The use of Arabic in Kuwaiti EFL classrooms: An exploratory study on the patterns and functions of language choice

Abstract This study examines the relationship between the functions and patterns of language choice in EFL classrooms in a Kuwaiti primary school. It applies the overall order model, specifically the medium of classroom interaction, to identify three patterns of language choice: an English monolingual medium, an Arabic monolingual medium and a bilingual medium. Then, an exploratory analysis of the relationship between specific patterns of language choice and broad functional categories of classroom code switching (CS) was conducted. No one to one relationship between a specific functional category and a specific pattern exists. One language pattern can serve different functions. From a detailed analysis, it is apparent that teachers in EFL classes try, as far as possible, to use the English monolingual medium to conduct pre-planned curriculum activities. However, when teachers and students departed from these prescribed activities, they used Arabic, either the Arabic monolingual medium or a bilingual medium, to serve different functions, such as accessing the curriculum, managing the classroom and establishing interpersonal relationships. On the other hand, the use of the prescribed medium (English monolingual) was never used to enhance interpersonal relationships.


ABOUT THE AUTHOR
Maha Alghasab is an assistant professor in the English Language Department, the College of Basic Education, Kuwait. She obtained her MSc in Applied Linguistics from Edinburgh University in 2011 and her PhD in Education from the University of York in 2016. Her main research interests are in discourse analysis (DA), computer-mediated discourse analysis (CMDA), computer-assisted language learning (CALL) and computersupported collaborative learning (CSCL), with a focus on wiki-mediated writing activities. She is also interested in bilingual interaction, especially bilingual classroom interaction and the use of L1 in Foreign language classroom.

PUBLIC INTEREST STATEMENT
This study aims to explore why teachers and students use their first language (L1) in English as a foreign language EFL classroom. The aim was to record and analyse their talk and then identify what types of talk they engaged in and how L1 is embedded in their English talk and why both teachers and students used it during the teaching and learning process. Case study research approach was conducted with fourth and fifth primary school students in Kuwait. The main finding of this study was that teachers used L1 for many purposes to explain the curriculum, to manage the classroom and to socialize with students. Another main finding was that English was used for pre-planned activity, whereas the use of Arabic language occurs in response to emergent problems related to students' understanding for the linguistics/content knowledge or to issue related to classroom management and socializing with students.

Introduction
Language alternation between two or more languages in bilingual classrooms has been receiving researchers' attention. The literature in this field has proposed many theories of interaction in bilingual classrooms, which have been developed in bilingual and multilingual communities. Some bilingual classroom studies have focused on identifying the functions of code switching (CS) (e.g. Adendorff, 1993;Camilleri, 1996;Ferguson, 2003Ferguson, , 2009Lin, 1996;Macaro, 2005;Metila, 2009;Paker & Karaağaç, 2015;Simon, 2001;Üstünel & Seedhouse, 2005;Walsh, 2002), while others have sought to characterize classroom talk and functions (Pennington, 1999) or identify the patterns of language choice in bilingual classrooms (Bonacina & Gafaranga, 2010). Ferguson (2003Ferguson ( , 2009) reviewed studies on the functions of classroom CS, and described these functions as fitting into three broad categories: CS for curriculum access; CS for classroom management and CS for interpersonal relations.
Drawing on the literature, Bonacina and Gafaranga's study (2010) was the first of its kind to focus on a different dimension; namely identifying and analysing patterns of language choice in bilingual classrooms. They applied the overall model (Gafaranga, 2007) and developed the notion of the "medium of classroom interaction", which was used to identify three patterns of language choice. Talk can be conducted in the prescribed medium of interaction (French monolingual medium), the English monolingual medium, or in both English and French (a bilingual medium).
The present paper will take the analysis a step further by exploring whether or not a relationship can be established between a specific functional category, in reference to Ferguson's CS functions, and a specific pattern of language choice. The conversational analytical (CA) approach, and the notion of medium of classroom interaction, will be used to analyse the patterns of language choice in EFL classrooms in the Kuwaiti context.

