A Case Study of Vietnamese EFL Teachers’ Conception of Language Output and Interaction

There exists a gap between what second language acquisition research has to say and what teachers do in the classroom. As part of an attempt to bridge this gap, this study is driven by the motive to understand how pedagogical innovations such as task-based language instruction can be influenced by teacher beliefs. Drawing on the perspective of research on teachers’ thinking which aims to inform language teaching pedagogy and teacher education, the study employed multiple data sources (focus group, lesson plan, and stimulated recall interview) to tap into the conception of six Vietnamese EFL instructors regarding language output and interaction. Cross-case analysis showed that most of the teachers geared language output and interaction activities towards achieving a targeted linguistic aim. Further analysis revealed that this view reflects a synthetic, product-oriented conception of teaching and learning by skill-building, and is in line with traditional approaches which emphasise transmission style and form instruction. This finding implies that constructivist perspectives on teaching such as task-based language instruction may run counter to teachers’ existing conception of teaching. The implementation of task-based instruction thus needs to consider negotiating between supporting teachers to focus on meaning and the need for form-focused instruction.


Introduction
The transfer from theory to practice is still generating research interest in language learning and teaching (Mackay, Birello & Xerri, 2018) and contextual influences have become a crucial issue. In Vietnam, English is a school subject, and English teachers are familiar with the transmission style of teaching and form-focused instruction (Nguyen, Le & Barnard, 2015). However, recent curricular innovations have embarked on meaningoriented, task-based language instruction that facilitates a constructivist approach to language teaching (Barnard & Nguyen, 2010). This innovation has attracted much attention among scholars and researchers. One assumption is that the new approach may run counter to teachers' existing beliefs and practices.
Decades ago, it was cautioned that teachers' beliefs constituted one of the strong forces that restricted the impact of educational reforms (Fullan, 1993). Such awareness has led to a boom of research on teacher cognition in an attempt to unveil the hidden side of teachers, which aims to offer implications for teacher education and development (Borg, 2006). Johnson (2006, p.239) has pointed out "an epistemological gap between how L2 teacher educators have traditionally prepared L2 teachers to do their work and how L2 teachers actually learn to teach and carry out their work".
Theoretically, knowledge about second language acquisition (SLA) principles is argued to offer teachers a tool for trial in their classrooms, and the opportunity to reflect and change (Johnson, 2006;MacDonald, Badger & White, 2001). However, research that focuses on the effect of SLA theory training has revealed little impact on teacher beliefs about language learning and teaching (MacDonald, Badger & White, 2001;Mattheoudakis, 2007;Peacock, 2001). In this respect, according to Berliner (2005), research on teachers' cognition should concentrate on "phenomena that have been found important from the perspective of the process-product research programme" (p.14). Thus, basic SLA principles such as language output and interaction in SLA theory merit an inquiry. Instead of investigating teachers' perceptions of task-based language teaching, this study explored their responses to these SLA-enabling conditions underlying the task-based approach. This bottomup approach to understanding teachers' reactions is expected to provide insight into how teachers perceive and uptake task-based language teaching (TBLT). The report in this paper is part of the researcher's doctoral thesis (Nguyen, 2011).

