported Swedish (their L1 or L2) as the most frequently employed language of thought; while English is reported to be activated considerably more often once the students turn to text-generating activities.
The evidence reviewed here supports claims put forward by the proponents of the translanguaging framework (García, 2009(García, , 2012Lewis et al., 2012;Williams, 1996) in that users of two or more languages rely on all of their linguistic repertoire by choosing flexibly and strategically from an integrated system of linguistic resources (Gort, 2006;Kibler, 2010). To be specific, Velasco and García (2014) examined the way young bilingual writers utilized translanguaging as a writing strategy in the planning, drafting, and production stages. The qualitative analysis of the five writing samples produced by the K-4 th grade Spanish-English and Korean-English bilingual students revealed that the young writers used translanguaging to organize ideas related to the topic and to engage the reader. In addition, translanguaging was employed as a vocabulary learning strategy (text annotations) and a discourse feature (inner speech). In a recent study by Turnbull (2019), the author investigated the effect of weak and strong forms of translanguaging on the production of academic and creative texts by 60 first-year Japanese EFL university students. The results indicate that engaging in the strong version of translanguaging at the planning stage of writing in English allowed the students to produce more concise and well-formed essays.

Learners' perceptions of translation and direct composition
Studies on students' perceptions of different writing conditions are scant and, to our best knowledge, focus exclusively on direct composition versus translation. For instance, Cohen and Brooks-Carson (2001) looked into the students' self-reported use of writing strategies in direct and translated essays. The participants were 39 university-level students who wrote two essays in French (their L2 or L3) using translation and direct composition. With regard to the students' perceptions of the writing modes, even though the direct writing mode received more positive feedback, the students reported translation to be helpful for generating ideas and organizing their texts. Further, Ismail and Alsheikh (2012) and Tavakoli et al. (2014) adopted the research design and the instruments from Cohen and Brooks-Carson (2001) for their investigations of direct versus translated writing. Together, the analysis in these two studies showed that the participants thought that writing in English was faster and that it helped them find English expressions. The students also reported that translation took more time, but was advantageous in terms of generating ideas and organizing the content.
In sum, previous research clearly indicates that all L2 writers make use of their L1 and other background languages if they are available. In order to examine language use of the students when they are encouraged to draw on their entire linguistic repertoire, we prompted our participants in the translanguaging group to employ translanguaging at a draft stage of writing an essay in English. It follows then that the current investigation of the students' language use and perceptions extends beyond research on direct versus translated writing and offers an opportunity to explore the interplay of languages across three different (at least on surface level) writing conditions.

Participants
For this investigation, we recruited 200 participants who were 15-16 years old at the time of the experiment. In advance of the data collection, the project was approved by the Norwegian Centre for Research Data. Due to the age of the participants, the written voluntary informed consent was obtained from the participants before the data collection process started. The questionnaire data were elicited by means of two online surveys. All personal information was anonymized.
The participants were first-year upper secondary school students. Since most Norwegians enter the education system at the age of six, and English education begins in the first year, most of our students had had ten years of English instruction by the time of the experiment. However, some students may have had some of their schooling outside of Norway. Norwegian schools are obliged to provide adapted language education to students who cannot follow regular curriculum. To the best of our knowledge, none of the students attended any language support classes outside of the regular curriculum. Therefore, it is reasonable to assume that the participants with a foreign background 2 had achieved the required level of proficiency in Norwegian and English during data collection. The data on language background and gender were elicited through the online questionnaire (see Appendix A). Table 1 is a summary of the collected responses to the online questionnaire:

