Triadic Dialogue in EFL Classroom : Embedded Extensions

Students learn languages through talking and there is a documented need for more student talk in the classroom. Through talk we learn not only structural components of a language but also the communicative application of it. Can standard classroom speaking strategies, embodied in typical predictable patterns, successfully serve these functions? And to which extent should those traditional patterns allow predictability and control in managing classroom interaction? In this paper the focus is driven towards the ways the classroom teacher can orchestrate and support a kind of classroom discourse that engenders more active student talk that leads to foreign language learning. A particular emphasis is put on the use of the Triadic Dialogue, known as IRF (initiation-response-follow-up) pattern, the value of which has been debated in writings on language education. It has attracted criticism for being ritualistic and restrictive, although recent research has pointed to the range of functions that may be fulfilled by the follow-up move. The paper examines the constituents and possible sub-genres of the three-part classroom exchange and aims to prove that a certain degree of freedom is possible within the constraints of the Triadic Dialogue. Drawing on recorded episodes of teacher-students interaction in adult EFL classroom, the paper will show that the three-part pattern allows spontaneous variations and extensions initiated both by teachers and students. The variety of forms that the basic IRF structure can take enriches the linguistic repertoire of choices in the co-constructed classroom reality.


Introduction
It is now generally accepted that proficiency in another language requires more than knowledge of a linguistic code (grammatical competence) (e.g., Cazden, 2003;Thornbury, 2005;McCarthy, 1991; not to mention the Common European Framework of Reference (CEFR) for Languages: Learning, Teaching, Assessment, 2001).Communicative competence entails the ability to interpret and enact appropriate social behaviours and requires the active involvement of the learner in the production of the target language (Boyd and Miller, 2000, p.165).Thus the learner needs to do more than supply one-word answers in the target language or recite isolated sentences.The learner needs to be actively engaged in constructing and clarifying meaning.Students can learn through talk and students can learn about the target language and through the target language by producing it.
The classroom context, in which a foreign language learning process is taking place, is recently viewed as defined by two sets of major constraints.On the one hand, as in any institutional discourse setting, such as courtroom or doctor's surgery, verbal behaviour of the participants in the classroom is goal-oriented and governed by certain rules.The rights and obligations of the participants, as well as their role relationships, are culturally accepted and prespecified by instructional goals stated in the curriculum and determined by other requirements of formal educational contexts in the era of high-stake testing.In typical classroom communication, the teacher is in control of turn-taking and topic management, and usually follows a plan of actions that aims to achieve certain instructional objectives (Byram, 2001, p.111).
On the other hand, language teaching is a relatively unpredictable phenomenon, co-constructed and defined by the participants.Language teachers work with people in specific situations at a given point of time.Institutional rules and restrictions do influence the reality of teaching but they are not full reality; they are simply elements of the complex dynamics of teaching and learning.Van Lier (1997 quoted in Tudor, 2001, p.9) refers to this as an 'ecological' perspective to teaching.In this view, learners are treated as active agents who, along with their teachers, engage in a form of conversational discourse that aims for the enhancement of understanding rather than the one-way transmission of teacher-directed instructional talk (Wells and Haneda, 2005).In this case the target language becomes the vehicle for communicating ideas rather than the means for instruction.
To combine the ritualistic and the spontaneous features of classroom discourse, the term "instructional conversation" (or "instructional dialogue") has been coined (Wells and Haneda, 2005;Leinhardt, 2005).As a notion, IC contains a paradox: 'Instruction' and 'Conversation' appear contrary, the one implying authority and planning, the other equality and responsiveness (Tharp & Gallimore, 1988 quoted in Wells and Haneda, 2005, p.151).
The questions to be addressed here are: Can this paradox be resolved?And, if so, under which conditions? Can the evolved with time structural patterns such as IRE (Initiation, Response, Evaluation) that were used mainly for instruction allow enough flexibility for conversational purposes?Is it overall realistic to expect conversation to take place in the classroom?http://dx.doi.org/10.5755/j01.sal.0.18.416 Hence the aim of the present study is to investigate if and how the traditional IRE structure (the Triadic Dialogue) can be extended and to see whether these extensions permit freer conversation-like teacher-student interaction.
The objectives of the paper are as follows: to analyse scientific literature related to the Triadic Dialogue; to find and investigate examples of extended Triadic Dialogue in the data recorded in adult EFL classrooms; to determine the constituents and sub-genres of the extended structure; to see whether there are extensions initiated by students; to draw conclusions and to provide recommendations based on the literature analysis and the investigation.
The research methods include indirect observation (recording) of lessons in adult EFL classrooms and qualitative analysis of the collected data using Discourse Analysis and Conversation Analysis approaches.
In Lithuania, Discourse Analysis is a relatively new field of scientific studies.Some researchers have turned to the areas of social discourse, political discourse, Internet discourse, general educational discourse.Both teacher and student variations on the Triadic Dialogue, and studentinitiated extensions of the structure are under-explored, let alone in the EFL teaching context.