Literature Review Conceptual Framework of Teacher Conceptions
Teachers' thinking has generated much interest in educational inquiry (Eley, 2006;Day, Calderhead, & Denicolo, 2012;Kember, 1997). However, the development of research on teachers' cognition has led to overlapping terminologies (Borg, 2006). Given this overlap, the current study employs the term 'conception' and Pratt's (1992) conceptual framework of teacher conceptions.
As noted by Entwistle, Skinner, Entwistle, and Orr (2000), in North America, the school education literature prefers such terms as 'beliefs' and 'knowledge', while the higher education literature uses the term 'conceptions' to indicate "researchers' ways of describing different aspects of teaching" (p.8). For example, Freeman and Richards (1993, p.194) referred to conceptions as systematic "bodies of knowledge about an object, idea, or phenomenon", based on which they identified scientifically-based, theory and values-based, and art or craftoriented conceptions. The term 'conception' in this sense reflects the epistemological belief of how knowledge grows and teaching should be conducted. In contrast, according to Entwistle et al. (2000), in the European literature, the term 'conception' is commonly employed in higher education to denote "teachers' own ways of thinking and their beliefs about teaching" (p.8).
A popular definition of conception in adult education is described by Pratt (1992). Accordingly, "conceptions are specific meanings attached to phenomena, which then mediate our response to situations involving those phenomena" (p.204). Pratt (ibid.) further explained that a conception consists of actions, intentions, and beliefs, together forming a dynamic relationship as presented in Figure 1.
Following Figure 1, one's beliefs, intentions and actions are intertwined, and together constitute one's interpretation. Thus, by examining beliefs in conjunction with intentions and actions, a conception can be unveiled. Pajares (1992, p.320) states that "beliefs and concepts are central to a conception". Benson and Lor's (1999) distinction between conceptions, beliefs, and approaches also aligns with Pratt's conceptualization. Accordingly, a belief can be investigated by data interpretation, while conceptions "call for a further level of analysis" (Benson & Lor, 1999, p.464). Conceptions and beliefs are translated into instructional approaches employed by teachers in a specific context. Therefore, based on the interaction between belief statements, intentions, and actions, teachers' conceptions can be identified (Pratt, 1992).
In language teaching, a similar conceptualisation of the term is found. To Tsui (2003), teachers' personal beliefs, assumptions, metaphors, images, and values constitute their conceptions. These conceptions, she contends, have an interactive effect on their teaching and development. This way of conceptualizing teacher conception is also employed by other researchers (Farrell & Lim, 2005;Freeman, 1991;Mangubhai, Dashwood, Berthold, Flores, & Dale, 1998;Sato & Kleinsasser, 1999;Shi & Cumming, 1995).
Teachers' conception in the current study is, therefore, defined as teachers' personal ways of making sense of new information or knowledge. It can be reflected in teachers' personal established beliefs, pedagogical intentions, and classroom actions. Teachers' conceptions interact with their teaching context and classroom experiences (Borg, 2006).

Language Output and Interaction in Second Language Acquisition
Researchers have recognised the indispensable role of language output and interaction in fostering second language acquisition (Long, 1996;Pica, 1994;Shehadeh, 1999;Swain, 1995). Language production, both noninteractively and interactively, have recognised values. In the light of Output Hypothesis (Swain, 1995), second language learners can test out their assumptions about the target language rules, become fluent, and extend their interlanguage discourse (Ellis, 2003;Skehan, 1998;Swain, 1995). When pushed to produce language, learners become aware of "the relationship of the forms and rules to the meaning they are trying to express" (Swain & Lapkin, 1998, p. 69). Furthermore, the Interaction Hypothesis (Long, 1996) states that through interaction, learners negotiate for meaning and receive feedback which pushes them to modify their language in such a way that makes it comprehensible. In this way, learners actively generated comprehensible input, which is useful for second language learning (Long, 1996;Markee, 1997). Long (1996) stressed the connection of interaction, comprehensible input, and output with selective attention as the facilitative conditions for SLA.
Convergent empirical evidence in SLA research has testified to the relationship between output, interaction and second language learning (Ellis & He, 1999;Ellis, Tanaka, & Yamazaki, 1994;Gass & Torres, 2005;Izumi & Bigelow, 2000;Mackey, 1999;Pica, Young, & Doughty, 1987). After conducting a meta-analysis, Kech, Iberrishea, Tracy-Ventura, and Wa-Mbaleka (2006), for example, concluded that interaction has some effect on the acquisition of certain lexical and grammatical features, especially when tasks force the use of essential features or elicit useful language for task completion.
Pedagogically, Ellis (2005) proposed ten crucial principles for instructed language acquisition. Two of these principles include creating opportunities for output and interaction in the target language. Proponents of TBLT also place a focus on communication and language use as the ultimate goal and principle of learning and teaching (Foster & Skehan, 1999;Prabhu, 1987;Skehan, 1996;Skehan, 2003;Swain & Lapkin, 2000;Willis & Willis, 2007).
In summary, theoretical, pedagogical and empirical arguments have confirmed that language output and interaction are potential facilitative conditions for second language learning. Therefore, from the SLA perspective, ESL and EFL teachers should engage learners in language use practice (Ellis, 2005), particularly when exposure to the target language is too inadequate to support natural language acquisition (Green, 2005). These theoretical positions, however, need to be examined against teachers' existing beliefs and practices.

