Why do teachers code-switch when teaching English as a second language?

In post-apartheid South Africa, 11 languages, including English and indigenous languages, were given equal status as official languages. Yet, more than 25 years after democracy, Black parents still believe that their children’s wealth and success depend on English, and therefore, send them to English-medium schools where they take English as a second language (ESL), known in South Africa as English First Additional Language (FAL). Many circumstances compel teachers to code-switch between English and learners’ first language. In the study reported on here we explored the reasons behind teachers’ code-switching in FAL classrooms in 4 rural high schools in KwaZulu-Natal, South Africa. Concept maps, open-ended questionnaires and open-ended telephone interviews were used to generate data. Findings indicate that FAL teachers use code-switching to clarify difficult concepts, enhance understanding of the content presented, and keep learners engaged during lessons. We argue that the use of learners’ first language may be crucial in the teaching and learning of a second language.


Introduction
Despite the Language in Education Policy (Department of Education, 1997) in South Africa aiming at redress of the malpractices of the past apartheid education system and giving equal status to 11 languages, parents, including those of learners in rural high schools, view English as a language of wealth and success, and opt for English as a language of teaching and learning for their children. These learners opt for English as a second language (ESL), known as English First Additional Language (FAL) in South African schools. For most learners in these schools, English is a second language (L2), and they are often taught by teachers who are L2 speakers of English themselves. While learners often struggle in their attempts to engage with the content that is presented to them in FAL classrooms, teachers know that they need to get learners to understand the language taught and used for teaching and learning.
To overcome the language barrier to teaching and learning, code-switching has become a communication strategy for teachers and their learners as they attempt to express themselves, thus breaking down and transcending the institutionalised ethnic barriers of apartheid (Finlayson & Slabbert, 1997). It is in such circumstances that teachers find themselves using code-switching to teach their learners.
However, despite a fairly progressive Language in Education Policy, studies indicate that even where teachers feel that they are justified in using code-switching for classroom interaction to scaffold the learning process, they feel that they are breaching not only the official language policy but what is presented to them as best classroom practice by curriculum specialists or the Department of Basic Education (DBE) (Kretzer, 2019;Probyn, 2009).
We argue that the teaching of the second language (L2) may be enhanced by the use of learners' home language (L1), since learners already possess a home language system with its communicative structure which has the potential to enhance the learning of the target language. We explore the reasons why teachers codeswitch in the context of EFAL teaching at four rural South African high schools and aim to answer the question: Why do teachers of English at four rural high schools code-switch when teaching the subject?

