THE THIRD TURN:
USING TEACHER UPTAKE TO INFLUENCE
FOREIGN LANGUAGE CLASSROOM DIALOGUE
KRISTINA L. THORNTON
S3645835
M.A. Thesis in Applied Linguistics
Faculty of Liberal Arts
University of Groningen
Word count: 15,133
Supervisors:
Prof. Dr. Wander Lowie
Dr. Audrey Rousse-Malpat
Declaration of Authenticity
MA Applied Linguistics - 2018/2019 MA-thesis
Student name: Kristina L. Thornton Student number: S3645835
PLAGIARISM is the presentation by a student of an assignment or piece of work which has in fact been copied in whole, in part, or in paraphrase from another student's work, or from any other source (e.g. published books or periodicals or material from Internet sites), without due acknowledgement in the text.
TEAMWORK: Students are encouraged to work with each other to develop their generic skills and increase their knowledge and understanding of the curriculum. Such teamwork includes general discussion and sharing of ideas on the curriculum. All written work must however (without specific authorization to the contrary) be done by individual students. Students are neither permitted to copy any part of another student’s work nor permitted to allow their own work to be copied by other students.
DECLARATION
• I declare that all work submitted for assessment of this MA-thesis is my own work and does not involve plagiarism or teamwork other than that authorized in the general terms above or that authorized and documented for any particular piece of work.
Signed______________________________________________________________________
Acknowledgements
I would like to extend my sincere appreciation to Nienke Smit for inspiring me to take on this topic, for sharing her network and framework for classroom observation, and for the many hours of insightful discussion that helped guide the investigation. To Dr. Wander Lowie for gently pushing me to tackle this challenge. To the brave teachers who agreed to participate. And to my wonderfully even-keeled husband Karsten Cramer for standing by my side throughout this complex, dynamic process.
List of Abbreviations CDST Complex Dynamic Systems Theory
CLIL Content and Language Integrated Learning COLT Communicative Orientation of Language Teaching CPRT Cambridge Primary Review Trust
EAL English as an Additional Language EFL English as a Foreign Language FL Foreign Language
HAVO Hoger Algemeen Voortgezet Onderwijs (Dutch higher general education track for secondary school)
IRF Initiation – Response – Feedback L1 First Language
SLD Second Language Development STT Student Talk Time
TT Talk Time
TTT Teacher Talk Time UoY University of York
VWO Voorbereidend Wetenschappelijk Onderwijs (Dutch pre-university secondary school education)
WT Wait Time
Table of Contents
0. Abstract ... 6
1. Introduction ... 7
2. Background ... 9
2.1 Relevant Principles from CDST and Their Application to Classroom Discourse... 10
2.2 Definition of Constructs Under Examination ... 14
2.3 Statement of Purpose ... 22
3. Method ... 24
3.1 Participants ... 24
3.2 Data Collection: Tools and Procedure... 25
3.3 Analysis ... 30
4. Results ... 31
4.1 Mode and Talk Time Distribution ... 32
4.2 Quantity of Teacher Uptake ... 34
4.3 Quality of Teacher Uptake ... 38
5. Discussion ... 41
5.1 Distribution of Mode and Talk Time ... 42
5.2 Quantity, Quality and Impact of Uptake ... 43
5.3 Limitations of Current Study ... 49
6. Conclusion ... 50
6.1 Summary of Findings ... 50
6.2 Directions for Future Research ... 51
6.3 Implications for Classroom Practice ... 52
Appendix A: Uptake Coding Scheme Part 1 ... 54
Appendix B: Uptake Coding Scheme Part 2 ... 55
Appendix C: Comparison of Modes and Talk Time Distribution ... 56
Appendix D: State Space Grids with Mode 3 Uptake Type, Student Utterance Level and Student Language ... 57
0. Abstract
This study focuses on the role of teacher uptake as a type of Feedback move in the IRF sequence to promote extended and complex student utterances as part of a balanced dialogic setting in a foreign language classroom. Uptake represents the responsiveness of the teacher and is critical in guiding classroom interaction into dialogic spells (Nystrand, Wu, Gamoran, Zeiser, & Long, 2003) that promote the learning of both content and language. The three research questions investigated talk time distribution, type and quantity of uptake, and which uptake co-occurs most often with desired student output. Uptake was operationalized according to a coding scheme adapted from Alexander, Hardman, Hardman, Rajab, & Longmore (2017) and plotted on state space grids (Hollenstein, 2007, 2013; Smit, Van Dijk, De Bot, & Lowie, n.d.) along with data regarding the quality of the subsequent student utterance based on a coding scheme by Smit and colleagues (n.d.). Talk time distribution was also investigated as part of the interactive context in which dialogic spells may occur. As a result of seven EFL/CLIL lesson observations, the most frequent type of teacher uptake observed was Expand Questions while the uptake type that co-occurred most frequently with subsequent high-quality student utterances was Backchanneling. The relevant constructs are defined and results interpreted through the lens of complex dynamic systems theory (CDST) (De Bot, Lowie, & Verspoor, 2007; Larsen-Freeman & Cameron, 2008; Seedhouse, 2015).
The third turn: Using teacher uptake to influence foreign language classroom dialogue “For the word (and, consequently, for a human being) there is nothing
more terrible than a lack of response” (Bakhtin, 1987b, p. 127).
1. Introduction
At one intersection of applied linguistics and education is classroom interaction, where the spoken discourse or “language in time” (Nystrand, Wu, Gamoran, Zeiser, & Long, 2003, p. 136) is complex and dynamic (Seedhouse, 2015), shaped by pedagogical goals that interact with social and institutional goals (Seedhouse, 2010b), and influenced by past events as well as its abstract and concrete immediate context in which it is embedded (Seedhouse, 2015). As classroom discourse emerges, with each utterance building on the last (Bakhtin, 1987a), and when the interaction between teacher and students involves unique contributions that build upon each other, the dialogue achieves meaning (see also Wells & Arauz, 2006). Meaningful dialogue in any classroom between teacher and students is engaging and leads to better learning (Bakhtin, 1987b; Cullen, 2002; Haneda & Wells, 2008; Lantolf & Johnson, 2007; Mortimer & Scott, 2003; Nystrand, 1997; Walsh & Li, 2013; Wells & Arauz, 2006). As Walsh and Li (2013) write, “…spoken interactions are used to both transmit and clarify new information and then to reflect on and rationalise new knowledge” (p. 249).
Classroom discussions co-created with independent and individual thoughts from both teacher and students are reflective of Bakhtin’s (1987b) concept of a dialogic model of interaction, within which the concept of responsiveness is elevated (see quote above). In learning situations, this idea connects well with Vygotsky’s (1978) zone of proximal development (ZPD) where interaction with an expert, the teacher, brings novices, students, to the next level of both first language (L1) and content competency through a supportive response to their needs.
In extending the idea of the power of learning through meaningful dialogue to second language development (SLD), a line may be drawn to such underpinning concepts as Interaction Hypothesis (Long, 1996), Noticing Hypothesis (Schmidt, 1990) and Output Hypothesis (Swain, 1985).
In fact, Swain (2000) makes the connection herself by linking her own Output Hypothesis to the notion of “collaborative dialogue” defined as “dialogue in which speakers are engaged in problem solving and knowledge building” (p. 102). Interaction, according to Long (1996), is considered beneficial for the natural process of negotiation of meaning that occurs as two speakers strive to understand each other, which also forces learners to focus their attention on forms and their meanings as per Schmidt’s (1990) Noticing Hypothesis.
