This web site contains naturalistic and experimental studies of face-to-face and electronic educational discussions in home and school settings conducted by Robert J. Beck and colleagues from 1993 to the present. Educational discussions are a form of institutional talk that is casual in the sense that there are no rigidly structured turns of conversation as there are in courts of law or church services (Edwards, 1997). However, educational talk does involve characteristic conversational turns, such as questions, responses, and feedback and is socially structured – it presumes speakers of unequal status, teachers and students, or the emergence of hierarchies in group discussions – and is intentional in that the purpose is to instruct, whether the subject matter is improvised or pre-determined. Because educational conversations are self-organizing, even as they follow broad cultural rules, they tend to vary greatly and the accepted method of research is to analyze turn-by-turn transcripts based on the audio recordings of actual dialogues or, in the case of electronic educational discussions, to analyze the sequence of recorded text communications. Educational discussions generally take place most effectively in one-on-one or small group conversations conducted by parents, teachers and coaches in any setting, including classrooms, that that are intended to instruct children or students.

While discussions are a ubiquitous pedagogy in educational practice of any theoretical persuasion, little is known about how it contributes to student learning, nor are there more than superficial guidelines for best practices – use a variety of questions, involve everyone, call on students by name, and provide meaningful continuous assessments. Recent studies of teamwork and collaborative learning have provided a few insights into the beneficial effects of dialogic communications on learning processes and on group task performance. Some research has shown that patterns of giving and receiving elaborated help are critical components of teamwork skills (Webb, 1993). In addition, giving explanations helps senders of messages to reorganize and clarify material (Bargh & Schul, 1980); receiving explanations can be beneficial by filling in gaps of understanding or correcting misperceptions and strengthening connections between new information and previous learning (Mayer, 1984; Wittrock, 1990). The best pedagogic advice comes from McKeachie (2002) who, at least, expands the guidelines to include such valuable suggestions as starting discussion with a common experience, using controversies, and problems or cases. He also thoughtfully considers barriers to discussion, among which are student passivity, fear of looking stupid, and perhaps, most important, “failure to see the value of discussion” (p. 40). There are probably three interrelated reasons for the dearth of generalized knowledge of discussions: the lack of research; the highly idiosyncratic nature of teachers’ approaches to conducting discussion; and the methodological difficulties of investigating discussions. However, Cazden and colleagues (Cazden, John & Hymes, 1972; Cazden, 1988) pioneered classroom discussion studies and found evidence of common patterns of questioning, response, and evaluation communications. In one of the few well-designed assessments of discussion, Nunn (1996) reported that typically only about 25% of students regularly participate in classroom discussion and teacher-directed discussions are normative. Teacher asks question, Student A responds; Teacher asks another question, Student B responds, etc. Nunn found that peer-to-peer communications constituted only a few per cent of all discussions.

The purpose of our research studies is to develop a pedagogy of discussion by analyzing non-traditional contexts, such as family discourse, mother-child conversations, electronic discussions in higher education and that supreme mode of teaching and learning, the Oxford tutorial. Formalization of the pedagogy remains a goal and our best synthesis is probably contained in the paper on the Oxford tutorial, which integrates some material from other of our studies, but is necessarily generalized only to discussions between tutors and one or two students.

Our approach is based on a communications model for educational discussion in that we assume that speakers are expressing intentions – those taking the role of teacher intend to teach and those who are students or neophytes intending to learn (or know they are being taught). Moreover, because educational discussions are a form of institutional talk, we also assume that cultural rules prevail in guiding the conversation. This does not mean that teachers are necessarily effective in their instruction or that students learn. In fact, in some of the studies students apparently failed to learn and may take the teacher role, attempting to instruct the nominal teacher. Neither does the adoption of the communications model mean that we eschew conversation analysis (CA), which is concerned with the sequential structure of all conversations for any social purpose. In fact, we are generally preoccupied with the sequential organization of conversational turns particularly those speech acts that have been analyzed in pedagogic discourse. But, because educational discussions are culturally institutionalized we do use conversations that are part of experimental designs and resort to cultural models and metaphors to analyze conversations. See Edwards (1997), however, for his reasonable justification of a scientific approach to conversation analysis that insists on strictly naturally occurring talk, uses “unmotivated looking” on the part of the observer, and consequently rules out the use of pre-determined categories of analysis. We do accept the extreme difficulty of identifying goals and intentions in samples of educational discussion, but argue that talk proves to be less indeterminate and ambiguous in social situations where teacher and student roles are relatively clear cut. We concede that intentions are elusive, and evidence may be inconclusive, especially in our studies of families and mothers and young children.