Research on Teacher Conceptions about SLA Principles
Studies that explore the effect of SLA and methodology training on teachers' uptake of SLA principles generally point to a limited change in their beliefs about SLA. Mattheoudakis (2007) found that Greece EFL pre-service teachers changed some of their beliefs significantly after a three-year programme. Most of the teachers, for example, changed their belief about focusing on grammar knowledge, which they strongly held. They also thought teachers should correct all errors made by beginner learners, but this belief became obsolete. In Hong Kong, Peacock (2001) investigated 145 ESL pre-service teachers learning in a three-year programme, and discovered some changes in their SLA beliefs; but the majority of teachers still held on to the view that learning a language means mastering vocabulary and grammar rules. Likewise, MacDonald et al. (2001) detected a shift in only some beliefs of 55 TESOL non-native speaker undergraduate and postgraduate students after an SLA course. Overall, they found that the teachers significantly moved away from the behaviourist view, particularly from viewing language input as grammatical knowledge. Nonetheless, these teachers remained doubtful of the benefit of learner-learner interaction to their language learning. MacDonald et al. (2001) attributed this reluctance to the teacher-fronted teaching culture to which these students were accustomed before attending the training.
Such inadequate trust was also expressed by the six Thai EFL instructors in McDonough's (2004) study. They were concerned that interaction cannot push learners to modify their linguistic output or provide useful feedback to peers, and that learners produce less accurate TL forms during interaction. Pair work and group work, for them, were the opportunities for students to practice certain targeted language items, believing that students should focus on intended forms during practice. They also reported some factors which constrained their practice of promoting interaction, namely how to manage and monitor interaction in large classes with fixed desks, and the need for preparing students for high-stakes tests. In a different way, Howard and Millar (2009) responsively introduced Ellis's (2005) ten instructed SLA principles to fifteen South Korean EFL teachers attending a four-week professional development course in New Zealand. The researchers found that the Korean teachers lacked the confidence in using English proficiency to conduct class activities in English. Although all the teachers viewed interaction as an important principle, only two of them reported implementing interaction in their classrooms. Inadequate training, students' limited proficiency and L1 overuse, large and mixedproficiency classes, and limited instructional time were the major reasons why most of them rarely generated opportunities for output and interaction. The biggest challenge involved the university entrance examination, which focused on grammar and reading rather than communicative skills. These findings, in general, resemble what McDonough (2004) reported.
In Vietnam, scant research has examined teachers' conceptions of SLA principles. Some studies reported findings relevant to the current issue under investigation. Chau (2009), with a survey of 106 EFL university instructors, revealed that the teachers gave more priority to grammatical accuracy than communication skills. Le (2008), in a different way, focused on high school teachers' beliefs regarding the new English task-based curriculum, an approach that concentrates on communication as well as language output and interaction. He concluded that the teachers' beliefs and practices remained the same. The teachers still focused on grammar and reading skills, which were to prepare their students for examinations. Case studies such as Barnard and Nguyen (2010), Nguyen (2014), and Nguyen, Le, and Barnard (2015) further confirmed that Vietnamese EFL teachers had a strong belief in teaching grammar, which restricted their focus on meaning and communication as required by task-based language teaching (TBLT).
The studies reviewed above provide a general understanding of EFL teachers' conceptions of SLA principles. However, most studies (e.g., MacDonald et al., 2001;Mattheoudakis, 2007;Peacock, 2001) concentrate on preservice teachers, subscribing to an etic perspective, which stresses a straightforward causal effect and treats participants as "objects rather than subjects" (Kalaja & Barcelos, 2003, p.2) without considering the mediational effect of teachers' cognition. Teachers' conceptions and beliefs are strong mediators of their practice and change.
A study by Howard and Millar (2009) relied on a context-responsive approach to teacher development to impact change, but the study employed interviews and questionnaires without classroom data. These methods fail to consider the correspondence between teachers' self-report and their real classroom practices (Breen, Hird, Milton, Oliver, & Thwaite, 2001). Although the case studies (Barnard & Nguyen, 2010;Nguyen, 2014;Nguyen, Le, & Barnard, 2015) have contributed some insight into Vietnamese teachers' responses to TBLT, further research using multiple methods is indispensable for gaining further insight. The current study relied on a bottom-up approach to working with EFL teachers at a workshop that focused on output and interaction, following them in their classrooms to record their lessons and interview them. From this, the study sought to answer the following question:

Method Design
The present case study employed multiple methods to tap into the Vietnamese teachers' conceptions, following their self-report on group discussions, lesson planning, teaching, and stimulated reflections.
As a case design, the study relied on the working concept of cases as "specific individuals, particular events, processes, organizations, locations, or periods of time" (Stake, 1995, p.2). It aimed to examine a single instance in detail, as this may be useful to predict recurrent characteristics of human systems (Sturman, 1994), and extend the observation to a wider population (Cohen, Manion, & Morrison, 2000). Contemporariness, real-life context, and boundary are three essential characteristics of a case study (Yin, 1994). Despite the blurred and indistinctive boundary of what a case is (Yin, 1994), as well as the tension in identifying uniqueness, complexity, and patterns, David (2006) contends that the power of case research is an in-depth exploration and description. In a case study, the results are not statistically but theoretically generalizable. According to (Yin, 1994, p.10), evidence from cases can serve to modify or develop theory, thus being generalised as "theoretical propositions" rather than "populations or universes".
In the present study, six individual cases of teachers are treated as subcases within a specific institutional case, a representative university. They serve dual purposes: to compare and reveal a holistic description of teacher conceptions, and to illustrate the meanings the teachers attached to language output and interaction in their specific context of General English (GE) classrooms. The study aimed to gain insight into how SLA precepts can be put to practice, a way of informing the gap between theory and practice. A combination of multiple data sources and cross-case analysis aimed to uncover "the complex "hidden side" of teachers (Freeman, 2002).