Code switching (CS) as part of bilingual interaction: Theoretical perspectives
Several decades of sociolinguistic research has observed that alternation between two languages in the form of CS is a widespread phenomenon in bilingual and multilingual communities (Martin-Jones, 1995;Turnbull & O'Cain, 2009). CS has become a characteristic feature of bilingual speech, rather than a sign of deficiency in one or other of the languages in use (Li, 2000). Language alternation can be understood from different perspectives; the grammatical perspective that is concerned with any linguistic constrains on CS (e.g. Poplack, 1980;Sebba, 1998), the interpretive and critical research perspective which relates micro-interactional functions of classroom CS to larger societal issues, such as the reproduction or sometimes contestation of linguistic ideologies in the larger society (Heller & Martin-Jones, 2001) and the socio-functional perspective (Auer, 1984;Gafaranga, 1999Gafaranga, , 2007Gumperz, 1982;Myers-Scotton, 1993) which treats CS as a discourse phenomenon since it is based on the assumption that the motivation for CS is fundamentally stylistic. Gafaranga (2007) further divided the socio-functional perspective into two broader perspectives: the identity related perspective, and the perspective of organizational explanation. Research based on the identity-related perspective has focused on the symbolic dimensions of languages, assuming that languages have social value (e.g. Gumperz, 1982;Myers-Scotton, 1993).
The organizational perspective, on the other hand, is more concerned with the contribution CS makes to the organization of conversation. The explanation provided by the organizational perspective can be divided into two categories: the local order model, which analyses language alternation by considering its sequential nature, i.e. turn by turn (Auer, 1984); and the overall order model (Gafaranga, 1999(Gafaranga, , 2007. Gafaranga criticized the local order model, stating that "language negotiation sequences would be meaningless if language choice acts were accountable only at the level of individual turns" (2007, p. 134), and he suggested an alternative framework, namely "the overall order model". He argued that in a bilingual conversation, there is a preference for same medium talk. The proposed model represents language alternation itself as the medium; "this corresponds to the normative use of two languages in the same conversation" (Gafaranga, 2007, p. 145).
In addition, language alternation can be an instance of deviance from the medium; this includes "medium repair", when alternation functions as a repairable deviance from the medium. According to Gafaranga's model (2007), the medium is not only monolingual but can also be bilingual, and, while deviance from the medium may be repairable, it may also be functional in terms of interactional othernesses. On this point, Gafaranga (2007) introduces the concepts of "medium suspension", which consists of temporary deviance from a medium that is not repaired and "medium switching", which occurs when speakers use one medium and agree to speak using a different medium.
In classroom context, Bonacina and Gafaranga (2010) developed the notion of classroom medium of interaction. While, CS refers to the alternation use of more than one linguistics code in the classroom by any of the classroom participants, the notion of medium of classroom interaction specifics which linguistic code is used by students and teachers while talking as opposed to the policy prescribed medium of instruction. The present paper applies the medium of classroom interaction to explore classroom interaction.

Functions and patterns of language choice in language classrooms
As mentioned before, studies in bilingual classrooms have, to varying degrees, provided lists of functional categories for CS, while few have focused on analysing patterns of language choice in classrooms as well as characterizing how classroom talk represents different frames. The following sections discuss these two research strands in depth.