Nevertheless, our strategy is to fit various scientific and metaphoric models to the dialogues that have been collected. It will come as no surprise to discourse researchers that discussions afford a variety of theoretical frameworks and the danger is that of any ethnographic enterprise, superimposing the researchers’ views over the participants’ views, who may have had very different meanings or, even more likely, may have had little self-consciousness of what they meant by their language. The test of this approach must remain in the eyes of the reader-beholder, whether the particular model explains, persuasively, what is taking place in the discussion. Thus, we deem our program, “Towards a pedagogy of educational discussion.” We invite critical response and discussion of our work.

The research program commenced with a naturalistic study of a family dialogue at home that was intended by parents to socialize their pre-adolescent sons after an aggressive altercation (Beck and Wood, 1993). This was followed by an experimental study of mothers and 5-year olds discussing a short movie to prepare children to retell the story (Beck & Clarke-Stewart, 1998), (Clarke-Stewart, 1999), (Beck, 2002; Beck & Bear 2003a). In the context of a teacher preparation program, we subsequently researched experimental peer electronic discussions that were aimed at improving their understanding of the California Standards for the Teaching Profession (Beck, Fitzgerald, & Pauksztat, 2003b), and enhancing electronic portfolios (Beck, Livne, & Bear, 2005). Finally, I conducted a literature and observational study of Oxford University tutorial discussions with a view to developing a pedagogy of the tutorial as a means for understanding its underlying dialogic learning theory (Beck, 2007).

In each of these studies our research approach was to record and transcribe, or observe educational discussions as the primary data. Then we adopted cultural or scientific models and metaphors as frameworks for the close analysis of the data. As guides to the studies, the frameworks are critical and will be presented in detail following. For each study the reader is invited to consult the publication to determine the effectiveness of the framework in creating meaning from the sample dialogues and in generating useful insights into the optimal pedagogy of discussions.

NB. All references are found in the publications.

The Studies

In Beck and Wood (1993), two parents and two children were discussing a fight that had just transpired in which, apparently, the older brother (11) hit the younger (9). On tape we captured the boys’ varying stories of their fight and the parents’ responses that questioned the veracity and the intent of these stories. The dialogues were used educationally to elicit and critically inquire into stories and to provide alternative stories by the parents. The parents used their stories to get  the boys to revise their stories particularly with respect to intentions (who started it?) and thus to rehearse and learn moral rules and behavior that would better prepare them for avoiding fights of this kind.

Two frameworks were used to analyze the dialogues of this study: Toulmin’s jurisprudential logic; and Schank & Abelson’s script theory

Court of Reason (S. E. Toulmin, 1963)

Toulmin (1963) described logic as a “generalized jurisprudence” consisting of pragmatics used to conduct social investigations or inquiries:

Logic is concerned with the soundness of the claims we make – with the solidity of the grounds we produce to support them, the firmness of the backing we provide for them—or, to change the metaphor, with the sort of case we present in defence of our claims…A sound argument , a well-grounded or firmly backed claim. Is one which will stand up to criticism, one for which a case can be presented coming up to the standard required if it is to deserve a favourable verdict. How many legal terms find a natural extension here! One may even be tempted to say that our extra-legal claims have to be justified, not before Her Majesty’s Judges, but before the Court of Reason” (pp. 7-8).

According to Toulmin, the basic layout of arguments in a “Court of Reason” consists of Data that lead to a Conclusion under the authorization of Warrants. Warrants, which are generally implicit, are “logical bridges” (p. 98) that allow one to argue a step between data and conclusions. Warrants may assume various forms: premises, assumptions, justifications, reasons, beliefs, or any form of pre-suppositional understanding used to authorize a conclusion.