Writing sessions
In total, 15 English classes were recruited for the experiment. Each class was randomly assigned to one of three writing conditions, namely, English only, translation, and translanguaging. The students were given the same task: to write a fantasy narrative essay. The translation group was asked to write a text in Norwegian (or another dominant language) and then translate it into English. The English-only group did all their writing in English. The translanguaging group could choose to use any language they wished (or a mix of languages) to write a draft which they then used to produce a text in English. All of the students could use up to 90 minutes in total, and all writing had to occur in class either on the same day or during the next English class in the same week. The task did not count to their final grade. The questionnaire data were collected shortly after the writing sessions. The same experienced university lecturer introduced the activity to all three groups to ensure consistency of the instructions as well as to reduce individual teacher impact. The language form of the writing prompts (see Appendix B) and the given instruction differed slightly from group to group, i.e. English only, Norwegian only, and a mix of English and Norwegian respectively. Through manipulation of the languages in the prompts and instruction, we attempted to apply Grosjean's (2008) concept of language mode. We tried to put the students in the English only group in a monolingual mode with English being the base language. In the participants in the translanguaging group we tried to induce a bilingual mode where both Norwegian and English would be activated to some degree. The participants in the translation group were supposed to start writing in the monolingual mode with Norwegian as the base language and then switch to English monolingual mode when it was time to translate. We acknowledge a major issue with this strategy: giving the participants instructions in one language (or a mix of languages) does have a potential to activate a specific language or languages as a base, but "does not guarantee a particular position in the monolingual-bilingual mode continuum" (Grosjean, 2008, p. 58). Although it is impossible to guarantee the students would use a specific language mode in a non-laboratory setting, the students in all groups responded to the prompts and instructions as anticipated. That is, the students used English, Norwegian, or a combination of languages in their drafts and responded to the survey questions about their language use according to their assigned writing condition.

Questionnaire
Two hundred students provided the answers to the online questionnaire "Use of languages and feedback on writing conditions" (see Appendix C). These answers constitute the data for the present investigation. Parts of the questionnaire were adopted from Cohen & Brooks-Carson's study (Cohen & Brooks-Carson, 2001) of direct versus translated writing. The subscale for the translanguaging group was designed specifically for the present study and contained the items eliciting the students' language use during various stages of the writing process, such as while reading the prompt, generating ideas for the essay, organizing the content, and structuring the text. In addition, the students in the translanguaging groups were asked to explain the choice of languages in their drafts and whether their thought that using other languages contributed positively to different aspects of their writing. The questionnaire was translated to Norwegian to assure that the students understood the items and could provide exhaustive answers to the open-ended questions. Most of the items were based on a 4-point Likert scale with four options available, i.e. "strongly disagree", "disagree", "agree" and "strongly agree". A few items required the students to indicate the extent to which a particular language was used ("very little" to "very much"). Finally, a few items were formulated as open-ended questions, specifically, the ones that were designed to elicit the students' attitudes.

Results
In this study we aimed to investigate the students' use of background languages and their perceptions of the English-only, translation, and translanguaging writing conditions. We start the section by outlining the principal findings regarding the students' self-reported language use in three writing conditions. Next, we present the results obtained from the evaluative items of the survey.