Studies of functions
There is a consensus among studies that the use of L1 in language classroom is necessary to perform different functions (Ferguson, 2003(Ferguson, , 2009Halliday, 1994). Relevant to this paper is Freguson's works, which provided a succinct summary of functional use of classroom CS. The first function is CS for curriculum access to clarify and negotiate the meaning of the written text to enhance students' understanding of the subject matter including linguistic and content explanation. Studies have reported that teachers use L1 to provide metalinguistc comments (Raschka, Sercombe, & Chi-Ling, 2009), to teach grammar explicitly and translate difficult words (Edstrom, 2006;Macaro, 2005;Simon, 2001). Others (e.g. Adendorff, 1993;Probyn, 2009) have found that the use of L1 has a cognitive function such as clarifying and responding to learners' limited English proficiency.
The second function is CS for classroom management. This signals a shift in footing (Goffman, 1974) teacher may switch to motivate, discipline, or negotiate task instructions. Studies have indicated that L1 serves a repetitive function; teacher gives an instruction in the target language then repeats it in student's L1 for clarification and comprehension (Sert, 2005) or for shifting a frame from focusing on lesson content to a disciplinary issue (Lin, 1996). Others (e.g. Edstrom, 2006;Macaro, 2005) observed that teachers switch to L1 for clarifying task instructions.
The last function is CS for interpersonal relationship to negotiate different identities and maintain social relationship. Studies have found that switching to L1 conveyed symbolic values; L1 is used to establish friendliness and warmth, whereas L2 indicates detached relationship (Camilleri, 1996;Edstrom, 2006;Macaro, 2005;Raschka et al., 2009). Others (e.g. Canagarajah, 2001) found that both teachers and students switch comfortably between the two languages to construct their bilingual identities and refuse English only pedagogical ideologies. Khresheh (2012) observed that in Saudi Arabian EFL classrooms, teachers' use of L1 was related to cultural norms; they considered it shameful to commit mistakes in front of students, and, as a result of this belief, they avoided speaking solely in English, and switched to Arabic to avoid grammatical mistakes as far as possible.

Pattern of language choice studies
Compared with the previous research strands, there were studies that focused on a different dimension, namely identifying and characterizing the classroom talk into categories that can be differentiated in terms of functions, language and participants (Pennington, 1999) or patterns of language choice (Bonacina & Gafaranga, 2010). Pennington (1999) among others, proposed three frames of classroom talks. The first one is lesson frame, which is used exclusively by students in role-play activity to speak L2 for the benefits of the teacher. The second is Lesson support frame is used by the teacher who speaks in English to regulate the talk that is occurring within the frame of the lesson or when one student switches to ask about a grammatical point. The last one is commentary frame, which is used by students to make evaluative remarks about participants, to ask questions about the content in lesson frame and this also includes students' laughing and background talking. Bonacina and Gafaranga (2010) carried out a study in a French complementary school with the aim of exploring language choice. Based on the overall order model of bilingual conversation (Gafaranga, 1999), they developed the notion of classroom medium of interaction, which refers to "the linguistic code that classroom participants actually orient to while talking, as opposed to the policy-prescribed medium of instruction" (Bonacina & Gafaranga, 2010, p. 12). This medium can be used by teachers and students. Thus, when analysing classroom talk, the researcher should not assume that the language prescribed by the school policy is usually used as the code. This argument was based on their observation that participants at "La Colombe" school could choose to speak French, English, or a mix of French and English. Therefore, their main aim was to identify the base code that participants could switch from to achieve different purposes. They identified three patterns of language choice. The overall order perspective accounted only for one type of language choice pattern, in which classroom talk conducted in French was "the prescribed medium of instruction". However, the notion of medium of classroom interaction successfully accounted for all types of language choice, as it explained both normative language choices and deviance from them. The first pattern is French monolingual medium; participants switch from the prescribed medium of instruction to speak English for functional purposes, this switch either repairable or non-repairable. The second pattern is English monolingual medium; students use English as the base code rather than the medium of instruction (i.e. French), students adopted English as the base code while completing tasks in French (e.g. they negotiate the task and switch to French words that were written in the text). The last pattern is bilingual medium of classroom interaction, which represents the alternate use of French and English. It has been noted that this pattern of language choice could occur in either a mixed mode or parallel mode. In the parallel mode, both English and French are used as the base code; however, one participant may consistently use French, while the other uses English. Regarding the mixed mode, the researchers observed instances in which individual turns were completed in one language, and inter-turn language alternation occurred in classroom interaction.
Based on the literature review, it is apparent that some studies have identified functions of CS in classrooms, while others have identified patterns of language choice. This study aims to extend this analysis further, to examine whether a relationship can be established between a specific functional category and a specific pattern of language choice. To achieve this, the following research questions are proposed: (1) What are the patterns of language choice in fourth-and fifth-grade EFL classrooms in Kuwait?
(2) Is there any systematic relationship between Ferguson's (2003Ferguson's ( , 2009) functional categories of CS and the pattern of language choice? (i.e. Can a specific medium fulfils a particular function?)