Literature Review
In his definition of code-switching, Gumperz (1982:59) identifies it as a "juxtaposition within the same speech exchange of passages of speech which belong to two different grammatical systems or subsystems." This means that an association of speech or discourse belonging to two different languages occur at the same time during a conversation. Extending this, Myers-Scotton (1993) regards code-switching as constituting a context where two languages within the same conversational turn are used. Furthermore, this is described as a conversational turn which incorporates an insertion of a word or phrase to enables mutual understanding between or among speakers while also considering the three contextual contexts, namely, the relationship among speakers, the setting where the interaction occurs, and the topic under discussion (Ariffin & Husin, 2011). Itmeizeh, Ibnian and Sha'fout (2017) and Wang (2016) reiterate that when speakers use code-switching, they use it in a nonambiguous, flexible and contextually-free manner and do not perceive their first language (L1) as a deficiency, but instead, as a resource that enables them to find a better way to convey meaning to each other.
While code-switching incorporates translanguaging, in which bilingual and multilingual speech communities use various languages instead of just two, our focus is on code-switching. The teachers are FAL teachers who speak isiZulu and English in the school context, and they are fairly competent in both languages. Their L2 learners are more competent in their first language, isiZulu, and attend English FAL classrooms.
Code-switching is a practice of using more than one language, usually two languages, in an EFAL classroom during any form of interaction. Swain andLapkin (2013, cited in Cummins, 2019), encourage learners' meaningful use of their home language (L1) to demonstrate cross-linguistic comparisons or to supply the meaning of abstract vocabulary items (p. 123). Swain and Lapkin suggest that learners should be awarded opportunities to use their L1 in collaborative interchanges or private speeches to facilitate their understanding and production of complex ideas in their preparation to produce the end product (oral or written) in the target language (pp. 122-123). Similarly, Bartlett and Garcia (2011) maintain that learners should have the freedom "to draw on the totality of their linguistic resources in carrying out academic tasks" (p. 23). Opić (2016) notes that as teachers use code-switching in ESL classrooms, they create positive interpersonal relationships with learners which in turn enable mutual understanding between teachers and their learners. Mutual understanding enables thorough understanding of the content presented in class and thus promotes positive relationships between teachers and learners. This further allows learners to be creative and successful in their academic work.
In a slight deviation from the ideas presented above, Auerbach (2016) indicates that teachers should not allow learners to use their L1 indiscriminately but should be selective in their approach. Examples of where learners' L1 may be used include group writing tasks, learners reading and translating stories from L1 to L2 to each other, learners paired together with those that are more fluent in English, learners' use of bilingual dictionaries for difficult texts, and use of family members to assist with school work in the learners' L1. Auerbach (2016) reiterates that by providing such opportunities, teachers, as knowledge generators, can construct instructional initiatives that challenge the exclusion of learners' L1 from the school.
However, code-switching usage has not been positively perceived by many early researchers, especially when used in the classroom situation. For instance, a study by Chaudron (1988) and Wong-Fillmore (1985) portrays code-switching as a hindrance to successful learning processes proposing that it forces learners to rely on codeswitching, which then reduces learners' exposure to English, and hinders their acquaintance with L2 subject terminology. Additionally, code-switching is deemed to negatively influence learners' communication skills in the L2 (Zhu, 2008), and it is seen to allow learners to commit errors while using the language without even realising it (Jingxia, 2010). Furthermore, the use of codeswitching in the teaching of L2 is perceived to make learners lose the eagerness to learn the L2 and learners fail to guess and infer in their new linguistic environments of L2 (Nordin, Ali, Zubir & Sadjirin, 2013).
However, more recent studies acknowledge the positive role that code-switching plays in an ESL teaching (Gulzar, 2014). Functions, such as classroom management, language analysis, rulesgoverned grammar, discussion of cross-cultural issues, giving of instructions, explanation of errors, and checking of comprehension have been associated with the use of code-switching in ESL teaching (Gulzar, 2014). Furthermore, studies highlight the role that code-switching plays in clarifying difficult concepts, checking learner understanding, reinforcing learners' vocabulary (Lin, 2013;Magid & Mugaddam, 2013;Mahofa & Adendorff, 2014), and linking learners' existing knowledge in their L1 to the new vocabulary and context in the target language to enhance mutual understanding (Songxaba, Coetzer & Molepo, 2017). In addition, code-switching has been found not just to enhance teaching and learning but to serve as an empowerment strategy to improve learners' performance (Maluleke, 2019). The studies cited above serve as an important review of studies done before and their shaping of the study under discussion in this article.

Theoretical Framework
Our study was underpinned by Gumperz's (1982) semantic model of conversational code-switching. While speakers participate in an interaction, they insert words or larger portions of language and can create multiple relationships between the language they use and social meaning (Gumperz, 1977). The speakers are awarded an opportunity to create different relationships or identities using linguistic means to create a shared context where both speakers involved in a conversation understand what is being conversed. During their interaction they also account for their choice of switching codes. This makes the Gumperz model central to this study as its purpose was to explore the reasons for teachers' code-switching in FAL classrooms. The Gumperz model also allows for the categorisation of code-switching instances into situational and metaphorical (Gumperz, 1982). This means that speakers are able to base their reasons for switching codes on the situation at hand or on metaphorical intent, such as when code-switching is used for quotations, reiterations, message qualification, interjections and addressee qualification (Gumperz, 1982).