Thus, student participation in classroom discourse is critical in any learning environment (see also Nystrand & Gamoran, 1997), and especially where pedagogical goals involve communicative competence in the target language as in classrooms for foreign language (FL) teaching and content- and language-integrated learning (CLIL).
Yet information on the quantity and quality of meaningful dialogue in FL and CLIL classrooms is lacking. “Dialogic classroom studies have thus far been mainly conducted in L1 classrooms. Llinares et al. (2012) note their relative absence in CLIL settings, suggesting that the potential for providing an ‘optimum environment’ (Haneda and Wells 2008) for developing dialogic CLIL classrooms remains to be investigated” (Coyle, 2014, p. 360). Nystrand and colleagues (2003) provide clues as to how the “optimum environment” may be achieved, stating, “…some of the ways teachers seek to kindle dialogic interaction include (a) asking authentic questions… (b) practicing uptake… and (c) using high-level evaluation to valorize students’ responses” (p. 172).
In pursuit of the “optimum environment” mentioned above, this study will provide a timely focus on the role of teacher uptake to promote extended and complex student utterances as part of a balanced, meaningful dialogic setting. Uptake represents the responsiveness of the teacher and is critical in guiding classroom interaction into dialogic spells (Nystrand et al., 2003) that promote the learning of both content and language.
An effort will be made to quantify and qualify dialogism in a foreign language classroom through responsiveness by way of investigating the phenomenon of teacher uptake, its interactional
context, and its impact on emergent classroom dialogue. The following research questions will be addressed:
1. Who is talking in FL/CLIL classrooms?
2. How much and what kind of teacher uptake is taking place in these classrooms, and how does this correlate with teacher talk time?
3. Which type of uptake co-occurs most frequently with subsequent complex student utterances?
The relevant constructs will be defined and results interpreted through the lens of complex dynamic systems theory (CDST) (De Bot, Lowie, & Verspoor, 2007; Larsen-Freeman & Cameron, 2008;
Seedhouse, 2015).
In recognizing through CDST that classrooms are complex and dynamic systems sensitive to initial conditions and context, and where iterative interactions emerge over time, data will be gathered from multiple sources to provide a description of the relevant phenomena and, building on the work of Smit, Van Dijk, de Bot and Lowie (n.d.), classroom interaction will be graphed and analyzed via state space grids (Hollenstein, 2007, 2013). This study will utilize data from two sources: 1) lesson observations from several 4th- and 5th-year EFL/CLIL lessons in the Netherlands, coded for
teacher talk time, student talk time, wait time, teacher uptake, and complexity of subsequent student utterance; and 2) informal open interviews conducted with some of the teachers before and after the observed lessons.
2. Background
This study is informed by a CDST perspective on the classroom as a dynamic and complex system thus outlined in this section and meriting a special attention paid to the discourse context of the micro-interactions at the heart of this study. Also explained below are the sources for the constructs used in the coding and analysis of the lesson observations.
2.1 Relevant Principles from CDST and Their Application to Classroom Discourse
CDST has migrated from mathematics, physics and biology through psychology to the study of language development and hence applied linguistics. As a metatheory (Hollenstein, 2013; Larsen-Freeman, 2018), CDST offers a general perspective on systems and their processes, with several key principles especially relevant to the complex, dynamic system of classroom discourse. Specifically, an understanding of the embeddedness of the classroom system, the significance of the initial
conditions of the classroom at the time of observation, the non-linearity and therefore
unpredictability of discourse in the classroom, and the emergent nature of classroom interaction will guide the interpretation of what is observed in this study.
Complex dynamic systems are embedded and interconnected (De Bot et al., 2007; Larsen-Freeman & Cameron, 2008; Seedhouse, 2015). Complex dynamic systems are composed of multiple elements that interact with and change each other, whether concrete or abstract, and are
embedded within other systems. One illustration of this is the traffic at an intersection. The elements or moving parts here may include cars, cyclists and pedestrians, who are all reacting to each other to avoid collisions. The system is open as energy moves in and out in the form of
elements that come and go, and it changes with time. The complex dynamics of this intersection are part of a larger transportation system that is influenced by the society in which it exists: a busy intersection in Taipei looks very different from one in Delhi. Moving from macro to micro, the various elements of the busy intersection may also be seen as systems in their own right. A person crossing the street is composed of psychological, physical and cognitive sub-systems, to name a few. Thus, as Larsen-Freeman (2012) writes,
Systems are said to operate at a number of different, but interconnected, levels, from a macro level, such as that of a whole ecosystem, all the way down to a micro level, such as subatomic particles. The system can thus be viewed at different levels of granularity, with each nested, one within another. (p. 75)
As adjacent systems interact across levels (e.g. a traffic jam at one busy intersection will affect the flow of elements at the next intersection) and timescales (e.g. local businesses at the intersection may grow over a span of years as the amount of traffic increases, which, at some point, may feed back into the growth of traffic), they change each other. Studying complex dynamic systems in an attempt to understand or at least describe them, then, involves a recognition of their context. Context is another interconnected system, or else it is where multiple systems are interacting on the “outskirts” of the open system under investigation. Larsen-Freeman and Cameron (2008) note,
An open system cannot be independent of its context since there is a flow of energy or matter between system and environment; the context is part of the system and its
complexity…Open systems…not only adapt to their contexts but also initiate change in those contexts; these systems are not just dependent on context but also influence context. (p. 34) Taking a classroom as an example more specific to this study, the physical embeddedness is evident in its place in a school, which is abstractly part of a society’s education system, which is interconnected with its cultural philosophy. Zooming in, the elements of a classroom are students and one or more teachers, interacting across multiple timescales and with a regular flow of energy in and out as students come and go, enter and graduate. The context of the classroom includes its school and education system, and also the local-level interactions between teacher and students, since the classroom system is influenced by changes in any of these adjacent systems. A CDST perspective acknowledges the impact of this broader context.
Complex dynamic systems have a sensitive dependence on initial conditions (De Bot et al., 2007; Larsen-Freeman & Cameron, 2008; Seedhouse, 2010a). Related to context is the significance of initial conditions, or the state of the system at the point in time from which observation starts. In the case of classroom research, conducting questionnaires as well as interviews can provide insight into the many conditions that have shaped the classroom-as-system at the beginning of the observation. For example, questionnaires can provide information on variables such as student engagement, teacher preparation, teacher expectations and goals for the lesson, all of which may
impact the state of the system and may contribute to an explanation of the system’s subsequent behavior. The same lesson taught by two teachers in the same school will exhibit variability within its classroom discourse; this may be due to the training, experience, or mood of the teachers or
perhaps the language backgrounds, relationship history, or engagement of the students. “A fuller trace of conversational system activity would include information from various subsystems, levels and timescales. For example, from the individual as system…from the social group level…from ontogenetic history…and so on” (Larsen-Freeman & Cameron, 2008, p. 172). While the full spectrum of variables will likely never be known, nor the system’s behavior fully explained, CDST recognizes that variability may result from seemingly minute differences in otherwise similar systems. This study will account for this by informing quantitative aspects of lesson observations with qualitative interviews with teachers.