Instruments
Three main instruments were used to tap teacher conceptions: focus group interviews, lesson plans and quick interviews, and stimulated recall interviews after teaching.

Focus Group Interviews
Focus groups were selected because this method can elicit experiences, attitudes, and opinions (Wilson, 1997). A focus group discussion was conducted to obtain teachers' initial understanding of output and interaction before the teachers wrote lesson plans and had their lessons video-recorded. A discussion guide was provided for two small groups of three teachers, an ideal number to keep the discussion focused (Cohen, Manion, & Morrison, 2000). Each group discussion lasted around 20 minutes and the teachers' discussions in English were video-recorded. The following questions were posed: 1. What does the term 'output' mean to you? 2. What role does it have in language learning? 3. What is good interaction? What role does it play in language learning? 4. To what extent do you create opportunities for output and interaction in your lessons? Why?

Lesson Plans and Quick Interviews
Lesson plans and interviews were employed to document the teachers' intentions and beliefs about output and interaction. They served as the initial door to look into the teachers' conceptions. A quick individual interview was conducted individually with teachers before their lesson was taught, aiming to clarify the lesson's procedure, activities, and objectives.

Stimulated Recall Interviews Following Video-Recording
Stimulated recall interview (SRI) was employed to engage participants in viewing their lesson recordings, and then recalling and evaluating their actions. This introspective method is suited to elicit data about "thought processes involved in carrying out a task or activity" (Gass & Mackey, 2000, p.1). The issues of reliability and validity in stimulated recall data were addressed. Following Gass and Mackey's (2000) suggestions, to reduce risks of forgetting and ensure 95 percent accurate recall at least, the stimulated recall interview was conducted less than 48 hours after the lesson. The effect of prompt questions on teachers' recall was addressed by designing prompts that focused on their reflection at the time a classroom event happened (Gass & Mackey, 2000) (e.g., What were you thinking? What was in your mind at that time?). The interview schedule (see appendix) was based on a sample of Mackey, Gass, and McDonough (Cited in Gass & Mackey, 2000) to assist the interview process. Each interview lasted around one hour and was conducted in Vietnamese by the researcher. The interview was conducted in an arranged empty classroom so that the teachers felt secure. The teachers watched their recorded lessons on the researcher's laptop. While reviewing the lesson, the researcher paused at classroom activities related to output and interaction, and used the interview protocol (see appendix) to prompt the recall.

Participants
Purposeful sampling (Glense, 2005) was applied to select the participants with different teaching experiences, ages, and degree qualifications. Six lecturers from the Department of English of a university in the Mekong Delta of Vietnam agreed to participate. As shown in Table 1, these teachers' workloads for General English (GE) classes varied because they taught students of various majors and of different levels (GE level 1: 45 hours, GE level 2: 45 hours, and GE level 3: 60 hours). They had a range of teaching experiences from less than one year to 12 years. Three of the teachers held a Master's degree: one TESOL, one American Studies, and one Educational Studies. All of them had completed a Bachelor's programme in teaching English as a foreign language.

Data Collection Procedure
Invitation letters were sent to nine lecturers in the Department of English of the university where the researcher is working. Six of them agreed to participate. A meeting was set up with them to talk about the project in detail and consent forms were signed. Following that, a focus group discussion was conducted in the first workshop about SLA facilitating conditions. They were divided into two groups, and each discussion was recorded with a digital recorder. After the workshop, the teachers planned and taught a lesson that incorporated their views about language output and interaction. Before teaching the lesson, they were asked to clarify their intentions in a quick interview. While the lesson was taught, the researcher recorded it with a handheld camera. To avoid disturbing them, the researcher sat in a corner, using the camera's zoom function. One day after the lesson was taught, a stimulated recall interview was conducted.