Paradigm, Approach and Design
To explore why teachers of English at four rural high schools code-switch when teaching the subject, an interpretivist paradigm, a qualitative approach, and a single case study design were used. The interpretivist paradigm was suitable for the study as it aims to understand human action and human experience, and to explore how participants make sense of their personal and social world (Cohen, Manion & Morrison, 2011). The interpretivist paradigm enabled an exploration and understanding of the reasons behind the use of code-switching by four rural high school teachers of English.
A qualitative approach worked well with an interpretivist paradigm as it allowed for comprehensive reports that allowed us to make sense of the feelings, views, and experiences of the participating teachers, and of the social contexts surrounding their teaching, and how these affected them in their real world or context.
Together with the interpretivist paradigm and qualitative approach, we used a single case study design involving four teachers in four rural high schools within a specified district on the south coast of KwaZulu-Natal. In this single case study we explored the reasons behind the four teachers' use of code-switching in their English classrooms. A case study allows for exploration and understanding of multiple facets of a phenomenon within its context by employing various data sources (Yin, 2009). The case study design allows researchers to answer the how and why questions, it does not manipulate the behaviour of the participants and it covers relevant contextual conditions of the phenomenon under study (Yin, 2009). The choice of a single case study design in this study was prompted by the understanding that there was no attempt at identifying similarities and differences among the teachers but instead we aimed at exploring a deep understanding of the reasons behind FAL teachers' use of codeswitching in rural high schools. This is in line with Siggelkow's (2007) contention that single case studies provide opportunities for deep understanding of a phenomenon.

Sampling
For this qualitative, interpretive case study, purposive sampling was used to allow a selection of a distinct group of participants for whom the research questions would be of significance. Rather than using random selection, the participants were hand picked to ensure that the best information was most likely to be obtained from a relatively small sample with known attributes and because of their experiences and insights in the field under study (Cohen et al., 2011).
Four African teachers aged between 35 and 53 were chosen to participate in this study. All participants spoke isiZulu as L1 and English as L2. The selection criterion was that they taught English as a second language to isiZulu-speaking learners, and also used English as the official language of learning and teaching to teach other subjects at their schools. Two of the participants were specialists in the field of study and held honours degrees in linguistics. The other participants were non-specialists and taught FAL due to a shortage of English specialists at their schools.
All four high schools were selected because they were rural and located on the south coast of KwaZulu-Natal. The communities from which the learners came had little or no English-speaking background but the carers opted for their children to attend English-medium schools for the perceived benefits associated with English. Two of these schools were categorised as no-fee-paying schools. These two schools did not have libraries and laboratories, among other resources, and learners came from poor communities. The two schools hosted between 200 and 400 learners each. The other two schools were slightly better resourced and hosted between 700 and 800 learners each who came from a diversified context, with some members of their communities having a good command of English.

Data Generation and Analysis
After piloting the instruments with a similar sample and making the necessary changes to the instruments, data were generated using concept maps, open-ended questionnaires, and audiorecorded in-depth interviews. The three different but complementary methods contributed to the triangulation of the evidence for this study (Cohen et al., 2011). To probe deeper and obtain enhanced responses from the respondents, comparable questions were asked differently in the three data generation instruments. These instruments were developed to present three diverse ways of generating data and thus ensuring triangulation.
In the first data collection instrument, the concept map, we used thought bubbles and arrows for participants to brainstorm ideas (Eppler, 2006) and to enable us to understand the participants' reasons for code-switching. Questions as prompts were provided and participants could provide responses in their own space within a month. This method proved invaluable as it did not demand a prepared form of narrative or impulsive answers. The use of concept maps allowed ideas to be "developed and reviewed quickly due to the fact that ideas are drawn in the form of keywords, shapes, and arrows" (Baugh, McNallen & Frazelle, 2014:4). In this process, Baugh et al. note that the brain is visually stimulated to extract ideas, giving participants the freedom to think out of the box and enabling them to remember using both images and words.
The open-ended questionnaires allowed for an inexpensive tool for gathering data and were easy to collect (Sekaran & Bougie, 2010). The participants were asked the same questions based on the research question of the study. The anonymous, open-ended questionnaires were distributed and collected by us to establish rapport for the interviews that would follow.
The qualitative, open-ended, audio-recorded interviews were designed for the researchers to probe participants to provide more information following the concept maps and questionnaires. The interviews provided us with authentic information about the participants' insights into and understanding of the reasons behind their use of code-switching (Cohen et al., 2011).
As data were generated, thematic analysis took place. This involved reading the data and identifying possible themes. It also involved analysing and interpreting the recurring patterns in the data (Braun & Clarke, 2006). Once the data from each instrument were analysed, thematic analysis took place across the instruments.