Methodology
This study was conducted in EFL classes (fourth and fifth grades) in a girls' primary school in Kuwait. The fourth and fifth grades were chosen because it was presumed that students would likely be more fluent in English than other students in earlier grades. Convenient sampling was employed, and three teachers who were Arabic native speakers were chosen to participate in this study. Their length of teaching experience ranged between five and seven years, and they were aged between 27 and 30 years old.
Audio recordings were made of 10 lessons over a two-week period, each of which lasted for approximately 35 min. The audio recordings of the 10 lessons were first transcribed, and then an exploratory analysis was conducted to identify the observed patterns using the notion of medium of classroom interaction (Bonacina & Gafaranga, 2010). That is, for each interactional sequence, the base code was identified and where appropriate instances of CS in each interactional sequence were highlighted. The CA method was used to analyse language alternation. CA considers all the attributions of CS motivation should be dynamic, constructivist and grounded in fine-grained turn-by-turn analysis, taking into account all pertinent contextualization cues. Therefore, the use of CA helps to adequately address the function of speaker's CS by examining all relevant contextualization cues in bilingual interactions (Auer, 1995). Gafaranga (2007) argues that language choice in bilingual interaction is a social action, and that in order to understand the order of speech in bilingual conversations, analysts must identify its medium by observing speakers' reactions to their language choices. In other words, "social interaction is a norm governed activity, in the sense that each interactional act is either instance of a specifiable 'scheme of interpretation', or an instance of deviance from it" (Gafaranga & Torras, 2002, p. 19).
The notion of the medium of the classroom interaction differentiates the policy-prescribed medium of instruction from the actual medium of classroom interaction. Thus, to understand instances of language alternation, it is necessary to consider the actual medium of classroom interaction in each episode, rather than referring to the overall classroom policy and the prescribed language of instruction.
The data from the audio recordings were transcribed in accordance with the recommendations of Gail Jefferson (see Appendix 1). In order to analyse the functions of each pattern, Ferguson's (2003Ferguson's ( , 2009) classroom CS functional categories were applied.

Results and discussion
It has been observed that teachers and students used Arabic to serve different functions. The use of Arabic was observed in the three patterns of interaction. For example, in the English monolingual medium, the talk was solely in English to access curriculum (extract 1) or to give task instruction (extract 2), students may switch to Arabic. However, this switch was either repairable (extract 3) or functional (as in extract 4). It seems that participants do have a joint agreement that the code base for interaction is English since it is the prescribed language of instruction in this class. This joint agreement is clear since their switch to Arabic was repaired by the teacher or to serve a specific function. Here, in these extracts, we can see that the classroom participants use English monolingual medium to access the curriculum and to manage the classroom. 148T:

Pattern 1: English monolingual medium
give them food (.) drinks In the extract, both the teacher and the student are speaking English. The teacher asked a question in English, which elicited a student response. In turn 144, the student answered the question.
However, the answer was not acknowledged by the teacher, who then provided an English explanation to help the students in identifying an appropriate answer. In turn 146, the student provided another answer, which was accepted by the teacher who asked for further explanation. The English explanation in turns 145 and 147 helped the student to self-repair and give an appropriate answer. Thus, it can be said that the English monolingual medium was being used here to access the curriculum and discuss the subject matter.