We propose that the court of reason, or the logical criteria for the proof of claims, provides normative standards for arguments in ordinary family educational dialogues as it does in legal courts.

Schank & Abelson’s script theory (1977)

This theory was conceived as a way of organizing the distinctive features of world knowledge, in which “understanding is defined as the fitting of new information into a previously organized view of the world” (p. 422). Script theory, which refers to the analysis of narrative understanding, has three structures: script; story; and plan. “A script, as we use it, is a structure that describes an appropriate sequence of events in a particular context” (p. 422). Their well-known example is the Restaurant Script, which describes culturally appropriate sequences of behaviors in a restaurant. Stories are reported actions that deviate from script requirements. Whoever is supporting the cultural validity of the script is compelled to correct the story. “Plans are responsible for the deliberate behavior that people exhibit….By finding a plan, an understander can make guesses about the intentions of an action in an unfolding story and use these guesses to make sense of the story” (p. 428). In Toulmin’s generalized jurisprudential logic, with which script theory is paired in this study, a script is a warrant that enables an observer to draw a conclusion from a story, considered as data. Plan or intentions may be data or warrants. Intentions may simply be part of the story (datum) element whose existence it is important to establish in making a proof. But when the meaning and definition of the intentions are in dispute, then the parties are addressing the backing of the warrant, that is the criteria by which the intention is classified and judged.

            In the second study, involving four papers (Beck and Clarke-Stewart, 1998; Clarke-Stewart & Beck, 1999; Beck (2002), Beck & Bear, 2003a), we researched  the problem of mothers helping their 5 year-old children to view, recall, and retell a short movie story about a wounded deer. The performance of these children was compared with a control group who did not receive help from their mothers. At issue in these contentious mother-child conversations was whether to shoot the deer, Prancer (mothers’ positions) or not to shoot the deer (children’s positions). We studied how spontaneous discussion, involving chains of questions, responses, and feedback were used to help the young children understand the different moral issues in the story. Mothers and children, who discussed the moral issues extensively, used a variety of questions, made elaborated responses to each others ideas, and offered repeated mutual feedback and assessments, were associated with children, who in a subsequent standalone performance were able to retell the movie story most accurately, comprehensively and with greater understanding.

            To interpret the educational discussions in this study we used four theories: scaffolding; moral scripts; dialogic inquiry; and “theory” theory.

Scaffolding is an adult- or expert-facilitated process that enables a child or novice to solve a problem, carry out a task, or achieve a goal that would be beyond his or her unassisted efforts (Wood, Bruner, & Ross, 1976).  Scaffolding includes both content and form, which are seen, respectively, in scripts and dialogue.  Wood and Middleton (1975) found that mothers who successfully scaffolded were those who had systematically changed their instructions on the basis of the child’s response to earlier interventions and were able to estimate the child’s current ability or readiness for different types of instructions.  Such mothers were therefore contingently responsive to their children and employed verbal communications within their intellectual grasp.  Hobsbaum, Peters, and Sylva (1996) argued that “scaffolding can take place only in one-on-one teaching situations because contingent responding requires a detailed understanding of the learner’s history, the immediate task and the teaching strategies needed to move on” (p. 32).

While researchers such as Wood and Bruner did not refer to the length of exchanges of turns between mother and child, it is apparent that, in order to “systematically change instructions on the basis of the child’s response to earlier interventions,” dialogues would need to consist of an extended series of turns on particular topics.  Moreover, there is a growing body of evidence that topically focused elaborated and extended exchanges between adults and children contribute to children’s narrative and language development.  Several investigators found that children included more material in narratives of personal experience when parents extended children’s topics rather than switching topics (McCabe & Peterson, 1991) or when mothers asked them elaborative questions (Fivush & Fromhoff, 1988; Reese & Fivush, 1993; Reese, Haden, & Fivush, 1993). 