A complex dynamic system self-organizes and displays non-linearity (De Bot et al., 2007; Larsen-Freeman & Cameron, 2008; Lowie, 2013). Over time, subsystems interact with each other while the system interacts with input from its context (Larsen-Freeman & Cameron, 2008; Lowie, 2013). This propels the complex dynamic system from state to state in its state space, which is a metaphorical, multidimensional space that represents every possible state that this system could occupy (Van Geert, 2008). The interaction at the lower level of subsystems may result in patterns on a more macro level (Lowie, 2013) such that some states of the larger system are more frequented than others; these attractor states influence the trajectory of the system in that they require an interaction with more energy (a perturbation) to move out of or away from. The attractors may exist due to internal (self-organized) or external factors, and as the system moves around its state space, the constant interaction within and with other systems renders its path vulnerable to change in any direction at any moment (De Bot et al., 2007; Larsen-Freeman & Cameron, 2008; Van Geert, 2008). What appears to be small input into the system, say, a teacher changing perfume, can result in a significant effect on the system: a number of students have allergic reactions for which the teacher must stop the lesson to care for the students. On the other hand, a seemingly large adjustment such
as a substitute teacher may result in very nearly business-as-usual for the system of that classroom. The change in the system is not proportionate to the input (De Bot et al., 2007). These scenario systems may well have the opposite reaction to the input—in another classroom, perhaps no one notices the teacher’s new perfume, or the substitute teacher’s classroom moves into complete rebellion and chaos. “Whereas convergence to a stable position…is one possibility, convergence toward a cycle of 2, 4, 8 or more different attractor states or growth towards a chaotic pattern may occur just as easily” (De Bot et al., 2007, p. 13). From the perspective of a teacher seeking to guide classroom discourse out of a question-answer pattern and into a discussion, the unpredictability of this nonlinear system may mean that even a seasoned teacher is unsuccessful in launching the desired sequence of interaction despite researchers’ attempts to find out why (see Seedhouse, 1996, for examples).
System change is emergent and iterative (De Bot et al., 2007; Larsen-Freeman & Cameron, 2008). The unpredictability of the system’s development is somewhat mitigated in that each change in the system is predicated on the previous state. In other words, the next iteration of the system may be predicted with enough knowledge of current conditions. Models or descriptions of a complex dynamic system are thus limited:
…the outcome of development over time can therefore not be calculated exactly…because the variables that interact keep changing over time and the outcome of these interactions, unless they take place in a very simple system, cannot be solved analytically. To follow a dynamic trajectory, the system has to be simulated by doing the iterations, for there is no equation that will directly give a value of the system at some later time. (De Bot et al., 2007, p. 8)
Classroom discourse may be organized, for instance, into classroom modes (Alexander, 2018; Walsh, 2006b) or classroom contexts (Seedhouse, 1996), but smaller patterns emerge as well. The IRF sequence (teacher initiates, student responds, teacher gives feedback) has become one of the most recognizable attractors in classroom discourse. The pattern’s emergence can be explained:
Teachers initiate since they are responsible for classroom control, turn to questions to elicit student participation, and then provide feedback, often an evaluation of the learning demonstrated in the student’s response (Larsen-Freeman & Cameron, 2008, p. 180-181). But this pattern may be disrupted and shifted into an unpredictable series of turns. Waring (2009) describes a student’s successful perturbation of the system out of the IRF attractor state which had an unpredictable, significant impact on the continuing discourse throughout the remainder of the lesson as the discourse emerged through other students’ unpredictable disruptions to the teacher’s IRF sequences.
The focus of this study lies on the IRF sequence as an attractor which may appear at any point in a lesson (Wells, 1993). Yet from a CDST perspective, the context of these sequences is critical and requires somewhat broader constructs to be described. The following section provides frameworks for understanding the context in terms of who interacts, how often, and at what point in a lesson. Additional detail is given on IRF sequences and the construct of teacher uptake.
2.2 Definition of Constructs Under Examination
Teacher and student talk time. At its most basic, talk time refers to any moment of a lesson in which a voice is heard and may be broadly divided into two categories, teacher talk time (TTT) and student talk time (STT). Overlap may exist in certain modes of the classroom, for example, during group work facilitated by the teacher. No distinction is made between one teacher or two, and STT represents the collective voice of the students whether there is one student speaker or several.
On a higher level than talk time as spoken utterances is the notion of floor control. A speaker may pause between utterances for emphasis, absorption, recollection, and so on, but maintain the right to continue speaking. The unit of analysis in this case may be termed a turn rather than an utterance. Due to overlapping terminology in the literature, both utterances and turns will be used. Note that neither turns nor utterances here include wait time (see below) as the floor is open for the student response awaited by the teacher, though the teacher may reclaim the floor in order to repeat, restate or redirect the question; in other words, wait time is neither TTT nor STT.
Alexander (2018) cites research indicating the prevalence of recitation mode, characterized by “closed questions, recall answers and minimal feedback” (p. 562), including a study of 23
American schools by Nystrand and Gamoran (1997) which reported the classrooms as
“overwhelmingly monologic” (p. 33). Moving a classroom away from a monologic mode, perhaps a strong attractor in its own right, involves a reduction and modification of TTT to encourage more student output. Not only is the increased quantity of this STT critical in a dialogic setting, but also its quality. Alexander (2018) writes, “To judge student talk merely by the length of utterances…is useful only as a preliminary or general indicator of quality” (p. 584). Additional measures of quality shaped by the goals of the institutional context, that is, the system of the language classroom, include cognitive and linguistic dimensions (see Seedhouse, 2015; Sert, 2015; Walsh, 2013).
Cummins’ Quadrants (Cummins, 2000) provide a framework built on two continuums, that of contextual support and cognitive demand, where language classroom activities and input may be described as 1) contextually supported, cognitively undemanding; 2) supported with little context, cognitively undemanding; 3) contextually supported, cognitively demanding; and 4) supported with little context, cognitively demanding. Quadrants 1 and 2, as cognitively undemanding, do not require students to use an academic register but rather “everyday language”: basic interpersonal
communication skills (BICS). Alternatively, the cognitively demanding nature of Quadrants 3 and 4 involves cognitive academic language proficiency (CALP) (Cummins, 2000). Smit, Van Dijk, de Bot and Lowie (n.d.) provide a scale of student utterance ranging from 0 to 3 which differentiates length and cognitive demand between utterances which are Off-task, Simple (one to three words, cognitively undemanding), Adequate (full sentence or longer, cognitively familiar), and Complex (longer than one sentence, cognitively demanding). In a dialogic foreign language classroom, then, complex student utterances may be said to feature the cognitive challenge of CALP and extend beyond a single sentence. Figure 2.1 below provides an illustration of the quadrants with examples based on the work of Smit and colleagues (n.d.).
Figure 2.1. Cummins' Quadrants (Cummins, 2000) with examples based on the work of Smit, Van Dijk, de Bot, and Lowie
(n.d.).
These criteria of complexity and extended duration may be combined with the scale of student utterances developed by Smit and colleagues (n.d.) to evaluate the quality of interaction occurring around uptake events.
Wait time. This term has been in use since the 1970s, as noted in Swift, Gooding, and Swift (1988), and generally refers to the time given for a student to respond to a teacher’s question. Alexander (2018) describes his encounter with teachers who prefer the label “thinking time”, though this term may be applied to other moments of silence in a lesson where the teacher may be allowing time for absorption. These moments are outside the scope of this study and will be considered part of TTT. For the present purpose, wait time occurs when a teacher’s utterance is designed to elicit a response such as with a question or with a form of uptake (see below) that is followed by a pause.