Extract 2: English monolingual medium
13T: girls, you are going to match the verbs with pictures (.) cover the expressions and look at the pictures (.) okay 14 Ss: okay, teacher 15 T: first work individually, and then discuss it with your partner Again, the teacher used the English monolingual medium in extract 2 to provide the task instructions. In turn 13, the teacher told the students what they were going to do, and the students, in turn 14, acknowledged her using the discourse marker (okay). Then, the teacher used the same medium to instruct them to first work individually. The English monolingual medium was used to manage the class by providing instructions. In this extract, the teacher and students are using English; the student provided the second pair part of the question initiated by the teacher in turn 62. Clearly, the student was aware that the appropriate code was monolingual English, as she started turn 63 in English. However, "in a bilingual conversation, a speaker may depart from an initial choice when facing a problem resulting from a lexical gap" (Gafaranga & Torras, 2002, p. 6). The student deviated temporarily from the previous medium (English) and spoke Arabic to fill the lexical gap; this is clearly evident in the trouble marker (pause) in turn 63, which signalled that she had encountered a problem producing the required English word.

Extract 3: English monolingual medium
In turn 64, the teacher did not accept the deviant act, and identified the Arabic word as a problem, asking the student to repair the deviant act. In turn 65, there is a noticeable silence, which indicates that the problem has not been repaired yet. This prompted the teacher to ask for a repair, and for help. The teacher asked another student to answer the question and initiate the required repair. In order to access the curriculum, the teacher used the monolingual English medium to conduct a question/answer activity. However, when a problem occurred in the students' turns, the teacher acknowledged the need to use Arabic as a learning tool for enhancing comprehension and ensuring understanding. Therefore, she switched the medium temporarily (medium suspension) to resolve the problem.

Pattern 2: Arabic monolingual medium
The second observed pattern of interaction is Arabic monolingual medium. It has been observed that teachers and students speak solely in Arabic with no switch to English. The adoption of this medium served different functions such as explaining task instruction (extract 5), explaining given information from the textbook (extract 6) and to engage in social talk that was not related to the lesson content (extract 7).

→47T: la lazem tshofon elsorah wa tkamleen el ahrof elnaqsah (.) elkalimat elmatqrfonah la tjawbonha)
(no, you have to look at the picture and write the missing letters of the words (.) ignore unknown words, I am going to explain them) 48 T: do the exercise quickly, I want to see who finishes first In extract 5, classroom talk was conducted in the English monolingual medium; however, it is clear that medium switching occurred. Gafaranga explained that "medium switching occurs when participants stop using a medium and negotiate to use a different one, for whatever reason" (2007, p.147).
As the medium is structural (i.e. aspect of talk), the organizational sequence consists of two parts; the first part of the pair was uttered by the student in turn 46, when she asked about the exercise. In turn 47, the teacher provided the second part of the pair, by answering the student's question. Thus, the sequence of adjacent pairs was conducted in an Arabic monolingual medium to provide task instruction.

Extract 6: Arabic monolingual medium
40T: so drivers need to drink juice before the race (.) to drink what girls?
41Ss: juice 42T: good (.) they have to eat a full breakfast and to wear helmets →43 S: ablah hal rah yseer shi etha ma elbso el khotha (teacher, if they do not wear a helmet will anything bad happen?) →44T: la bs ashan ykono amneen akthar wo hatha shart fi shbaq (no, but to be safer they have to wear them, and this is part of the race conditions) As a result of a student asking a question in Arabic about wearing helmets in the race (turn 43), the teacher responded to her question in Arabic, and initiated the second pair part in Arabic. In this situation, the teacher was explaining a text called "car races" in English. However, in order to help the student understand the reason for wearing helmets she switched and answered the student's question in Arabic monolingual medium.

Extract 7: Arabic monolingual medium
→06 T: Nora wakiran dawamti shfeej kaybah (finally, Nora is here today, what happened to you, why were you absent?) →07 S: ablah thrsi kan y'aworni wa aboy wadani eltabib (Miss, I had a tooth pain and my father took me to the dentist yesterday) →08 T : shloonk elheen (how are you now?) →09 S: hamdellah bekir (I am fine, thank you) 10 T: fine (.) okay girls, look at the board, who can guess what we are going to talk about today? In order to find out what had happened to the student, the teacher spoke to her in Arabic to express her care and concern. She initiated the first pair part with a question asking about the reason for the student's absence, and the student provided the second pair part, which was the answer. In turn 8, the teacher initiated another Arabic turn, when she asked about the student's current well-being, and the student answered her in turn 9. In turn 10, the teacher shifted the frame; she emphasized the move from the socializing sequence to the lesson by asking a question about the lesson.