Reliability, Credibility, Transferability
To ensure reliability, well established data generation methods were used. To build trust and rapport, a sound relationship was established with the schools and teachers involved in the study. To ensure credibility of the data, interview transcriptions and findings were provided to the participants for checking. Since generalisability cannot be obtained through a single case study (Rule & John, 2011), transferability is aimed for. This is done by keeping detailed information about the schools, participants, data generation methods and sessions, and contexts. If readers deem the contexts described to be like theirs, the findings may be relative to their own contexts.

Limitations
Certain limitations of the study were evident. Firstly, as the study involved only four rural high school teachers in KwaZulu-Natal, South Africa, the findings could not be generalised. Future studies could consider larger numbers of participants over a greater geographical area. Secondly, it is possible that participants shared what they considered to be the "right" answers. We tried to address such bias by using multiple data generation instruments. Thirdly, despite piloting, questions in the questionnaire might have been misunderstood. Again, the multiple instruments aimed to alleviate such. Finally, a limitation of the study is not using ethnographic observations which might have provided rich insights. Addressing this limitation could be considered for future research.

Discussion of Findings
The findings in this study are presented thematically and the participants are referred to as ED1, ED2, ED3 and ED4. The themes identified to answer the research questions are: Learners' Limited Exposure to English; Code-switching as an Aid to Understanding; and Code-switching to Create an Enabling Environment.

Learners' Limited Exposure to English
The interviews with the four participants revealed an insight into their learners' limited exposure to English. Referring to English, ED2 noted that "our learners are not familiar with the language." ED1 was clearer and explained that the learners "come from Black rural communities where English is not used", and "they only use English when they are at school." The statements from the two participants indicate the realities of learners' linguistic contexts. Learners do not speak English at home, evidently choosing to speak their home languages. A possible predictor explaining the reason for code-switching was ED1's understanding that learners only experienced English at school. The responses above emphasise findings in Hibbert and Van der Walt's (2014:213) study which highlights the challenges that African-language speakers encounter when entering an English-dominated environment, which then requires of them to be competent in the language of learning and teaching.
Explaining learners' limited exposure to English, a respondent noted in a questionnaire that "learners have a poor English background from primary" and another noted that, in primary schools, "these kids are taught in mother tongue, so when they come to Grade 8 you have to take them slowly from mother tongue." The respondents' statements in the questionnaires place the blame for poor English language proficiency on the primary school teachers for failing to teach learners the language. It, therefore, becomes clear that teachers would need to code-switch to bridge the language gap and fill the gap allegedly left unfilled by the primary school.
However, in an interview, ED2 noted that "at high school, you find that only the English teacher teaches in English. The rest of the teachers in the school teach in the mother tongue." This indicates that teaching is continuing in the mother tongue by high school teachers who are meant to be teaching through the medium of English. It is possible that effective teaching and learning is taking place in the mother tongue but if learners are to be assessed in English and have not had practice using the language, they would possibly be severely disadvantaged. Clearly, ED2 did not believe that code-switching was taking place in the classrooms cited.
A lack of resources also characterised learners' limited exposure to English. In response to the questionnaire one respondent noted that "learners do not have dictionaries" and another respondent noted that "they depend on Government for learning materials, but they are not provided with them." It became clear that the learners' limited exposure to English was made worse by not receiving learning materials that could assist them.

Code-switching as an Aid to Understanding
In a concept map, a respondent noted that codeswitching was used "to aid understanding." The idea of code-switching as an aid to understanding permeated many responses across the instruments. Questionnaire respondents noted that codeswitching was used when "trying to explain things to the learners, when they need further explanations; when the educator wants to consolidate understanding; to explain and clarify terms; to explain difficult parts in a lesson; to explain questions so that learners know what to answer." It is clear that the reasons proposed for code-switching were designed to enable effective teaching and learning. Not using code-switching could potentially thwart learners' understanding. This finding is in line with the literature on codeswitching (Mahofa & Adendorff, 2014;Then & Ting, 2009).
In the interviews, participants were more specific, with ED2 noting that code-switching "fits well in literature", especially to "explain figures of speech like irony, the difference between an oxymoron and a paradox, and provide examples of what an oxymoron is." ED3 explained that codeswitching helped to explain "phrases, idioms, and proverbs", and ED1 indicated that code-switching was used when learners were faced with a "new or foreign setting in a comprehension passage which presents a problem." In an English classroom, new concepts and terminology may prove difficult to many learners and using code-switching could assist them to get acquainted with new vocabulary, understanding of concepts, and the content of comprehension passages, as similarly identified by Lin (2013) and Magid and Mugaddam (2013). Such examples of code-switching in the study and literature reflect situational code-switching where the reasons for code-switching are based on the situation at hand and metaphorical code-switching which is used for reiterations and message qualifications (Gumperz, 1982).
In the interviews participants also identified code-switching as useful to make connections with and understand learners' realities. In an interview, ED4 noted that "to progress from the known to the unknown, people depend on their home language. We must relate to their contexts." Later in the interview ED4 noted that "code-switching helps learners make connections from their past to their present learning experience through using codeswitching in informal discussions." Stated differently, ED3 stated that a teacher uses codeswitching with learners "to relate to their real-life situations." Both ED4 and ED3 highlighted the importance of using code-switching to recognise learners' home languages, the contexts in which they live, and their prior learning experiences. These ideas are reiterated by Siong and Min (2017) who emphasise the value of incorporating learners' prior and existing knowledge in their L1 through code-switching.