Wait time is a powerful element of a dialogic classroom as it contributes toward “space for learning” when extended, and can increase student participation in the discourse (Walsh & Li, 2013). As part of a project designed to improve language teachers’ classroom “communicativeness”, Thornbury (1996) further highlights the benefits of an extended pause:
Nunan (1991) quotes studies showing that when teachers are trained to wait three or four seconds, instead of the customary one, not only do more students respond, but there is an increase in the average length of their responses. The proportion of student-initiated
questioning also increases. All of these adjustments would seem to be worthy objectives in a communicative classroom. (p. 282)
Measuring wait time alongside TTT and STT will provide an overview of the distribution of dialogue in the classes under study and therefore another layer of context around the systems under investigation. Classrooms tending toward a more balanced ratio between TTT and STT are expected to evidence more wait time than those with a high TTT-to-STT ratio, as increased wait-time of three to five seconds has historically led to more student engagement and longer utterances (Rowe, 1974; Swift et al., 1988; also see Thornbury, 1996, above).
In the dynamic system of classroom discourse, wait time under certain conditions may fit the description of the edge of chaos as explained by Larsen-Freeman and Cameron (2008), in that the system is unstable and the trajectory unpredictable in this period of silence. When a teacher presents a question to the class and does not nominate a speaker, the level of predictability is low regarding the next state of the system. As wait time is neither TTT nor STT, the system is
uncontrolled until the next utterance. How long will wait time last? Who will speak next? Will the next utterance answer the question, ask another question, or pivot to another speaker? Despite these unknowns, there are patterns within classroom discourse that provide a degree of stability between wait time, TTT and STT, as described in the next section.
Modes of classroom interaction. Classroom discourse occurs in several combinations or modes of teacher-student interactions. In contrast to Walsh’s (2006a) L2 classroom modes which are defined by their pedagogical goals and interactional features, Alexander (2018) provides five modes purely based on interaction:
Whole class teaching (teacher-student) Group work (teacher-student, teacher-led)
Group work (student-student, student-led) One-to-one (teacher-student)
One-to-one (student-student pairs) (p. 567)
This list provides a starting point from which to narrow down the type of interaction examined in this study. As the focus here is on promoting a dialogic classroom where both teacher and students are contributing to discussion, the modes will be distinguished as Mode 0 for no interaction or no opportunities for interaction; Mode 1 for off-task or unstructured interaction such as during
transitions or when the teacher is giving instructions; Modes 2 and 3 for interaction related to lesson content, with Mode 2 for groupwork and Mode 3 for whole-class teaching (see also Appendix B). The talk time distribution and class time allocated to Modes 0-2 will be reported, but these modes will be excluded from analysis in terms of IRF sequences and uptake events.
In Nystrand and colleagues’ (2003) groundbreaking study on classroom discourse via event history analysis, arguably from a CDST-compatible perspective by its investigation of shifts in emergent interaction over time, whole classroom dialogue is further divided into three types: recitation, discussion, and dialogic spells. Recitation mode is primarily TTT and centers around IRF sequences and display questions, whereas discussion and dialogic spells involve co-construction of meaning, implying much more balanced TTT and STT. Dialogic spells, specifically, feature a mix of non-display questions and engaged responses and comments from teacher and students. The discourse turns which instigate the transition of the discourse mode are known as monologic or dialogic bids. In CDST terms, they provide the perturbations that shift the discourse out of the attractor state of IRF-recitation mode, and Nystrand et al. (2003) list two possible elements of a teacher’s F-turn that may shift the discourse into a dialogic spell:
• Responding to and taking up ideas and observations that students introduce, for example, through uptake and authentic questions.
• Withholding evaluation in such a way as to encourage discussion and conversational interaction. (p. 151)
Both possibilities for the F-turn are discussed below.
IRF/E sequence. One of the most recognizable and widely-studied patterns in classroom discourse is the three-turn IRF sequence, Initiate – Respond – Feedback/Follow-up, first delineated by Sinclair and Coulthard (1975). Mehan’s (1979) competing model proposed “Evaluate” as the third turn but has recently revised this to “Feedback” to reflect the range of possibilities beyond a
teacher’s evaluation of the previous utterance (Mehan & Cazden, 2015). The use of the term “Feedback” also raises the importance of this turn in keeping with Hattie’s (2009; see also Hattie & Timperley, 2007) findings of very large effect sizes related to teacher feedback in the classroom.
The IRF sequence functions as an attractor state in the system of classroom discourse. Its prevalence in the classroom, as Nassaji and Wells (2000) suggest, is due to 1) discourse among a group of people requiring facilitation no matter the topic, and 2) the teacher’s responsibility to ensure learning, accomplished by facilitating discussion and offering evaluative comments. The sequence is also flexible in its function and can take place within any mode of classroom interaction, including between students.
Despite its ubiquity in classrooms, the IRF sequence is controversial. Lemke (1990) identifies common uses of what he terms “Triadic Dialogue”: “for the purpose of a Review,…to discuss new topics, to go over the homework, and even to work step-by-step through the solution of a problem” (p. 49); however, he also notes that this pattern benefits the teacher by providing control but at the expense of opportunities for student initiative and challenges to the teacher. Mercer (1995),
meanwhile, argues IRF sequences are necessary to check if students understand “procedural, factual matters” (p. 38), citing, for example, ensuring student safety as good cause for the use of IRF
sequences. Wells (1993) avoids condemnation of the pattern, writing, “…in itself, triadic dialogue is neither good nor bad; rather, its merits—or demerits—depend upon the purposes it is used to serve on particular occasions, and upon the larger goals by which those purposes are informed” (p. 3). Waring (2009), while highlighting a case of student-led, successful departure from IRF sequences, also argues that learning can result from these sequences.
If the goal of the teacher is a classroom discussion or dialogic spell, the strong attraction of the IRF sequence may thwart such an intention. Swift et al. (1988) report instances of teachers referring to lessons as “discussions” while talking over 80% of the lesson time or asking mostly memory-based questions. Yet moving out of the cycle is critical if students are to learn by
participating in discourse. By its nature, IRF yields a ratio of two-to-one for TTT to STT, in terms of turns, skewing the balance of talk time. Repeated questioning in whole-class interactions, especially of display questions which solicit regurgitated information rather than thoughtful processing, threatens the meaningfulness of a discussion (see Swift et al., 1988). Multiple solutions to this problem have been proposed and will be discussed below.