Pattern 3: Bilingual medium
The last observed pattern of interaction is bilingual medium whereby teachers and students mixed both languages without seeing them as different realities. In this pattern, it has been observed that there were instances in which students and teacher adopted mixed mode; turns were completed in one language and inter-turn language alternation occurred to explain grammatical rule (extract 8) or to explain task instruction (extract 9). →47T: okay, add s hak elfa'l go wa qoli eljomla kamlah (okay add "s" for the verb "go", and say the full sentence 48S: she (.) go (.) goes (.) she always goes to school 49T: yeah, yes, very good

Extract 8: bilingual medium
In this extract, the classroom discussion was conducted in English; the teacher initiated the first pair part of the question, asking for an English sentence. The student provided the second pair part in turn 42, using the same medium. However, the teacher regarded her answer as repairable, and therefore asked for a repair, indicated in her raised intonation in turn 43. Clearly, the student, in turn 44, indicated that the problem has not repaired yet. In turn 45, the teacher switched to using Arabic and English as one language to explain the grammatical rules, and in this way helped the student to self-repair. The student's silence in turn 46 indicated that the student was still having problems in understanding her mistake, and in response to this silence, the teacher produced another bilingual turn (47) in order to help students. In turn 48, the student initiated the required repair by uttering the full English sentence. In this case, the inserted repair sequence was conducted in a bilingual medium (see also extract 9).

Extract 9: Bilingual medium
T: banat (.) please look at your books wo look at page 18 in the work book, el tamreen talib minkom thwdoon el odd words (.) banat fi a'ndakoom four words and you have to circle one (girls (.) please look at your books and look at page 18 in the work book, this exercise requires you to circle the odd words (.) girls there are four words, and you have to circle one, girls) As shown, the teacher switched the mediums and used the bilingual medium to provide task instructions to help students to fully understand what was required to accomplish the task successfully.