Literature Overview
The IRE cycle is a teacher-led, three-part (Initiation, Response, Evaluation) sequence that begins with the teacher asking a student a question or introducing a topic for the purpose of finding out whether the student knows the answer.In the IRE pattern, the student's answer is evaluated by the teacher.Then the interaction ends.The pattern was first identified by Sinclair and Coulthard at the University of Birmingham (1975).Here is an example from Sinclair and Coulthard's analysis of an interaction within a lesson (Coulthard, 1992, p.2): Sinclair and Coulthard (1975) analyse this part as an eliciting exchange, since its primary function is to elicit information.The exchange is made up of moves: the first utterance in line 1 by the teacher is an initiating move; the pupil then makes a responding move in line 2; and the teacher then makes a follow-up move in line 3, when s/he repeats the pupil's utterance to confirm that this was the answer s/he wanted.S/he then goes on to make another initiating move which constitutes a new exchange.Moves are made of acts which are defined by their function.In line 1 the teacher says 'yes', which could be classified as a nominating act because the teacher clearly chooses a pupil to answer.In line 3, the same word 'yes' has a different function and could be classified as a confirming act because like the repetition of the pupil's utterance by the teacher, it serves to indicate to the pupil that the answer is acceptable.
This labelling of discourse units has proved very helpful but it is also problematic: there seems to be no particular justification for some of the labels chosen.As Mills (2004, p.142) points out, the very fact of naming a discourse unit means that you have already decided that the unit has one particular function.Sinclair and Coulthard (1975) in their early work developed a very limited number of terms to label discourse units, but over time these terms increased in number, so that it became very difficult to decide which function a particular unit had.Thornbury (2005, p.72) gives the following joke, in which the student fails to recognize the function of teacher's follow-up: Teacher: What's the protective outer layer of the tree called, Tom?
Tom: I don't know.
Teacher: Bark, Tom.Bark! Tom: Woof, woof!The Birmingham model is certainly not the only valid approach to analyzing discourse but it is a relatively simple and powerful model (McCarthy, 1991, p.12) which tries to capture the larger structures of discourse.IRE was later recast into Triadic Dialogue by Lemke (1990 quoted in Gourlay, 2005, p.404).The three-part structure's rigid pattern is valid in traditional, teacher-fronted classrooms where teachers and students speak according to very fixed perceptions of their roles.Conversation in this case is very one-sided, with the teacher asking all the questions and the students answering them.Moreover, the teacher questions are most of the time display/closed questions, i.e. questions that require the learners to show knowledge that is already known by the teacher.Such questions usually require only short answers and do not generate adequate language output by students.The spontaneous, two-way interactional and interpersonal features of conversation are almost totally absent from this kind of classroom discourse.Thornbury (2005, p.80) though marks that this does not disqualify it as form of discourse of its own right: this kind of 'teacher talk' has a long tradition and serves a very useful pedagogical purpose.But it can no way be considered a valid model for -or practice of -casual conversation.
Likewise, White (2003) notes that a less rigid, discussion oriented classroom setting creates difficulties in applying the IRE model.He observes that in small language classes with relaxed perception of teacher/student roles, the turntaking becomes free and results in an excess of student/ teacher exchanges without a new initiating move made by the teacher, i.e. the structure then follows the pattern of IRFRF or IRFRFRF.Nunn (2001), on the other hand, argues that teacherfronted classroom discourse, as opposed to communicative language teaching, is sometimes devalued.He agrees that due to its highly repetitive structuring in the flow of discourse teacher-fronted language is a 'ritual' antithetical to classroom discussion, but in his opinion, rituals are necessary and useful, and can serve as 'framing devices' to embody 'a repertoire of choices.' Thus Nunn allows 'spontaneous improvisations on basic patterns of interaction', taking the IRF model (where F stands for 'follow-up', as the function of this move can be not only evaluative) as a pattern.On the basis of school data corpus, he enriches the IRF structure with a new element called 'negotiation' and marks it as N resulting in elaborations of the model like INR, IRNRF, INNR and so on.He claims that teacherfronted classroom rituals do not exclude negotiated choices and that they are 'potentially useful and flexible educational tool.' Nassaji and Wells (2000) question assumptions that the prototypical function of the third turn in the IRE sequence is primarily evaluative of the preceding student turn, arguing that a wider range of options is available to teachers.They point out that in this way teachers may extend the sequence with 'dependent exchanges' until the sequence is completed with an answer accepted by the teacher.In their paper, they exemplify and explore the various ways that the teachers in their data used the followup move in order to achieve a variety of different pedagogic purposes.They discuss the variations of the third move to promote open dialogic classroom interaction, as opposed to using Triadic Dialogue as an instrument of control in the discourse.
These freer interpretations of the three-part pattern let us understand the IRF approach as a contrast to the IRE pattern defined above.Thus in this paper the following understanding of the IRF structure will be applied: IRF structure is a sequence that begins with either the teacher or student asking a question or introducing a topic.After a response is given, the initiator then uses the response to move the conversation forward.This conversation can continue for as long as the participants wish to talk about the subject, and may include contributions from many people in class (Byram, 2001).
Teachers and learners adjust their use of language according to the task they are involved in.While certain classroom activities (for example, grammar explanations, checking episodes, teaching content-based subjects) may involve very little learner participation and greater portions of traditional teacher-centred talk, others (like eliciting learner responses, discussions) may result in more active learner participation and more complex teacher-learner interaction.The important point is to be aware of the particular goal at a given moment and to be able to match the pedagogic purpose to the language in use.