Teacher uptake. In order to successfully launch into a dialogic spell, teachers may manipulate their two elements of an IRF sequence, the Initiation and Feedback turns. Wells and Arauz (2006) write,
…the single most important action a teacher can take to shift the interaction from monologic to dialogic is to ask questions to which there are multiple possible answers and then to encourage the students who wish to answer to respond to, and build upon, each other’s contributions. (p. 414)
The encouragement offered by teachers to increase student participation can take many forms, including follow-up questions, but a key attribute of a productive F-turn is the allowance of the student’s R-turn (response to the original question) to influence it. In fact, in contradiction to Wells and Arauz’s (2006) prominence of open questions, Boyd and Rubin (2006) found it was not openness or authenticity of teacher questions but “their contingency on previous student utterances” (p. 141) that led to extended student talk. Wells (1993) further highlights the significance of the F-turn: “…the third move functions much more [than evaluation] as an opportunity to extend the student’s answer, to draw out its significance, or to make connections with other parts of the students’ total experience during the unit.” (p. 30). In other words, linking the F-turn to previous utterances is
critical to instigating the emergence of or maintaining a dialogic spell. Nystrand and Gamoran (1991) offer a preliminary definition of this practice:
…in high-quality classroom discourse…many of the teacher’s questions are partly shaped by what immediately precedes them. This process of teachers’ incorporating students’ answers into subsequent questions is called uptake (Cazden, 1988; Collins, 1982, 1986), and it is an important way in which teachers engage students in probing discussion. (p. 266)
Building on this idea of teacher uptake, Fröhlich, Spada, and Allen (1985) in their Communicative Orientation of Language Teaching (COLT) observational tool searched for evidence of “Incorporation of preceding utterances” (p. 41) but went further to specify the categories of no incorporation, repetition, paraphrase, comment, expansion, elaboration. The main drawback to this tool is that the dynamic, iterative nature of interaction is not considered; the types of uptake are not weighted or otherwise measured for their impact on proceeding student utterances. In a language classroom, repetition or paraphrasing may offer a linguistic model for students, but from the perspective of a dialogic classroom, teacher utterances of expansion or comment do not have value inherently but rather retroactively, contingent on the quality of the next turn.
In a mid-scale project undertaken by Cambridge Primary Review Trust (CPRT) and the University of York (UoY), a 20-week intervention program trained in-service teachers in dialogic teaching with the aim “to improve the quality of classroom talk and thereby increase pupils’
engagement, learning and attainment” (Alexander et al., 2017, p. 1). The 2,493 nine- and 10-year-old students of the teachers in the intervention demonstrated two months’ progress in English and science ahead of those without the intervention. Especially relevant for the study at hand is that roughly half of the students spoke English as an additional language (EAL), hinting at possibilities for similarly-achieved success in CLIL classrooms. To monitor the implementation of dialogic teaching practices following the training intervention, the research team used a coding scheme that reported on three broad categories of teacher talk during whole-class teaching: questions,
the previous definition of teacher uptake: questions that add on, agree/disagree, expand, rephrase, revoice, ask why, and challenge (Alexander et al., 2017). Setting apart this coding scheme is the separation of talk moves that instigate another utterance from the student, the dialogic follow-up moves, versus those that close the IRF sequence and extend the teacher’s control of the floor to the next turn, termed feedback/evaluation moves. This distinction more effectively recognizes the emergence of interaction, elevating moves that explicitly require another student utterance.
As Alexander and colleagues described the foundation for their coding scheme to be based on “some of the key verbal indicators of typical classroom talk, both traditional and dialogic” (Alexander et al., 2017, p. 4), some similar results are expected to be found between the control group in their project and the teachers observed in this study. Namely from among “dialogic talk moves”, that questions which add on, expand and revoice will occur more frequently than those that ask why, challenge, or ask about agreement.
2.3 Statement of Purpose
The purpose of this study is threefold: 1) to describe the context of interactions during discourse of Dutch secondary FL classrooms in terms of who is speaking more often, teachers or students, and at what point in the lesson; 2) to investigate how, within teacher-led interactions, the discourse is influenced by the teacher’s response to the utterances of students (Can Daşkın, 2015; Cullen, 2002; Jarvis & Robinson, 1997; Lee, 2007; Nystrand et al., 2003); and finally, 3) to search for an indication that a particular type of teacher response is more likely to inspire a student to produce a high-quality contribution to the dialogue (Alexander, 2018; Cummins, 2000; Smit et al., n.d.).
On the first point, the research question is framed as follows:
1. Who is talking in Dutch EFL/CLIL classrooms? In other words, how is talk time distributed between teacher and students?
An investigation into classroom discourse cannot ignore this point as it provides context for the complex dynamic system in the form of a basic measure of the immediate interaction environment (Larsen-Freeman & Cameron, 2008). Fundamentally, if a teacher is talking for over 90% during the
whole-class teaching portion of the lesson, expectations for an extended, quality discussion would be low as students’ contributions would number fewer than one in ten utterances. The interaction certainly could not be considered dialogic, and the number of back-and-forth interactions between students would likely be lower than in lessons where the balance between teacher and student talk was 50-50 during whole-class teaching mode. Measurements of talk time distribution also contribute to a broader picture of how EFL/CLIL English classrooms are managed in the Netherlands.
The second and third research questions center around a particular interaction event: that of the third turn in an IRF sequence, when a teacher responds to a student’s answer. As noted above, the third turn is powerful and can be used to perturb the system out of the IRF attractor state and launch a dialogic discussion (Alexander, 2018; Can Daşkın, 2015; Nakamura, 2010; Nystrand et al., 2003). Though even as discourse emerges in a classroom and is co-constructed by teacher and students, the IRF attractor state can prove strong. In the cases where it is avoided, what has the teacher done differently in the third turn that has led to a shift away from the attractor state? Or rather, being sensitive to the distinction between causation and correlation, what can be observed in terms of co-occurrence between the teacher’s third move and a student utterance that follows it and thereby breaks the IRF sequence? As noted above, the notion of teacher uptake has been identified as a possible initial condition from which a quality dialogue may emerge (Nystrand & Gamoran, 1997; Nystrand et al., 2003). Thus, the second question initiates an exploration of occurrences of uptake in classrooms:
2. How much and what kind of teacher uptake is taking place in Dutch EFL/CLIL English classrooms, and how does this relate to teacher talk time?
Beyond that, what kind of influence on the quality of the dialogue can teacher uptake have? If a teacher simply repeats a student’s answer to a question and then pauses, is the student more or less likely to fill the silence with a complex, academic utterance? How does this compare to when a teacher asks a student to explain their answer to a question? And so, finally:
3. What type of uptake occurs most often with subsequent complex student utterances?
Certainly, a follow-up request for explanation or reasoning would be expected to produce a more complex and extended answer from a student as opposed to a teacher acknowledging a student’s answer with a simple “yes”. This study will attempt to categorize uptake and assign it a level of likelihood of leading to a complex, extended student utterance, while following the example of Smit and team (Smit et al., n.d.) by plotting the observed interactions on Hollenstein’s (2007, 2013) state space grids.
3. Method
This descriptive study involved observations of classroom interaction across seven lessons, which were then coded and analyzed for interaction mode, talk time distribution, teacher uptake and proceeding student utterance with the intention to identify which types of uptake correlated more often with complex, extended student utterances.
3.1 Participants
This study used data from seven EFL and CLIL English lessons filmed between 2016 and 2019 in six Dutch secondary schools. The teachers filmed had between 2 and 38 years of experience (M = 12.14, SD = 13.03) and class sizes were from 15-29 students (M = 21.14, SD = 4.45). The names of the teachers have been changed to protect their identity. Teachers were contacted as part of a network involved with a larger, separate study (Smit et al., n.d.).
The criteria discussed with teachers while scheduling a lesson observation were that the lesson was business-as-usual and involved a reading text. Topics ranged from modern American literature to articles on current events. Teachers’ styles varied: some favored lecture style, some instigated groupwork and peer teaching, and some led class-based discussion.