Discussion
Three patterns of language choice were observed, all of which could be accounted for by the notion of medium of classroom interaction, confirming the findings of Bonacina and Gafaranga (2010). Classroom talk can be conducted in monolingual English, the prescribed medium of interaction, participants can deviate from this medium via a deviant act, which could be described as either functional or repairable. The second observed pattern was the use of Arabic monolingual medium, and the third pattern was a bilingual medium in which participants used both English and Arabic as the base code for their interactions, and the alternation itself was the medium of interaction.
The data analysis sought to establish whether any correlation between a specific function and a specific pattern could be identified. It was impossible to identify a systematic relationship between a pattern of language choice and Ferguson's (2003Ferguson's ( , 2009 functions. In many situations, a pattern can fulfil three different functions. For example as illustrated in the extracts, the Arabic monolingual medium has three functional categories (accessing curriculum, managing the classroom and maintaining interpersonal relationship) and the bilingual medium and English monolingual medium can be used both to manage the classroom and access the curriculum.
Rather than establishing a one-to-one relationship, it has been observed that the teacher used the English monolingual medium for a prescribed curriculum activity. This is similar to Pennington's (1999) lesson frame where students were speaking in L2 to fulfil a role-play activity, which was part of their prescribed curriculum. When teachers departed from prescribed activities they used Arabic, either in the form of Arabic monolingual or as a bilingual medium. Teachers adopted a new medium to either establish social relations, access the curriculum (e.g. explanation and illustration based on student's responses and answers or needs) or to manage the classroom talk. The idea of "accessing the curriculum" refers to prescribed activities that teachers have planned to do in English. It was only when departing from this prescribed routine that they use Arabic. Thus, no one to one relationship could be identified between the pattern and the function of their language choice.
The use of Arabic medium aroused unplanned, as a reaction to emerging problems or situations the teacher needed to respond to the problems raised, and to the students' needs, as they occurred. Emerging situations, trouble markers, incorrect answers in relation to content and language or unplanned activities, such as responding to students' queries about the tasks, socializing with the students occasionally forced the teacher to switch medium and adopt either a monolingual Arabic or bilingual medium to control and respond to particular situations. Once these activities had been completed and problems had been resolved, the teacher immediately switched back to the English monolingual medium to move on to the next prescribed activity.
The use of the Arabic monolingual medium was not exclusively for solving emerging problems, but was rather intended to reinforce social relationships and to build a rapport with individual students and foster personal warmth to encourage pupil participation and motivation. By contrast, the prescribed medium (English monolingual) was used to conduct the speech acts that teachers had preplanned before entering the classroom but never used to build interpersonal relationship.
The deviant acts in the English monolingual medium and the use of Arabic monolingual medium observed in this study lends further support to the findings of previous studies (Eldridge, 1996;Khresheh, 2012;Macaro, 2005;Pennington, 1999;Probyn, 2009;Sert, 2005) that have reported that CS is readily available as a learning strategy in classrooms. However, it cannot be assumed that the Arabic monolingual medium is the only medium that can be used to access the curriculum, as in many cases the English monolingual and bilingual mediums can also be employed to facilitate accessing to the curriculum. Although previous studies (e.g. Willans, 2011) have demonstrated that two codes are in use in bilingual classrooms, where interactions in which the teacher was involved were found to be predominantly conducted in English, while the mixed code was used in students' interactions, in this study the data showed that teachers themselves used three different patterns of language choice to deal with emergent situations as they occurred. Arguably, this is because teachers are expected to use all possible means to fulfil their duty of teaching students a foreign language appropriately, and correctly.
The findings of the study are in line with those of Camilleri (1996) and Lin (1996), in the sense that the participant teachers never used the English monolingual medium or a bilingual medium to build interpersonal relationships. Rather, they used the Arabic monolingual medium to establish social relationships, which supports the findings of the aforementioned studies. Such studies have successfully shown "how the classroom is not only a place for formal learning, but also a social and affective environment in its own right" (Ferguson, 2003, p. 43). They all recognize that the foreign language invokes a more distant relationship, whereas the use of the shared language invokes solidarity and reinforcement.

Conclusion
The main finding of this study is that teachers try to use a specific medium for pre-planned activities; however, emerging situations and problems may force them to respond using a different medium, either an Arabic monolingual medium or a bilingual medium. Based on this finding, it is important to remember that writing expected problems and solutions in teacher's plan is highly important if they want to use L2 extensively in their language classroom. Teachers should lists a number of different problems related to student's behaviours, task instruction and linguistics and content understanding and try to think ahead how they respond to these as they emerged. In this way, teachers could minimize their use of L1. This does not mean that L1 use should be avoided as we have seen in some extracts the use of L1 plays an important role in enhancing student's understanding. What teachers need to know is that English monolingual medium can be the right medium to adopt in language classroom context. Using L1 is acceptable, however, teachers should either try to repair it or to use it to fulfil specific functions. Then, an immediate switch is necessarily to help students to be exposed to L2.
The main pedagogical implications of this study are the importance of conducting professional development sessions to increase teachers' awareness of the differences between the medium of interaction that may occur in their language classroom. Teachers should understand that the medium of interaction should be informed by emerging classroom situations and events. In other words, teachers should employ an "interactive decision-making" strategy, whereby in every lesson they will need to make fast decisions about which medium to use in order to address emergent problems, whether these are related to understanding the subject matter, classroom management or establishing a good rapport in the classroom.
As the current study had a sample size of only three teachers, it may not be possible to generalize the findings to all EFL classrooms. Therefore, further research is recommended to examine the relationship between patterns and functions of language choice with a larger number of teachers, as a more comprehensive investigation of the issue may help to generalize the conclusions drawn in this study. In addition, as the present study focused on the patterns and functions of language choice in a teacher-centred classroom, it would be useful for future studies to examine student-centred classrooms, in order to study student-student interaction rather than solely focusing on teacher-student interaction, in which the teacher controls classroom talk.