In a study of shared book reading, Haden, Reese, and Fivush (1993) found that children, whose mothers embellished and elaborated on indirectly specified information in the storybook, understood and retold the story better (although these differences were not statistically significant because of the small sample studied).  As part of maternal training programs, Arnold, Lonigan, Whitehurst, and Epstein (1994), Dale, Crain-Thoreson, Notari-Syverson, and Cole (1996), Lonigan and Whitehurst (1998), and Whitehurst, Falco, Lonigan, and Fischel (1988) found that when mothers employed numerous questions, followed children’s answers with questions, shadowed their interests, and expanded what they said, this promoted children’s language development.  Scaffolded extension and elaboration of children’s story topics appears to be critical in the development of children’s understanding of narratives.
A relatively ignored characteristic of conversational scaffolding about narrative subjects is that it not only consists of speech acts, such as questions and corrections that are used to probe and assess understanding, but also that these verbalizations are concerned with a particular topic or theme (Mehan, 1979; McCabe & Peterson, 1991).  Lemke (1993) argued that an adequate account of an episode must address the thematic content as well as the activity structures.

Morality Scripts

Morality scripts are used during interactions in which adults respond to children’s misbehaviors or moral understandings through scaffolding.  Scripts help children to develop an understanding of relevant moral concepts, rules, and norms.  Narrative structures dictate the kinds of thematic content needed to make sense of the story.  These structures, such as character intentions and feelings, as well as moral rules, might complement the dialogic moves in developing a comprehensive model of scaffolding.  As Bruner (1986) theorized, story comprehension consists of integrating the dual landscape of story actions and characters’ consciousness or intentions.  Beck and Clarke-Stewart (1998) found that the critical cognitive development issue for young children in retelling stories was not so much the recall of the so-called causal chain of objective actions, but rather the comprehension of characters’ intentions. 
Importantly, the understanding of intentions is a key developmental milestone in children’s moral socialization (Blasi, 1987).  The children in the present study may not have yet achieved this milestone and, as such, might not understand the intentions of characters in the movie story.  Newman, Griffin, and Cole (1989) considered the asymmetry between adults and children in culturally organized activities in which the dominant task definition was one of movement toward the adult system.  Because of the complexities of the moral issue in the movie, it was expected that the mothers and children of this study would have asymmetric views of the moral issue based on the latter’s developmentally challenged understanding. 

Several studies of young children’s moral socialization, using mother-child dialogues as data, have analyzed the role of implicit morality scripts in which adults respond to children’s misbehaviors with scaffolding that supports their understanding of relevant norms and rules (Emde, Johnson, & Easterbrooks, 1987; Edwards, 1987; Much & Shweder, 1978).  In Beck and Wood’s (1993) study of a fight between two pre-adolescent brothers brought to a family discussion, the parental moral scripts referred to the boys’ communications concerning their intentions toward each other, the history (stories) of their aggressive interactions, and moral standards of verbal and physical aggression.  Parents used the scripts both to inquire into, and to repair, the children’s misunderstandings or lack of understanding of their actions.  As such, in the present study, it was expected that mothers would use script-based arguments to overcome the asymmetric moral understanding engendered by the complex moral problem of this movie story.

Scaffolding as Dialogic Inquiry: IRE Sequences

Stone (1998) criticized research that employed unidimensional coding systems such as counting types of parental questions and their responses to children’s understandings.  He suggested that such an approach was likely to miss important “communications dynamics” in optimal patterns of scaffolding.  Stone felt that the study of communicational processes, rather than frequency of individual scaffolds, would be more likely to yield understanding of rules governing well-formed scaffolding.  Explicit in the research of scaffolding, as extending and expanding children’s responses, is the idea that the exchanges of communications or dialogic moves need to be analyzed to provide a more detailed model of knowledge construction in instructional dialogues.  Halliday (1993) also stated that exchanges of dialogue were the meaningful units to analyze, not individual moves.  Only exchanges could adequately show how collaborative knowledge artifacts were co-constructed by participants in a dialogue.
Wells (1999) formulated a dialogical inquiry model of semiotically-mediated activity.  His model involves co-participants, with varying degrees of skill, who are engaged in jointly solving a problem.  In the process, cultural artifacts, e.g., norms and reasoning, are generated, which  may be used to mediate the solution.  Finally, an “object,” such as the story of the video and its moral in the present case, is created in the process of formulating a solution.  From an analysis of a large corpus of classroom dialogues, Wells concluded that inquiry-response-evaluation moves in instructional conversations offer evidence of a ubiquitous conversational structure, with variations, that contributes to progressive knowledge building.