Dutch students begin formal study of English at the age of 12 at the latest and must receive a passing grade at the end of their sixth year of secondary school in order to graduate. The students
in these lessons had completed several years of study as well as, anecdotally, had considerable exposure to international media. This may explain their high levels of proficiency and the challenging content of the lessons. The estimated CEFR level for all lessons observed was B1-B2. The length of lessons ranged from 31-55 minutes (M = 41.86, SD = 7.71). See Table 3.1 for additional teacher, class and lesson characteristics.
Table 3.1 Participant Characteristics Teacher Teacher Gender Teacher Years of Experience Group Type* Class Type Students’ Year in Secondary School Total Number of Students Age of Students Lesson Duration**
1 Amy F 9 VWO EFL 4 and 5 19 15-17 55
2 Beth F 2 VWO CLIL 4 and 5 18 15-17 45
3 Claudia F 15 HAVO EFL 5 23 17-18 43
4 Dorris F 38 HAVO EFL 4 and 5 22 15-17 40
5 Enid F 17 VWO EFL 5 29 16 44
6 Flora F 2 VWO EFL 5 22 16 35
7 George M 2 VWO EFL 5 15 16 31
*VWO is the highest variant in the Dutch secondary school system and is the precursor to university; HAVO secondary schools are in the middle stream and provide general education to prepare students for polytechnics. **In minutes.
3.2 Data Collection: Tools and Procedure
Coding Scheme Part 1. The lessons used in this study were filmed over the course of three years, from 2016-2019. Using MediaCoder software developed at the University of Groningen (Bos & Steenbeek, 2017), the videos were coded and timestamped according to the events specified in Part 1 of the coding scheme: teacher questions which initiated new IRF sequences, teacher uptake, other teacher talk time, wait time, and student talk time. Events outside of this scheme such as group work or episodes of classroom management were given a dummy code to exclude them from analysis (see also Appendix A). The following descriptions of each event provide additional details.
Teacher questions. Teacher questions were coded whenever they initiated a new IRF sequence, indicated by a change of topic or a question which was not contingent on the previous student utterance. This is exemplified in Table 3.2 below by the question in line 1 distinguishable from the follow-up question in line 6. The former question is derived from the text which had just
been read aloud, while the latter question extends the line of discussion as evidenced by the ellipted form, “What about you, [student’s name]?”:
Table 3.2
Example from Dorris2 of Extended IRF Sequence with Uptake Question
Speaker Utterance Coding Category
1 Teacher: What about your culture, [student’s name]? Is it considered to be polite to leave food on your plate?
Question initiating IRF sequence. 2 Student: No. Yeah, if you have food left on your plate— Student response.
3 Teacher: Yeah? Teacher uptake (backchanneling).
4 Student: —you take it home. Student response.
5 Teacher: So that’s the same. Teacher uptake (comment).
6 What about you, [student’s name]? Teacher uptake (expand question).
7 Student: Oh, I, um, eat everything. Student response.
8 Teacher: Yeah. Teacher uptake (acknowledgement).
9 You eat everything. Teacher uptake (repetition).
10 Yeah. Teacher uptake (acknowledgement).
Teacher uptake. Any instance of teacher uptake was coded individually, including consecutive instances. As explained above, uptake is a teacher talk move that incorporates or is directly contingent on the previous student response. Key indicators for any type of uptake are the use of pronouns that refer to the student’s previous utterance (Nystrand et al., 2003) or repetition of one or more of the student’s words. A simple test of uptake is to ask, “Could that (teacher’s third-turn) utterance exist if no student had responded to the original question?” For situations where uptake was immediately followed by teacher talk time, the boundary between the two was distinguished by changes in prosody, speed, intonation, or with markers such as now, ok, so, right.
Teacher talk time. All other teacher utterances outside of questions and uptake were coded as a form of TTT unless they occurred during pair- or groupwork. This was primarily due to technical limitations as teachers were often inaudible when moving around the room to facilitate students’ work.
Student talk time. STT was coded for any student utterance whether on-task or not, initiating a question or responding, or reading text aloud. STT was not coded during pair- or groupwork for the same reason TTT was not coded during these modes.
Wait time. WT was coded whenever there was a gap or pause between a teacher’s question and a student’s response.
Other. As noted above, the interaction occurring during periods of pair- or groupwork was inaudible and also overlapping, rendering sequential coding of utterances impossible.
Coding Scheme Part 2. The data from the timestamps of each lesson was then exported to an Excel spreadsheet where a partial transcript and additional codes were added. Part 2 of the coding scheme involved assigning classroom modes and, building on the coding scheme proposed by Smit, Van Dijk, de Bot, and Lowie (n.d.), level and language codes to each of the utterances as described below (see also Appendix B):
Classroom mode. Modes were coded based on the type of spoken interaction occurring at any point in the lesson. A code of 0 was given when there was no spoken interaction, for example, during a video or if there was a pause between utterances of over five seconds. This mode also included instances of students reading sections of text aloud since these utterances were neither students’ original thoughts nor co-constructed as part of a dialogue or discussion. Mode 1, managerial mode, occurred when the teacher conveyed administrative information such as exam schedules, giving instructions, or while managing individuals or groups. Mode 2 was coded for pair- or groupwork. Whole-class teaching, including discussion and lecture, was coded as 3. The
separation of the modes allows for a comparison of TTT, STT, WT, and types of uptake across periods in the lesson while also providing the total distribution of talk time for the entire lesson. Mode 3 was of particular interest as providing the best opportunities for dialogic learning since both language and content goals would serve as the teacher’s pedagogical focus.
Type of uptake. Teacher uptake was assigned a code from 0-6 as adapted from Alexander and colleagues (2017). Figure 3.1 below provides an overview of the modification of the original coding scheme to the one used in this study. The original full coding scheme from Alexander et al. for teacher talk includes two codes for teacher questions, either open or closed; these codes were excluded from this study as the focus here is on the third turn (teacher uptake) rather than the
questions used to initiate IRF sequences or dialogue. Code 0 was added for a missed opportunity, i.e. the teacher gave no direct reaction when a student answered a question, and code 4 was added for backchanneling. Backchanneling responses, called receipt tokens by Gardner (1998; also see White, 1997), may be considered dialogic in that each vocalization is an interactive move designed to encourage the interlocutor to continue speaking. These vocalizations are cited as being more likely to precede utterances of the other speaker when rising intonation is used (Gardner, 1998).
Backchanneling utterances may sound similar to acknowledgement, e.g. “Yeah”, but the contrast lies in whether or not there is a pause afterwards; a pause follows a backchannel utterance, which allows the student to maintain control of the floor.
Alexander and colleagues’ original seven dialogic talk moves were consolidated into two groups, expand and justify questions. The main reason for this is that Alexander et al.’s (2017) original coding scheme, while based on “key verbal indicators of typical classroom talk, both traditional and dialogic” (p. 4), was designed to precisely highlight differences between the control and treated groups in their intervention study. As there is no intervention taking place in this study, a more general categorization of uptake type is appropriate.
Level of uptake. Each instance of uptake was given a rating from 0-2 with 0 indicating a missed opportunity, where a student answered the teacher’s question but was not given a direct response by the teacher. The three teacher moves labeled by Alexander et al. (2017) as “Teacher feedback/evaluation talk moves” (p. 10), i.e. acknowledgement, praise, and comment, are not prompts and thus have the function of closing the IRF sequence and maintaining floor control for the teacher. These moves are in contrast with uptake types 4-6 which are vocalizations or questions intended to yield back the floor to students for a response. Level 2 uptake types would be considered the most desirable in a dialogic classroom as they break out of the IRF sequence and
elicit additional student utterances, thereby contributing toward a balance of TTT and STT.