Language(s) employed in writing
To begin with, we approached the first research question by looking into the use of L1/dominant language in the English-only condition because of the amount of L1 use reported in previous studies of the L2 writing process. Only 18 out of 81 participants (22%) chose the option "agree" and "strongly agree" when responding to the item about thinking in Norwegian and translating into English while writing the text in English. Furthermore, 64 participants (79%) in this group reported using "very little " or "fairly little" Norwegian to organize their texts. Finally, 63 participants (78%) reported thinking "fairly" or "very little" in Norwegian when writing in English. As anticipated, the students' responses suggest that most of them were in the monolingual mode (Grosjean, 2008) with English as the base language.
As pointed out in the method section, the participants in the translation group were supposed to start writing a draft in the monolingual mode with Norwegian as the base language and then switch to English for the final product. We asked the participants in the translation group whether they found themselves thinking in English when writing their draft in Norwegian. Thirty-five out of 58 participants (60%) gave a positive answer. In sum, the answers from the English-only and translation groups indicate that the participants employed English to a large degree not only in the English-only condition, but also when writing their drafts in Norwegian.
The language use reported by the translanguaging group is illustrated by Figure 1, which provides a more nuanced account of the students' strategic use of languages at different stages of the composing process when flexible language use is encouraged: These results indicate that different language modes (Grosjean, 2008) are associated with different stages of the composing process, i.e. the majority of the participants report to be in the English monolingual mode when they start writing the text, whereas Norwegian becomes heavily activated when the students engage in idea-generating, content-organizing, and text-structuring activities. A smaller number of the participants remained in the bilingual mode (mixed languages), in which mainly English and Norwegian but also additional languages were activated throughout most of the writing process. Interestingly, in this group, mixed languages was reported to be the second most preferred mode (after the English-only mode) at the draft stage.
In order to understand what motivated the students' choice of a particular language (or a combination of languages) at a draft stage when they were given the opportunity to choose freely, we added the following open-ended item to the translanguaging subscale: Which language-(s) did you use in your draft? Please explain why you did it in this way.
Twenty-six (44%) out of 59 participants of the translanguaging group who answered the question reported that English was the only language of choice in their drafts. Twenty-one out of 26 participants provided their reasons 5 for using English only: • It was faster and easier to stick to one language in the draft and in the final product (11 respondents).
• English was a natural choice considering that the prompt was partly in English as well (1 respondent).
• It was a personal preference to use English as much as possible (7 respondents).
• English was used to avoid translation since it would result in bad English (2 respondents).
Fifteen participants (25%) chose to write their drafts in Norwegian. Nine of them explained that they chose to stick to Norwegian because it helped them generate ideas for the text or/and made writing easier in general. Four participants used Norwegian to enhance the presence of the author's personality in the text. One participant reported avoiding mixing the languages which presumably would resemble oral speech in writing. Another participant took advantage of the  opportunity to write in a local Norwegian dialect, as opposed to one of the two written standard dialects, because it was faster and easier to do so.
The remaining 18 participants in this group (30%) engaged in translanguaging as they mixed either Norwegian and English or Norwegian, English and additional language(s). 13 out of 18 participants explained why they chose this strategy: • To solve vocabulary or/and grammar issues (3 respondents).
• For stylistic and rhetorical purposes, i.e. story was placed in certain linguistic and cultural settings or to engage the reader (3 respondents).
• To generate as many ideas as possible regardless of the language of thought (3 respondents).
• To take advantage of the rare opportunity to try out different languages since it was allowed (4 respondents).
The languages reported by the participants include English, Norwegian (standard and local dialect 6 ), Spanish, French, German, Greek, Icelandic, Japanese, and Nyanja (a Bantu language spoken in Malawi, Zambia, and other countries in Southern Africa).

Students' perceptions of the English-only, translation, and translanguaging writing conditions
We obtained the students' feedback on the respective writing condition both through Likert scale and open-ended items. We start this section with the analysis of the Likert scale items in each of the three groups.

English only group
The three feedback-related items for the English-only group were as follows: (1) I think that writing directly in English helps to learn the language.
(2) I think that writing directly in English helped me focus on English expressions.
(3) I think that thinking in English during the whole process is better than translating. 7 Figure 2 contains the results of the analysis of the items in question: We can see that most of the participants in this group agree with the statements about writing directly in English.

Translation group
For convenience, we repeat the items targeting the students' perceptions of the translation condition: (1) I found it easier to write first in Norwegian and then translate than to write directly in English.
(2) I found it difficult to translate my Norwegian essay into English.
(3) I think it is easier to write directly in English than to translate.
(4) In my opinion, thinking in English during the whole process is better than translating.
(5) I felt that I had time pressure to complete my translation into English.
The analysis of the items on the translation subscale is presented in Figure 3: It is clear from Figure 3 that overall the participants found translation to be more challenging than direct composition. Furthermore, half of the participants reported experiencing time pressure when translating their drafts to English. (1) The following two items on the translanguaging subscale aimed to evoke the students' opinions:

Translanguaging group
(2) I believe that use of other languages in the writing process contributes positively in such areas as (a) vocabulary, (b) grammar, (c) content, (d) structure.
(3) I prefer to stick to English during the whole writing process. Figure 4 contains the analysis of the responses to the items above: Figure 4 illustrates a striking discrepancy in the participants' perceptions of the potential benefits of translanguaging, as many of the responders found translanguaging to have a positive effect on vocabulary, grammar, content, and structure of their texts. Conversely, most participants reported that they preferred to stick to English during the whole writing process, which contrasts with their positive assessment of translanguaging on the first item.
We now turn to the analysis of the final open-ended items eliciting the students' feedback on the alternative writing conditions, i.e. translation and translanguaging. The participants in the translation group answered the following question: "Would you consider using translation more often when you write a text in English? Why/why not?" First, we divided the total number of the responses (N = 53) from the translation group into two categories: 41 negative (77%) and nine (17%) positive ones. Three participants (6%) formed a separate category as they pointed out both positive and negative sides of the translated writing.  Then the negative responses were broken down into subcategories based on which kind of reasoning the participants provided in their answers to the question. Out of 41 negative responses, 25 participants (61%) found translation to be a time consuming and strenuous process, while 13 participants (32%) thought that translation would have a negative effect on their English essays in terms of vocabulary choices, grammar and sentence structure. Specifically, one of the participants noted that use of translation could result in "Norwenglish". The remaining three negative responses provided reasons outside of the given areas.
The positive feedback from nine responders (17%) seemed to revolve around the idea of achieving a better structure and content in the English essays as a result of having a Norwegian draft as template. A few examples of the negative and positive comments are given in Table 2 8 : As mentioned earlier, three participants recognized both advantages and disadvantages of translation, namely, that it might be a good way to develop better writing skills in both Norwegian and English provided there was enough time to complete the task.
The students in the translanguaging group answered a similar question: "Would you consider using other languages more often when you write a text in English? Why/why not"?
Fifty-one participants in the translanguaging group provided answers to this question. Thirtyfour participants (67%) gave negative feedback; 14 participants (27%) found translanguaging to be   (2) I got a better structure of my text.
(3) … Those struggling with English can get some help. (1) It takes twice as long to finish the task.
(2) It was much harder to translate than to write in English from the beginning. (

3) It [translation] cripples my vocabulary horribly.
(4) My experience was that I had to concentrate really hard to prevent my English text from containing lots of Norwegian grammar, words and expressions.
(5) I personally like English better than Norwegian. That is why I think it is better to write texts in English. I feel less pressure since the English text doesn't have to be "perfect", which is required of the Norwegian text because Norwegian is my mother tongue.
(6) It is important to be able to think in English […] considering written and oral exams. beneficial in certain areas; one participant noted both advantages and disadvantages, and two participants answered "I don't know". The negative responses appear to be attributed to three distinctive factors: (a) some participants found mixing language to be confusing and demanding, (b) some preferred to use English as much as possible, and (c) some did not see any purpose of mixing languages in terms of learning outcomes. For example: (1) I don't like to switch to another language because then I have to change the way I think, and it is hard. (a) (2) It is very confusing and annoying. Fourteen participants (27%) who gave a positive feedback on translanguaging stressed the creative side of the process, as well as its potential to make texts more engaging and exciting. Some also pointed out that use of more than one language could help write longer texts in English. Below are some examples of the positive responses:

It opens up the possibility for including jokes and other stuff in the text.
It gives a more multicultural feeling to the text, and also makes it more fun to write and read.
(1) I think it was a new and better way to write. I liked that it[text] turned out to be much more creative.
It engages more parts of your brain and makes you think in a more varied way.
It helps to write longer texts.
I'd like to be able to use Norwegian … because I feel that I have more control over the text.
(1) The way you express yourself in English and Norwegian is rarely the same. Sometimes it might be good to use Norwegian words in English texts and vice versa to create the desired "effect". There is a lot you just cannot translate directly, so you have to rewrite it to make it fit another language, and then it loses its effect.
One participant pointed out that it could be fun to mix languages, though it could also be a bit strenuous to juggle two or more languages at the same time.