Data and Methodology
Eight teachers of EFL (most of them were from an international private language school in Vilnius) agreed to take part in this study.Each was asked to make anonymous audio recordings of one of their lessons using a digital voice recorder.(Coulthard, 1977).One of the first and most apparent framing, mechanisms of classroom discourse discovered by DA analysts was the above-mentioned IRE pattern.
Conversation Analysis (CA), which by some authors is seen as just a branch of DA (McCarthy, 1991;Schiffrin, 1994), on the contrary, is claimed to be a 'bottom-up', almost non-theoretical empirical study which describes conversation in terms of turn-taking, sequences and repair practices, and observes how people orient to the demands of a particular speech event.In contrast to DA, Conversation Analysis does not impose any analytical categories on the analyst and does not attempt to 'fit' the data to preconceived notions.The classroom context, under the CA methodology, is viewed as being dynamic, changing from one stage of a lesson to another according to the goals of the participants.
It is also said that CA is able to capture the dynamic and complex interplay between the different levels of language interaction as conversational analysts believe that no detail of conversation can be neglected as unimportant (Seedhouse, 1997 quoted in Gewehr et al., 1998).
Nunn in his article 'A Dual Approach to Classroom Discourse' (2003) argues that 'no single approach is likely to be fully adequate for analyzing something as complex as classroom discourse' and shares his experience in using both discourse and turn-taking analysis to analyze the same data.He claims that together they provide insights which neither approach can provide alone.This is the attitude adopted in this study -the analyst should feel free to choose the most suitable approach or a combination of approaches.After all, what matters is not which particular method of analysis we use to describe or structure elements of discourse, what matters is their pedagogic function.

Results and Discussion
The data collected constitute a varied range of options of how real teachers orchestrate classroom talk.From the analysis of the recordings and transcripts, it becomes obvious that some teachers, whether knowingly or instinctively, consistently create opportunities for learner involvement because the language they use matches the pedagogic purpose they have in mind.Out of the eight recorded lesson extracts provided by eight different teachers, one was noticeably more successful than the others from the point of view of the amount and quality of linguistic output produced by the students.This extract can serve as a model for self-reflection and application for other teachers.Most of the examples provided in the discussion that follows have been taken from this particular extract.
In this part of the paper, first the constituents of the extended Triadic Dialogue will be determined, and then examples of the sub-genres of extended or student-initiated IRF found in the data will be provided.For simplicity and economy, only shorter chunks of the transcripts have been selected to illustrate certain features of expanded IRF structures.

Constituents of an extended IRF
As it is shown in the mid-point column of Table 1, the main prerequisites for an expanded teacher-led IRF pattern (Triadic Dialogue) constitute the following: initial mostly referential teacher questions that inspire divergent student answers, follow-up questions, and multifunctional followup moves.Here they will be characterized and illustrated with examples from the data.