Data Analysis
The SRIs were transcribed and translated into English by the researcher. This allowed for a more precise translation because of the researcher's familiarity with the contextual information. Feedback on the translation was obtained from a native speaker and a colleague so that readability could be improved and information about the context could be fully included. Upon obtaining the feedback, the researcher returned to the recordings to check the information reported and fixed the English translation to ensure a match of representations of information.
Data analysis followed the procedure suggested by Braun and Clarke (2006). Data were initially coded with short descriptions and then reduced into themes. A cross-case comparison of themes was made to obtain a holistic view, and individual cases were examined to gain further insights and develop thick and detailed descriptions. Lesson plans were described in terms of objectives and activities, and documented into tables. A triangulation

An Initial Product-Oriented Conception
The data from focus group interviews showed that all the teachers initially held a perception of output as language production, viewing it as an end-product rather than a process to facilitate language learning. For example, Teacher 4 stated, "The language output is what students can use, can produce after a lesson". Likewise, Teacher 1 said output means "How much progress students make in picking up the language".
The teachers also identified features of good interaction congruent with the literature. All of them viewed effective interaction as two-way communication or information exchange. Teacher 2 said, "If they do not understand, they should ask, and by asking questions to be clarified, that is a good interaction." Besides recognizing negotiation for meaning as the key feature of good interaction, Teacher 1 added effective interaction a communicative goal: "They can achieve the aim that they want." Teacher 6 and Teacher 4 also perceived interaction as an essential condition for peer feedback and learning. Teacher 4 clarified: I think good interaction means that students can work effectively with their partners and they can learn from their partners, and their partners can find out their mistakes, adjust themselves, and correct their mistakes themselves.
All the teachers thought that interaction is required for English learning. Teacher 4 emphasised that, "…there must be interaction in the classroom. Interaction helps students to produce or to perform what they have studied; it is to prepare them for language performance outside the classroom".
Most of the teachers agreed they regularly created opportunities for students to practice language output and interact with one another. For instance, Teacher 4 used role cards to encourage students to make a conversation. Teacher 2 employed discussion before her students did main reading or listening tasks. Besides, Teacher 5 shared her way of providing output practice from a controlled to a free manner: With controlled practice, I give students handouts, some information gap activities or some charts, so that I can check at least that they can use English under the teacher's control, and then move on to free production. I provide many choices, many questions, or many prompts in order for the students to be able to talk about what they want, based on their acquisition.
Only Teacher 1 said she rarely engaged GE students in group or pair interaction in the classroom because of their limited and mixed-level proficiency, willingness to participate, and especially the time restriction. Teacher 1 explained that with only 30 or 45 hours, with that number of lessons [ten units], she found it hard to create interaction in the classroom. Teacher 2 also reported that for output like writing she usually asked her students to do it at home because the activity is time consuming.
In general, the focus group data showed that all the teachers initially shared a product-oriented conception of language output. They understood the nature of interaction, and believed language output and interaction are important for language learning. However, some contextual factors could affect their practice such as time restrictions and students' characteristics.

Conception of Output and Interaction in Practice
An analysis of the lesson plans and stimulated recall revealed some intentions and actions across four major themes: target linguistic content-oriented, language accuracy-oriented, institutional restrictions, and student characteristics.

Target linguistic content-oriented
Target linguistic content orientation describes the teachers' control of language production activities in such a way that focuses on particular language items they were teaching, although the way output and interaction were manipulated varied slightly among the six teachers. Their lesson plans partially uncovered their intentional focus.
As shown in Table 2, oral output and interaction was implemented in all the lessons, whereas written output was also given in Teacher 1's and Teacher 2's lessons. Most of the lesson objectives, however, showed the production of a particular TL content. Teacher 3, Teacher 2, and Teacher 6 clearly gave a priority to the linguistic objective. Teacher 1's plan put the topical content on top, and the linguistic objective followed as a support, which was confirmed in the lesson plan interview: "The grammar use [final objective listed] supports all the output parts above." Teacher 5's planned focus appeared to resemble Teacher 1's. Teacher 4 did not specify any objectives. -Talk about how they [students] get to school/work and the trip they have just taken.
-Listen to some people's talks on travelling.
-Write a paragraph describing how they get to school/work/ travel.
-Use grammatical points such as articles, present tense, or past tense in both writing and speaking. The teachers' intentions were further revealed in the way they organised output and interaction activities in Table 3  Table 3 Procedure