Discussion
Our first research question sought to understand the students' use of languages in different writing conditions during the drafting and final stages of the composing process. Our analysis showed that English as a metacognitive language of choice had a strong presence in all writing conditions. It is reasonable to assume that thinking in English had to be a prerequisite of composing a final product in English even when other language(s) was/were employed at a draft stage in the translation and translanguaging writing conditions.
Interesting findings in terms of language use were obtained from the translanguaging group because translanguaging allowed the students to employ the language(s) of their choice while the English-only and the translation modes limited the language of draft to either English or Norwegian (or another L1/dominant language). Both language modes (Grosjean, 2008), i.e. monolingual (English-only or Norwegian-only), and bilingual (mixed languages) were reported by the participants in the translanguaging group. The participants employed crosslingual writing mostly for idea-generating and writing a draft. This indicates that some of the participants adopted translanguaging as an alternative approach to compose a text in English even though it was an unfamiliar way of writing an essay in a typical English classroom in Norway. In this respect, it is important to note that the context of teaching English in Norway is monolingually oriented, i.e. in addition to the English-only format of nationally administered exams, the purpose of introducing other languages to ELT is limited to fostering metalinguistic awareness between English and L1 (The Norwegian Directorate for Education and Training, 2019).
In light of Wang and Wen (2002) model of the L2 writing process, the correspondence between various language modes and various stages of the writing process is of interest. A large proportion of our participants reported utilizing substantially more English at the task-examining and textgenerating stages (writing the final product), whereas idea-generating, content-organizing, and text-structuring activities are reported to be strongly associated with Norwegian. These findings support Wang & Wen's model, according to which the aforementioned facets of L2 writing are labeled as L1 dominant. In addition, the strategic use of L1 for organizing ideas and structuring information is consistent with the empirical findings in the field (Gunnarsson et al., 2015;Manchón et al., 2007;Wang & Wen, 2002;Wolfersberger, 2003;Woodall, 2002). However, the indicated use of L1 is associated with the responses of the translanguaging group, whereas the students in the English-only condition reported using Norwegian for the said purposes to a small extent.
The participants in the translanguaging group also answered an open-ended question about their choice of language(s) at a draft stage. The responses show that the participants are highly aware of their choices of language(s) for certain purposes. For instance, the choice of the Englishonly approach for the draft and the final product was explained in terms of saving time and effort. It is noteworthy that the participants who decided to stick to either English or Norwegian in their drafts had concerns about "bad English" or "Norwenglish" and thus purposefully avoided mixing the languages. Their negative attitude towards crosslingual writing suggests that the students are not encouraged to engage in such practices in school settings on a regular basis.
An important aspect of the participants' language preferences had to do with self-presence, rhetoric, and stylistic facets of writing. As a case in point, the students who wrote their drafts in Norwegian felt that it helped them strengthen their sense of authorship. This might be beneficial to Norwegian learners beyond our sample since, according to Horverak (2015), Norwegian upper secondary school students lack confidence in their English writing skills. Enhancing stylistic and rhetorical features of the texts is another facet of translanguaging mentioned by the participants, i.e. incorporating other languages into the fabric of the text gave the students an opportunity to engage the reader and convey desired atmosphere.
To address the second research questions, we asked the participants in the translation and translanguaging groups whether they would consider using the respective writing approaches in the future. Roughly two thirds of the participants in both groups expressed a negative attitude towards the use of either translation or translanguaging at the draft stage of writing. For instance, some described translation as time consuming and demanding, while mixing languages or translanguaging was confusing and distracting. These responses may reflect a prevailing monolingual orientation in L2 instruction in the Norwegian context. Considering that these conditions require writing two texts, providing the students with more time and opportunity to practice both translation and translanguaging may be a way of counteracting some of the expressed negative attitudes and persuading the students of the value of experimenting with translation and translanguaging while composing in an additional language (e.g., Prilutskaya et al., 2020;Turnbull, 2019). After all, about 17% of the participants in the translation group and about 27% of the participants of the translanguaging group (responses noting both negatives and positives of both conditions are not included) point to several important positive outcomes, namely, achieving a better structure and content because of the scaffolding function of translation attributed to having a draft in L1 as a starting point. When it comes to translanguaging, the participants noted the creative side of composing a text in this way; they also felt that translanguaging resulted in longer and more engaging writing. Importantly, when asked to reflect on the effect of using more than one language in their drafts, the participants noted positive effects of translanguaging on vocabulary, grammar, content, and structure of their texts. Finally, despite a large proportion of negative responses, some students in the translanguaging group expressed their willingness to engage in translanguaging for purposes of exploration, thus demonstrating their openness to experiment with languages. Openness to crosslingual writing practices in classroom settings conveyed by our participants can be viewed as a potentially effective tool for creating engaging and stimulating learning activities, which may facilitate students' ability to employ their linguistic repertoire in more innovative and learner-oriented ways.