Initial questions (in turn 132 Ex1)
Initiating move is obligatory in any kind of IRF sequence and signals the beginning of it.It proposes an issue for discussion and requires the students to contribute to the issue in response.Certainly, the choice of initiating question has an important influence on the way in which the sequence develops (Cullen, 1998, p.180).Referential (new information) questions -questions for which teachers are not seeking one particular answer -are more likely than known information questions to elicit substantive student contributions.Because of their leading role, Wells classifies the very initial teacher-student exchanges as nuclear (Nassaji and Wells, 2000) and claims that, if successfully developed, the initiating question of nuclear exchange can give rise to a number of the following topically related bound exchanges.
The following are examples of successful initial questions from the data.The teacher modifies the initial question by respeaking a rather complex and formal initial formulation in informal register: Ex 1 T: now a question to everyone -I believe that any international company would have some kind of INHOUSE MAGAZINE -to what extent do you believe information that is published there is relevant to the employees within the company?-like -would you expect people to hold their breath waiting for another issue of the magazine to arrive?S4: I can say from my personal experience -that I don't care so much about this magazine T: uh-huh -so WHY is that -does it fail to meet your expectations in any respect?

Variations of the follow-up move Comment (in turns 47, 49 Ex3)
Apart from a follow-up question (which actually is a follow-up move), the teacher may choose to add a comment on the student's contribution.In the context of promoting natural and conversational language use, spontaneous comments and personal (often humorous) reactions create an atmosphere in the classroom which is conducive to learning and likely to promote learner involvement.
Ex 3 S2: and the problem is that this magazine is only for you and for one sort of all our employees 47.T: I guess the employers are getting cheeky now! (laughs) S1: yes -you must think before you say T: [right think twice -right!(laughs)]

Ex 6
S4: maybe we should can use or TRANSLATE -it will be not so expensive -to print -and to put something in the Internet and to use T: uh-huh S4: and just click -maybe three -ok or four -four languages T: so you're saying -you're saying we should aspire to make some savings -allocate the money differently yes T: probably get more translators -interpreters -uh-huh?
In turns 115 and 117 (Ex 6) the teacher not only reformulates and elaborates the student's response, but also turns the reformulated and expanded ideas into questions.Introducing her revoicing by the phrase 'so you're saying…' she also maintains the student's right to evaluate the correctness of her interpretation thus making the interaction more equalized and more natural.In turn 132 (Ex 1) she reformulates the question to change its register from quite formal to informal.

Active listening, back-channeling (in turns 113 Ex 6)
One-word remarks, such as 'uh-huh', 'mmmm', 'right', 'OK', 'aha' mean that the teacher is attending to the message but not actually taking a turn.Like in a natural conversation, they indicate the interest, comprehension and minute-by-minute attention to what students have to say.

Extended wait-time (teacher 'back seat')
As the discourse progresses, the teacher in the most successful of the transcribed extracts takes more and more of 'back seat', which is characteristic of the most open discourse pattern (the far right of Table 1).The learners successfully manage their own turn-taking without the teacher's intervention, the teacher only makes statements in relation to what a student has just said (e.g.'I agree.I think it is more likely that X is the cause.')and provides for a student question (e.g.'Think of the question that's still bothering you about that.').In classroom contexts where the stated aim is to increase oral fluency, deliberate appreciative silence, or extended wait-time, which many teachers might consider to be a sign of weakness, or a sign of not being able 'to do their job', is actually desirable and leads to more complex learner answers or an increase in student/ student interaction (Thornbury, 1995, p.282).

Sub-genres of an extended IRF
This paper does not aim to put forward a taxonomy of the sub-genres of Triadic Dialogue.It will only describe three most open variations of the pattern that were found in the recorded data and present excerpts from the transcript to exemplify the three issues.This lesson segment is a part of a role-play based on the material in the course book which the class had to read at home.The students are supposed to play a team of executives of an international company having a meeting to discuss some problems connected to their company's inside magazine.The simulated discussion is to be led by the President of the company (i.e., Student 1, marked as S1).