Figure 3.1. Development of codes for uptake type (based on Alexander et al., 2017) and level.
Language of teacher uptake. Also based on a coding scheme by Smit, Van Dijk, de Bot, and Lowie (n.d.), teacher language was coded from 0-3 with 0 for no utterance, 1 for a turn in the native language, 2 for a mix of the native and target language, and 3 for full use of the target language.
Level of student utterance. Student utterances following teacher uptake were coded on a scale of 0-3, as per the work of Smit and colleagues (n.d.). 0 was used to indicate if the teacher uptake failed to result in a student response or if the teacher left no pause for a student response. Student responses of one or two words were given a code of 1. If a student responded with one or two complete sentences, or an extended response using non-academic language or BICS (Cummins, 2000), the code was 2. An extended student response using academic language or CALP, as per Cummins (2000), was awarded a code of 3.
Language of student utterance. 0 indicated an off-task or non-response. Utterances in the native language were coded with 1. A mixed response of native and target languages was coded with 2. A code of 3 was given where the target language of English was used exclusively.
Reliability. An abbreviated version of the uptake coding scheme was tested with four fourth-year female students, all aged 16. Following one hour of observer training, the students were given a transcript to code while watching a brief clip of a lesson from one of the teachers included in this study. Agreement of 81% was reached on identifying instances of uptake as well as categorizing levels of uptake and student utterances. The most frequent source of disagreement was on differentiating between uptake questions and questions that initiated a new IRF sequence. As seen in Table 3.3 below, the question in line 4 appears to continue the topic addressed in the previous lines, yet for two reasons it is not uptake: 1) it is not exclusively contingent on the student’s previous utterance but also on the teacher’s, and more importantly, 2) it is a closed question, neither an expand nor justify question as per the coding scheme.
Table 3.3
Example from Dorris3 of Two IRF Sequences with Uptake
Speaker Utterance Coding Category
1 Teacher: Uh, two possible meanings of fair. I know that you found three. Could you tell them?
Question initiating IRF sequence. 2 Student: Beautiful, honest and just, or just and fun fair, so
kermis, carnival.
Student response. 3 Teacher: And so, uh, beautiful, honest, and fun fair, a fête, a
fate, um, a festivity.
Teacher uptake (paraphrase). 4 Which description fits the poem? Question initiating IRF sequence.
5 Student: Beautiful. Student response.
6 Teacher: Beautiful. Teacher uptake (repetition).
3.3 Analysis
After applying codes from Parts 1 and 2 of the uptake coding scheme, the data was sorted in Excel and the talk time distribution calculated for all modes and for Mode 3 (whole-class teaching) only. Subsequently, the data from the coded observations was adjusted such that consecutive instances of uptake were collapsed into one instance as defined by the highest-numbered type and level in the sequence. For instance, the uptake sequence in Table 3.4 below would be analyzed altogether as paraphrasing, Type 3, Level 1.
Table 3.4
Example from Dorris2 of Uptake Sequence to Be Collapsed into One Line for Analysis
Speaker Utterance Coding Category
1 Teacher: Rich. Teacher uptake (repetition, Type 1, Level 1).
2 Well done. Teacher uptake (praise, Type 2, Level 1).
3 Yes. Teacher uptake (acknowledgement, Type 1, Level 1).
4 So when you’re well-to-do you have a lot of money.
Teacher uptake (paraphrase, Type 3, Level 1).
The data was then realigned such that each collapsed instance of leveled teacher uptake was paired on the same line with the subsequent student utterance and its level, even if the student utterance was non-existent (Level 0). This data was uploaded to GridWare versions 1.1 and 1.15 (Lamey, Hollenstein, Lewis, & Granic, 2004), a software program designed to support CDST-informed research by including metrics on processes, time-series data, or, as in this study, event-based data. GridWare is used to produce state space grids, “two-dimensional state spaces with at least two mutually exclusive and exhaustive categories on each dimension. The intersection of these categories forms a grid of cells representing each categorical combination” (Hollenstein, 2013, p. 14). GridWare provides a variety of descriptive statistics at the levels of whole-grid, cell, and cell groups, including mean event duration and dispersion.
4. Results
The results of the study are organized in three parts according to the research questions: First, in pursuit of the first research question on who is speaking in Dutch EFL and CLIL classrooms, a table and pie graphs are used to highlight the distribution of modes and talk time across lessons with more details provided for Mode 3, whole-class teaching. The second part visualizes the data
collected on instances of uptake and student responses using tables and state space grids pertaining to the second research question on quantity and types of teacher uptake taking place. The
correlation between uptake and TTT is also provided. The final part provides results of analyses conducted to find relationships between types of uptake and extended student utterances in order to answer the third research question regarding the type of uptake which co-occurs most frequently
with high-quality student utterances. Percentages rather than seconds are primarily used for the sake of comparison as the duration of each lesson differed. The students’ language during uptake events in George’s class is presented at the end of this chapter.
4.1 Mode and Talk Time Distribution
Table 4.1 below reveals the differences in teaching style and type of lesson, showing how modes were distributed in each lesson according to the following code: Mode 0 for periods where interaction could not take place; Mode 1 for episodes where interaction was not related to content such as classroom management, giving directions, or noisy transitions between activities; Mode 2 signifies pair- or groupwork; and Mode 3 represents whole-class teaching (also see Appendix B for more information on modes). Percentages were used instead of minutes due to differences in lesson duration.
Table 4.1
Percentage of Lesson Time Spent in Each Mode
Teacher Mode 0 Mode 1 Mode 2 Mode 3
Amy 0.21 0.04 0.00 0.75 Beth 0.19 0.23 0.23 0.35 Claudia 0.09 0.16 0.40 0.35 Dorris 0.11 0.15 0.25 0.49 Enid 0.00 0.43 0.17 0.40 Flora 0.22 0.35 0.23 0.20 George 0.34 0.26 0.00 0.40
The relatively high percentages for Mode 0 represented listening activities such as the video that Amy played, or silent reading or writing activities as in the cases of Beth, Flora and George. High percentages in Mode 1 reflect more time managing the classroom by giving instructions or administrative information, or in lengthy, unstructured transitions between activities such as in Enid’s class. Mode 2 represents pair- or groupwork, which was 40% of Claudia’s lesson and
approximately a quarter of Beth’s, Dorris’s and Flora’s lessons. For Mode 3 the range was 0.20-0.75 (M = 0.42, SD = 0.17), showing considerable variability reflecting differences in teaching styles and lesson types. For four out of seven lessons, more time was spent in Mode 3 than in any other mode.
The pie graphs in Figure 4.1 below represent the sums of all utterances, regardless of quality, observed over the course of seven lessons. In other words, 78% of all utterances observed were spoken by teachers and 19% by students. The ratio is nearly the same for all utterances observed in Mode 3 where the percentage of observed STT lost 1% to WT; this is largely due to Modes 0 (no interaction) and 2 (group work) yielding no measurable utterances.
Figure 4.1. Combined distribution of talk time for all teachers in all modes of the seven lessons and then in Mode 3 only.