Teachers Lessons Lesson procedure
Teacher 1 How do you get to work? (Level 2) -T introduces vocabulary on transportation and travel.
-Ss practice in pairs, asking and answering questions on how to get to school.
-Ss read a passage on transportation.
-Ss notice the use of articles in the passage.
-T explains and Ss complete practice exercises in the book.
-T introduces vocabulary about vacation activities.
-Ss listen to people talking about their vacations.
-Ss talk about their vacations in pairs. -Ss write a paragraph describing how they go to school/work or their vacation.
Teacher 2 I'm going to save money. Table 3 further shows that the lesson procedures of Teacher 3 and Teacher 6 were aligned with their specified aims, while those of Teacher 1, Teacher 2, and Teacher 5 were not. In contrast, in Teacher 1's plan, the language focus was ordered before the language output activity, which was incongruent with her objectives reported in the lesson interview. Teacher 2's lesson apparently proceeded in the way that prioritised meaning expression before the simple past tense was taught, which differed from the objectives listed (Table 2). Teacher 5 clearly indicated a focus on teaching target linguistic forms prior to providing output practice in the subsequent activities. The target forms included 'self' and common expressions used at a shop. In contrast, Teacher 4's procedure spelled out her intention to prioritise meaning; that is, students' speaking about New Year's resolutions and plans for weekends.
Further evidence from the SRIs uncovered a form-oriented linguistic performance. In Teacher 3's video-recorded lesson, the form 'there be' was explicitly instructed first. Then a transformation drill was used to develop students' automatic processing and accurate production of the form. A question-answer drill followed to further prepare the students for an information gap task later that was provided as an opportunity for contextualised practice. The teacher explained that the last task was to evaluate whether the students used the structure accurately. She recalled: My purpose at this stage was to give them practice on how to make and answer the questions. Therefore, this was the preparation for them to do the production stage later on. It was a bit inauthentic but useful to them, I think, because they repeated the structure, and they would memorise that structure. (Emphasis added) Likewise, Teacher 2 provided opportunities for language production late in her lesson. After presenting the structure 'be going to', she asked her students to write down their own weekend plans, and then move around to share this with their classmates orally. For Teacher 2, these activities allowed students to reproduce the language, or "apply the language they had just learned to talk about their weekend." She explained that this was aimed at helping her students be able to use 'be going to' to talk about their weekend plan, which was in line with her lesson plan.
For Teacher 1, the first language practice, the ask-answer practice on means to go to school, was the opportunity when "they can make a small conversation; for example, they would ask about how to go, how long it takes, why they like to take a certain transport, or so on." This explanation reflects a linguistic focus on certain questions. A more authentic task, she added, was students' talking about their own vacations in pairs, which represents the same linguistic focus. She recalled: …in pairs, they more or less had a chance to speak English, but in terms of using these questions in real life, I am afraid that this activity was not enough….So in the next lesson, I planned to give them another situation where they will role play going to a travel agency to ask, for example, if they want to go to Dalat City, they will ask 'How can I go there?' or 'How far?' or 'How much?' so they can know how to ask questions. (Emphasis added) Teacher 6 provided two language practice activities. The first one was an interview with cue cards where students acted as journalists interviewing the daily routines of a celebrities, followed by a brief oral report. In this activity, she recalled, "students can both ask and answer and finally come to the same outcome," an opportunity for them to practice the target language items taught earlier in the lesson. The second activity was the students' interview about their daily routines where they practised using the questions they had studied, "First, they asked about the activities, asked what, and second how often. Those who asked would learn how to ask, and those who answered would use adverbs of frequency, and then focusing on time, they would use prepositions of time to answer questions." (Emphasis added) However, Teacher 6 believed language should bear a resemblance to real life, instead of what the textbook provided. This urged her to introduce the input close to her students' daily routines. For her, this activity generated more authentic language use because it freed students so that "they could choose to ask and answer about their daily activities". Her explanation evidently revealed the need for teaching certain language items targeted in the lesson, but she adapted to create a meaningful language practice.
Overall, there was obvious evidence that the teachers had a tendency to gear language output and interaction activities towards the linguistic content specified in the textbook in the control-free process. This predisposition is linked to their concern for controlling students' language use to impact accuracy.