Negotiations of meaning
Turns 53 to 68 (Ex 7) show an extended IRF pattern that contains a piece of negotiation between S1, S4 and the teacher about the meaning of 'communication manager'.The pattern is started with a teacherinitiated initial question ('What is your perspective, right?') which is posed in a very friendly informal manner and in which she also encourages the President (S1) to perform her 'duties' (Turn 53).It is important that the negotiation about the meaning is initiated by S1 who is not sure what responsibilities the position of 'communication manager' includes.The student does not actually ask a question, she just starts thinking aloud whether it might include this or that and the teacher helps her by supporting her opinion (Turn 57) and suggesting other possibilities (Turn 58) until S1 completes the utterance (Turn 59).The teacher does not provide a ready answer, she turns to S4 with a question instead thus involving him into the 'search' for the meaning.She tactfully (in whisper) guides him to where to find the necessary information in the book and greets the desired answer with affirmative 'ok' and a joke (Turn 68).
The whole structure could be viewed as a traditional IRF as it starts with an initial teacher question and ends with an evaluative teacher follow-up move, were it not for the element of negotiation in between which so very much resembles a naturally occurring conversation.
Student-asked questions (in turn 292 Ex8) Student asked questions seldom happen in the data gathered for this paper.It shows that Lithuanian classrooms, as perhaps most classrooms in the world, tend to be teacher-led.This particular question (Turn 292) appears after a series of questions (which could be viewed as a series of IRF structures extended with teacher follow-up questions) the teacher asked a few students in turn about their pets.Students were telling the teacher stories based on their personal experiences and then one of them asked the teacher to share something from his own life.By asking a question, the student 'steps out' of Triadic Dialogue (Gourlay, 2005, p.408).As it is seen from the transcript, the teacher exchanges the roles willingly thus creating a more dialogic and more equal mode of participation in the classroom.
The only other case of student asked questions and also some student/ student interaction in the data could be observed in the context of university environment after student-given presentations.When one or more students are presenting to the rest of the class, the role of the 'manager' of the discussion is sometimes handed over to a student.It is most important to allow students some interactional space -as long as the discussion is more or less within the scope of the main topic, students may talk about issues that they find intriguing and that are related to their personal life experiences.To promote this epistemic student talk, the teacher explicitly builds on what the student has introduced and extends the IRF pattern.By doing this, she incorporates the student-proposed connections with what is being discussed (topic shifts) into the classroom discourse and makes them socially significant.By selectively affirming and incorporating student utterances, the teacher can negotiate class objectives while also allowing students the opportunity to direct and elaborate on the topic and scope of discussion.This leads to more content-driven teacher-student interaction, more conversation-like atmosphere in the classroom, and hence more learner involvement.

Implications
If the pedagogic aim is to teach students conversation, it is important to expose them to more open classroom discourse, i.e. to involve learners in classroom activities that will generate output as close as possible to naturally occurring talk.The reality of classroom context, however, is predetermined by two sets of constraints: academic or institutional tasks and social participation of learners and teachers.Complete naturalness is therefore hardly possible in the classroom.Classroom environment can provide only simulations of real life talk which aim to practice conversation and, in many respects, resemble a conversation.
The IRF classroom discourse sequence (Triadic Dialogue) is a valued and traditional classroom language pattern which, as the data for this paper confirms, pervades most ESL classrooms in Lithuania.Depending on the openness of discourse required for the pedagogic purpose, this format can be applied for a wide variety of tasks, ranging from checking episodes where the classic IRE structure is sufficient to more open and more conversation-like extended IRF structures that allow embedded elements such as negotiation, student-initiation and student-induced subtopics.Thus the extended IRF pattern can be considered as a mid-point between a ritual and spontaneity.
In case of successfully applied conversation-like IRF structures, the teacher is no longer a 'primary knower', the teacher then makes moves that enable both students and teacher to contribute and co-construct understanding of an issue for which there is no one single correct answer and in which the goal is to arrive at a consensus after considering a variety of alternatives; i.e. the teacher then supports student utterances by playing roles of 'initiator', 'affirmer', 'clarifier', 'questioner', and 'summarizer'.At the same time the teacher remains a 'manager' responsible for the discussion and for the ensuring that time in the classroom is well spent.
Much could be done by individual teachers themselves to improve their understanding of the relationship between teacher talk, interaction and learning opportunity.Teachers need to be conscious of the need to use language appropriate to their teaching aim.They can find out about their language in classroom by making audio-and video recordings of their lessons which they can analyze with the help of either methods of Discourse Analysis or Conversation Analysis, or the combination of both.An increased understanding of classroom discourse can result in more measured and controlled use of classroom language.

Table 1 .
Participant Frameworks in Close and Open Discourse

Open Questions
As the object of this study is the extended, more freely interpreted IRF structure, the investigation will be placed in the realms of mid-open and open discourse depicted in the continuum in Table encourages the student but also shows the teacher's genuine interest in the newly introduced sub-topic.