As Mode 3, whole-class teaching, was the focus of this study, talk time (TT) for this mode was further broken down for each teacher by teacher talk time (TTT), student talk time (STT), and wait time (WT) as seen below in Table 4.2.
Table 4.2
Distribution of Talk Time in Mode 3*
Teacher TTT STT WT Amy 0.94 0.04 0.02 Beth 0.62 0.34 0.04 Claudia 0.73 0.24 0.03 Dorris 0.58 0.37 0.05 Enid 0.96 0.04 0.00 Flora 0.66 0.19 0.15 George 0.62 0.32 0.06
Again, differences in teaching style or lesson type are evident, with Amy and Enid talking for 94% and 96% of their respective lecture lessons in contrast with Beth and Dorris who only spoke for 62% and 58%, respectively, while their students had the floor for 34% and 37% of Mode 3. George also had a relatively low TTT at 62%, but he had the second-highest WT at 6%. Flora spoke for 66% of the lesson but had the highest WT out of all seven teachers, with 15% of Mode 3 spent waiting for students to answer, more than double the next-highest WT of 6% from George. Appendix C provides more visual data on modes and TT across teachers.
4.2 Quantity of Teacher Uptake
In answer to the second research question, Table 4.3 provides an indication of the overall quantity of uptake occurring in the observed lessons, shown in percentages rather than number or duration of uptake events.
Table 4.3
Uptake Interactions Observed Per Lesson
Teacher Percent of Lesson Percent of Mode 3 Amy 0.05 0.06 Beth 0.05 0.13 Claudia 0.11 0.28 Dorris 0.15 0.28 Enid 0.00 0.00 Flora 0.03 0.13 George 0.11 0.26
The percent of lesson time occupied by instances of uptake and subsequent student response ranged from 0.00-0.15 (M = 0.07, SD = 0.05). In Mode 3, this percentage increased as this is where most uptake occurred. The percent of Mode 3 occupied by uptake and subsequent student response ranged from 0.00-0.28 (M = 0.16, SD = 0.11). Claudia, Dorris and George spent more of their whole-class teaching in uptake interactions than other teachers, with a respective 28%, 28% and 26%.
As an additional investigation into the relationship between TTT and uptake events, the number of uptake events for each teacher was converted to a rate per hour. This was then correlated with the percentage of TTT in Mode 3 and recorded in Table 4.4 below:
Table 4.4
Rate of Uptake Events Compared to TTT in Mode 3 Teacher Rate per Hour % TTT
Amy 41.45 0.94 Beth 64.00 0.62 Claudia 118.60 0.73 Dorris 129.00 0.58 Enid 2.73 0.96 Flora 51.43 0.66 George 135.48 0.62
A Shapiro-Wilk test showed normal distribution for both rate of uptake events per hour (W = 0.912, p = 0.410) and percentage of TTT (W = 0.821, p = 0.066). But the result of a Pearson correlation test, while expected to be negative, was not significant (r(5) = -0.742, p = 0.056).
It is important to note here, as stated in the previous chapter, that in cases where utterance of teacher uptake occurred in a sequence without pauses, the sequence was collapsed and
categorized based on its highesttype of uptake. As seen in Table 4.5 below, by raw numbers the most common type of uptake observed was Expand Questions with a total count of 121 instances, 919.02 seconds, over seven lessons. The second most common type was Paraphrase/Comment with 79 instances totaling 480.15 seconds of interaction.
Table 4.5
# Duration # Duration # Duration # Duration # Duration # Duration # Duration
3 0 0.00 0 0.00 2 34.20 1 14.50 6 99.81 7 129.43 2 23.52 2 0 0.00 10 65.56 1 4.48 8 50.86 15 117.40 42 395.22 13 111.74 1 1 0.08 29 74.49 0 0.00 14 63.58 11 29.34 56 315.75 4 20.27 0 1 1.44 34 59.67 24 54.80 56 351.21 3 2.84 16 78.61 3 14.89 2 1.52 73 199.72 27 93.48 79 480.15 35 249.39 121 919.02 22 170.42 *Duration in seconds.
Number and Duration* for Uptake Events in Mode 3 for All Teachers
S tu d e n t U tt e ra n ce Le v e l TOTAL 0 1 2 3 4 5 6
No Uptake Acknowledgement/ Praise Paraphrase/ Backchannel Expand Question Justify Question
A summary state space grid for all observed lessons provides visual insight into the
frequency and duration of uptake-response events (see Figure 4.2 below). Each instance of uptake interaction is marked with a dot, and the size of the dot expresses its relative duration. The rate of dispersion, the distribution of instances throughout the grid (Hollenstein, 2007), for all uptake interaction in Mode 3 is 0.815.
Figure 4.2. State space grid showing a summary of uptake type and its resulting student utterance level for all lessons.
While the type of uptake has not yet been assigned a level on the grid above, the
concentration of numerous smaller dots toward the lower left side of the grid indicates that these types of uptake, Paraphrase/Comment excluded, tended toward uptake events that were shorter and lower in quality. The last three types of uptake were much less often followed by a Level 0 student utterance in comparison and were also longer in duration.
Another observation from this grid is that the vast majority of uptake events occurred in Mode 3, which shares the reasons behind Figure 4.1 above, that the interaction in Modes 0 and 2
.
was not measured. Mode 1 interaction ranged from 4% to 43% of each lesson observed, but this was much less than the 20%-75% of each lesson spent in Mode 3 (see Appendix C).
A preliminary visual inspection indicates possible attractors in Paraphrase/Comment uptake with Level 0 student response as well as Expand Questions with Levels 1 and 2 responses. These patterns of higher frequency and duration may indeed qualify as attractor states by meeting certain criteria. Hollenstein (2013) writes, “Strong attractors will have frequent visits, long mean durations, short return times and return visits, and low first entry values” (p. 77). Table 4.5 below highlights the cells which rise above average for Visits and Duration but below average for Return Time, Return Visits and First Entry.
Thus, the three strong attractor states identified by this set of criteria are 1) Paraphrase/Comment with a Level 0 student utterance; 2) Expand Questions with a Level 1 student utterance; and 3) Expand Questions with a Level 2 student utterance, confirming the previous prediction based on Figure 4.2 above.
4.3 Quality of Teacher Uptake
The third research question asks which type of uptake correlated most highly with
subsequent complex student utterances. By utterance count, only 5% of total teacher uptake moves were followed by a Level 3 student utterance. This rises considerable to 25% for Level 2 student utterances. Level 1 student utterances followed 32% of total teacher uptake utterances, and Level 0 student utterances were most common by following 38% of teacher uptake moves.
Figure 4.3 below shows that 39% of Level 3 student utterances followed Expand Questions while 33% proceeded Backchanneling from the teacher.
Figure 4.3. Types of teacher uptake preceding Level 3 student utterances.
Using the information in Table 4.4 above, the proportion of each type of uptake that led to a Level 3 student utterance can be calculated: 0% of No Uptake and Acknowledgement/Repetition; 7% of Praise; 1% of Paraphrase/Comment; 17% of Backchanneling; 6% of Expand Questions; and 9% of Justify Questions. Thus, teachers’ Expand Questions accounted for more higher-quality student utterances, but this occurred for only 6% of the Expand Questions asked. 17% of teachers’ Backchanneling utterances, on the other hand, preceded 33% of Level 3 student utterances.