Language Accuracy-Oriented
A concern for language accuracy was observed among the six teachers. It was manifested in a variety of ways they ensured students' accurate performance. Table 3 shows that most of the teachers delayed 'free production' until the end of their lessons. In practice, the teachers initially taught students the vocabulary items or grammatical structures necessary for their language output later. Presentation was stressed as being essential for students as it can give them various sources of language input. Teacher 5 said, for the usual utterance 'I need', students can be taught to say 'I am looking for or I'd like". Teacher 6 revealed this belief when she reviewed some grammatical patterns to guarantee her students' accurate language production: "I wanted them to do it [make questions] by themselves before they did the task because I heard some students asking inaccurate questions". (Emphasis added) The second way to ensure accuracy involved pronunciation practice before pair work practice. This helps students develop both fluency and accuracy in using the target forms in subsequent output activities. Teacher 2 explained why she had students repeat some words aloud, "I wanted to prepare them for the speaking activity". Teacher 6 shared the same worry in her recall session, "I wanted to ensure they had correct pronunciation before they produced output".
Another example lies in teachers' reasons for initiating interactions with their students. Teacher 2 viewed her intention to interact with her students not as a way to maximise learning opportunities for students, but as a classroom management strategy. She commented on an exchange with some students after their pair work talking about daily routines: "It was not to find out about their peers' routines, but to check who had worked and whether they could use accurate grammatical structures or pronounce an accurate final -s." (Emphasis added) Teachers' authority is manifested in their control of how students wanted to use the language. The power distance in the cultural context of Vietnam (Hofstede, 2019) has entitled Teacher 2 to dominate her students and forced them to stay on the track she intended. She recounted: "This morning I crossed out a sentence a student said. I wrote a clear model introduction: 'Hello, everyone, I would like to tell something about a US holiday.' The student began with 'Today I introduce to you a holiday.' I wrote it on the board, crossed it out, and told him never to use it again." In summary, although the six teachers exhibited an initial understanding of the merit of language output and interaction, a common thread among them was their manipulation of lesson activities to target the accurate production of particular planned linguistic forms. Such practice reflects a conception of language accuracy and form-oriented teaching and learning, and is further mediated by various contextual constraints that will be discussed below.

Institutional Constraints
The way the teachers implemented output and interaction was restricted by the institutional syllabus restriction, which was based on the textbook. Their lesson plans represented the textbook units they had to teach. Evidence from lesson interviews further showed that the prescribed content somehow influenced how they taught despite its relevance to the students' lives. Teacher 1, for example, recounted, "From the title 'How do you get to work?' when you introduce the lesson, you can ask this question, so the students would have to notice the question and how to answer it. The question lies in Exercise 3. I turned this exercise into an oral practice." Time restrictions created difficulties for optimising output and interaction in the GE classroom. According to the GE syllabus, ten units had to be completed within forty-five standard hours, each lasting 45 minutes. Thus, a particular activity like sharing opinions about some statements in pairs was assumed to be potentially timeconsuming and was replaced with a teacher-fronted discussion by Teacher 2.
Before listening, there is a discussion section 'What do you think about these statements?' so I prepared them for this discussion. Here in the book there was 'work in pairs' but I did not do that... If I had had time I would have let them do that.
Also, because of the time limit, accuracy emphasis right from the beginning was thought to be effective. Teacher 2, for instance, recalled a moment in her lesson that "My purpose was to let them listen first, so that they could imitate the conversation, so they could be correct from the start. If I had let them think and do the work by themselves, they would have used Vietnamese or asked me and it would have wasted time."

Student Characteristics
The teachers perceived student characteristics as a barrier to fostering language output and interaction. These characteristics involved students' limited and mixed levels of English, students' passive attitude to learning, and students' failure to initiate communication.
First, students' limited and mixed levels of proficiency made the teachers cautious. As mentioned, all the teachers had a strong belief in language instruction for language performance, mostly manipulating student production by giving a model and delaying activities for free language production until the later phases of a lesson. Teacher 1 said in the recall, "For this level [GE], if you do not give a model and just give a task and ask them to talk... even though I gave them a model, as you see, their output was so bad". The teacher's orientation towards language output accuracy reflects a cost-benefit analysis view in teaching English to mixed-ability classes. Teacher 2 expressed a similar mind-set, "You will become very tired of correcting them. Here they were all prepared, and they just needed to speak." She expressed her strong belief that without a dialogue model, they would have made mistakes. This concern represents a product-oriented perspective on learning where trialand-error is discouraged. She continued, "They did not have to try and make mistakes. I was afraid they would make mistakes… For GE students I think we should introduce the correct form right from the start, so they could follow the track." (Emphasis added) Students' unfamiliarity with active learning was an additional reason. This implied a common assumption that Vietnamese students in GE classes are often passive learners. Teacher 5 commented, "If they had been more active, I would have asked them to talk in front of the class and make their decisions". Teacher 2 complained, "They just listen and wait to be asked to give an answer, but they seldom ask questions." Teacher 1 described it as "the students' culture of learning, an established study habit that was still not learner-centred". With such a learning culture, Teacher 1 said, " [Teachers] have to give them a model and then give them time to practice," or "teachers need to scaffold lessons and can't allow students complete freedom." This culture is related to students' lack of competence in handling a conversation. Teacher 2 attributed this to the culture where "people do not like to ask questions", and stressed that "students can't communicate or start a conversation, and they will get confused and begin to speak Vietnamese". Such a perception might have induced the teachers to model interaction for their students. For example, Teacher 5 said "I began to ask sample questions, so they would know how to maintain a conversation".
Overall, most of the EFL teachers in the current study had a tendency to control output and interaction activities mainly for the desire of accurate performance. Their responses reflected a language form and accuracyoriented teaching view. Factors in the educational context, including both institutional factors and students' characteristics, mediated their conception.