Concluding remarks, limitations, and suggestions for further research
In this study we examined Norwegian upper secondary school students' self-reported use of linguistic resources while composing a text in English under three different writing conditions, and what they considered to be relative advantages and disadvantages of the assigned writing condition. The results indicate a strong presence of English as a metacognitive language of choice in all three writing conditions, the participants' strategic use of L1 for organizing ideas and structuring information, and their willingness to experiment with languages to enhance certain aspects of writing. Based on the results, we argue that translation and translanguaging may help balance the goal to develop students' English writing skills with their need to explore and utilize their linguistic resources, and thus engage in more holistic multilingual practices. Furthermore, making space for crosslingual writing practices offers learners a wider range of individualized writing strategies that may enhance certain aspects of writing and promote metalinguistic awareness and self-recognition of bi/multilingualism as a resource.
This study has its limitations. First, all data reported in this current study were based on self-reports and thus relied on the participants' understanding of the questions, their honesty, their introspective ability, and willingness to provide well-thought and adequate answers. We chose to elicit self-reports because of the nature of the phenomena being studied: internal processes and student perceptions. Therefore, it is not surprising that the students' responses were the primary source of data suitable for the purposes of this investigation. Another limitation concerns the data collection process, i.e. based on the feedback provided by the students who wrote their essays under the translation condition, 90 minutes may not be enough to complete the essays. Providing the students with enough time to complete essays might contribute to a more positive attitude to the writing condition in question. Finally, we employed descriptive statistics and a qualitative approach, which does not allow for generalization of the results to a larger population. However, replication studies in similar and dissimilar contexts may verify the methodology and findings of the study.
The main contribution of this study lies in its focus on the students' use of their linguistic resources as well as their perceptions of crosslingual writing practices. Our findings indicate that the students are strategic and inquisitive in the way they use their linguistic repertoire, and that there is more work to be done to harness the existing potential for the development of more individualized and engaging pedagogical practices. Specifically, empirical research is needed to explore the potential of translation and translanguaging as crosslingual scaffolding techniques in teaching writing in a target language as a legitimate alternative to the prevailing monolingual orientations in English language teaching in Norway and elsewhere. Future studies may benefit from expanding the methodological apparatus to include think-aloud protocols, focus-group interviews and stimulated recall data in order to obtain a more comprehensive understanding of the dynamic processes in the mind of emergent bi-and multilingual learners when they compose in a target language. Future research may recruit participants who have more experience with translanguaging in educational settings and beyond since their perceptions of translanguaging and the way they employ their linguistic repertoire when writing in a target language may differ from what the participants of the present research conveyed. Finally, a fruitful area for further work may include focusing on the utility of translation and translanguaging in classroom assessment design and practice due to the growing demand for innovative and inclusive approaches to language assessment in multilingual settings (see, for instance, García, 2009;López et al., 2017;Schissel, 2014).

Translanguaging (Continued)
Would you consider using translation more often when you write a text in English? Why/why not?

Statements
Which language(s) did you think in as you were reading the writing prompt?
Which language(s) did you use as you generated ideas for your essay?
Which language(s) did you resort to when you organized the content of your essay?
Which language(s) did you use when you thought through the structure of your essay (e.g., introduction, main part, and conclusion)?
Which language(s) did you use when you started to write the final text?
Which language(s) did you use in your draft? Please explain why you did it in this way.
I believe that use of other languages in the writing process contributes positively in such areas as: Vocabulary Grammar Content Structure Would you consider using other languages more often when you write a text in English? Why/why not?