Code-switching to Create an Enabling Environment
The use of code-switching was also identified as useful to create an enabling environment. In the interviews, ED1 shared that "what causes discomfort is that learners don't pay attention when you teach. You have to code-switch using isiZulu." While ED1 identified a discomfort, there was a solution found in code-switching which appeared to assist with learners' lack of attention. If code-switching helped to refocus learners' attention, it would probably assist in creating an environment conducive to teaching and learning. Gulzar's (2014) study, too, found that codeswitching serves as a useful means when managing a classroom.
In a questionnaire, a participant noted that s/he used code-switching to create an enabling environment. The participant used code-switching "for learners' self-fulfilment and meaningful participation." Later in the questionnaire, the participant noted that "I use it for inclusivity purposes because language shouldn't be a barrier to learning." The participant pointed out the importance of creating a learning environment where learners thrived and interacted meaningfully. The participant was also aware of the importance of including all learners, not just for effective teaching and learning, but to enable learners' self-worth and confidence, possibly the best way to ensure success in a classroom and elsewhere. This finding is in line with that of Maluleke (2019) who recognises the empowering nature of code-switching and supports Gumperz's (1982) assertion that codeswitching creates a shared context.
Another idea on how code-switching may help to create an enabling environment was revealed in two concept maps. In one concept map it was revealed that code-switching was "useful for free conversations, especially because people learn better in an informal environment." In a second concept map is was noted that code-switching helped a learner "engage better if s/he has that little freedom." The sentiments expressed in the two concept maps revealed that the participants understood the power of code-switching to enable an effective teaching and learning environment. They also understood that such an environment was not restrictive and was characterised by the freedom to express and engage. This is in line with Gumperz's (1982) contentions that code-switching can create relationships between the language learners' use and the social meaning.
Finally, ED1 noted in an interview that "I use code-switching because my learners are Zuluspeaking and I'm a Zulu-speaking educator." It is possible that ED1 used code-switching as a form of solidarity and identification with the learners, as is seen in a study by Jingxia (2010).

Conclusion
With an understanding of the language imperatives in South Africa's policies, with this study we sought to understand the reasons for four teachers' use of code-switching in their rural high school English classrooms. The findings indicate that the contexts from which learners come result in a limited exposure to English which hinders successful teaching and learning. The sampled teachers thus use code-switching to aid understanding and to create an enabling environment in which to teach and learn. These stated reasons indicate that teachers know that they code-switch and recognise the value thereof. This is contrary to literature (Probyn, 2009) that shows that teachers who code-switch feel that they are going against policy and best practices for teaching.
If South African teachers have concerns about the possible negative effects of code-switching, the study reminds them and all other language practitioners that they must become au fait with the language policies of the country which do, in fact, encourage code-switching. If they recognise that code-switching is a legitimate language learning strategy, they will use it with more confidence. They should also be empowered via training workshops, and other means, to learn about language-related policies and debates, and to learn how to integrate code-switching effectively into their teaching strategies to enable maximum effectiveness of the practice.
If we try to negate learners' home languages, we will be perpetuating an ideology that certain languages and certain knowledges are worthier than others. If we try to exclude learners' home languages as they learn a second language, we will ensure that learning of the second language becomes significantly more difficult. We re-affirm the argument that the teaching of a second language will be appreciably assisted by the learners' home language.