Discussion
Unlike the studies of MacDonald et al. (2001) and McDonough (2004), the current study showed that the six Vietnamese teachers believed in the importance of learner language production in second language learning, a similar outcome reported in Howard & Millar's (2009) study. The teachers' tendency to manipulate output and interaction activities for the mastery of certain targeted linguistic content is in line with previous research in a similar context (McDonough, 2004). Although the Vietnamese teachers did not believe that peer interaction could lead learners to make errors as their colleagues did in McDonough's (2004) study, they implicitly signalled disbelief in the contribution of student talk in pairs and groups, given their mind-set of language output accuracy.
To foster learner output and interaction, most of the Vietnamese teachers created activities for contextual and communicative practice of certain targeted language items, the kind of activities resembling what Ellis (2003) called linguistically focused tasks designed to support meaningful language practice. Such teaching practice is depicted as a weak approach to using tasks (Skehan, 1996), which is favoured by many Asian teachers in Hong Kong primary schools (Carless, 2003(Carless, , 2007 and the Thai tertiary classroom context (Todd, 2006).
The conception of the teachers in the current study described above represents Vietnamese EFL teachers' belief in grammatical accuracy over communication skills (Chau, 2009). It also resembles what previous research on curricular innovations regarding the necessity for explicit grammar teaching (Barnard & Nguyen, 2010), and preparing students linguistically before task performance (Nguyen, Le & Barnard, 2015). Such established practice that builds around particular language forms reflects a deep belief about form-focused instruction, and a lack of focus on meaning aligned with TBLT (Nguyen, 2014).
Conceptions of learning have a correlation with and underpin instructional approaches (Goodyear & Hativa, 2002;Lamb & Kember, 2006;Richards & Rodgers, 2001), and conceptions of learning are related to conceptions of language teaching (Benson & Lor, 1999;Freeman & Richards, 1993). Therefore, we can infer the conception of the Vietnamese EFL teachers about English learning and teaching based on their conception of language output and interaction. In this way, a synthetic approach (Long & Crookes, 1992), a reflection of a productoriented conception of teaching, could better capture these teachers' views. It represents the perspective of language pedagogy primarily oriented towards the presentation and practice of discrete language items or forms, as opposed to communicative meaning or fluency-oriented pedagogy (Wilkin, 1976). It can also be likened to focus-on-forms (Ellis, 2001), although their conception is not entirely structural at the expense of communicative meaning. The orientation of these teachers is likely to have its roots in their established experiences in teaching GE as well as professional backgrounds. It is further shaped by contextual constraints, namely institutional constrains and student characteristics. Such a conception seems to compromise the use of communicative tasks in that language production opportunities serve to prime the internalization of TL forms.

Conclusion
Despite its limited generalizability, the study offers useful implications. Considering that language pedagogy needs to investigate teaching conceptions to benefit language teacher education and development (Freeman & Richards, 1993), it is imperative that teachers' preconceptions, which could be shaped by prior training or classroom experience, be tackled and brought to explicit discussion and reflections. In that way, theory and practice could be brought closer. The conception of Vietnamese EFL teachers in the current study uncovers a form-focused perspective on language teaching, which is reflected in the present-practice-produce model they are familiar with. This conception seems to fit into their teaching context. Thus, for bottom-up approaches such as TBLT which advocates a constructivist and trial-and-error view of learning to be incorporated in an Asian context like Vietnam, this teacher conception should be addressed if the gap is to be narrowed.
In conclusion, the current study employed multiple sources of data to unpack what six Vietnamese EFL instructors believed about language output and interaction. The case-based thematic analysis revealed that the teachers held a conception of language teaching in the light of skill building theory (McLaughlin, 1987). This empirical evidence is corroborated by similar studies in other Asian contexts. The finding implies that the Western idea about a flexible communicative approach with a strong focus on meaning and a constructivist view of learning needs to work in tandem with the conception of learning and teaching in the light of skill acquisition and building in EFL settings. In these settings, teaching and learning are not supported in terms of an authentic language environment, institutional imperatives, and students' characteristics compatible with a constructivist learning approach. The limitation of the current study is the small sample size biased towards gender. Further research could employ a larger sample which involves a wider range of variables.