It has been argued that a dialogic inquiry model of scaffolding is supported by the typical IRE sequences found in formal classroom dialogues.  In these sequences, a teacher Initiates with a question, the child Responds, and the adult Evaluates the response.  IRE sequences were first recognized and labeled as such by Sinclair and Courthauld (1975) and have been found in naturalistic observational studies of verbal behavior in high school classrooms (Bellack & Davitz, 1963; Amidon & Flanders, 1963; Flanders, 1963; Cazden, 1988).  The researchers found that, in these dialogues, the IRE sequences were used to start teacher-student exchanges.  These IRE sequences then formed the basis for later collaborative elaborations in the conversation. 

Mehan (1979) specified a particular variety of extended educational exchanges that followed a theme, which he termed “Topically Related Sets,” consisting of multiple basic IREs and conditional IREs, e.g., IRs.  Several studies have found that IREs are effective strategies in contributing to children’s learning (Flanders, 1963; Beck & Clarke-Stewart, 1998).  Lemke (1990) identified a ubiquitous generic structure in learning in which English is the language of instruction as the triadic dialogue, which consists of multiple repeating sequences of IREs.

Attention has been drawn to the role of the third term, E, in IRE exchanges (Wells, 1999).  Sinclair and Courthauld (1975) refer to this term as follow-up to a response, while  Mehan (1979) has emphasized that the move is an evaluation, which may be the most common use.  Wells (1999) theorized that the third term served to provide feedback that extended the student’s answer to draw out its significance or to make connections with other parts of the student’s total experience of the unit (p. 200).  Specifically, “ . . . in the third move of the IRE exchange—when this discourse genre is used effectively—it is in this third step in the co-construction of meaning that the next cycle of the learning-and-teaching has its point of departure” (Wells, 1999 p. 207).  Beck and Wood (1993) also found that questions embedded in the third turns in moral socialization dialogues served as feedback to extend discussions.

Wood, Bruner, and Ross (1976), Dillon (1988), and Herrenkohl and Guerra (1998) have argued, however, that teachers’ use of IRE sequences in classrooms was simply ritual recitation techniques and was not productive for student learning.  It was suggested that, in using IREs, teachers merely asked questions to which they knew the answers, i.e., to simply test and not build upon or extend student prior knowledge.  In naturalistic cultural appropriation discourse, however, when parents use IREs, it is probably rare for them to be simply testing children.  Beck and Wood (1993) found that, while parental evaluations during moral socialization dialogues were used to correct children, their evaluations also addressed selected parts of communications to which children needed to pay attention because they apparently misunderstood the issues.  Thus, evaluations served to elaborate, not bring closure to, the inquiry.

Graesser, Bowers, Hacker, and Person (1997) carried out a naturalistic study of tutoring involving 13 tutors and 40 tutees in middle school and high school research methods and mathematics courses.  As a means to identify particular strategies and conversational styles of individuals, detailed microanalyses were performed on the tutorial dialogues, including the speech acts within each turn and the feedback that speakers gave each other’s contributions.  The results indicated that tutors who engaged in collaborative question answering and problem solving were highly effective.  Specifically, the dialogic explanations improved tutees’ comprehension and memory for material.  Collaborative strategies involved joint tutor-tutee elaborations of IRE sequences.  Thus, in 4th and 5th steps beyond standard three-step IREs, tutors and tutees collaboratively improved upon the initial levels of understanding.

“Theory” theory (Gopnik, 2001)

The purpose of the study (Beck & Bear, 2003) was to develop a model of children’s narrative theories of characters’ emotions in a movie story that would explain and systematize previous findings and serve as a guide for closer examination of “emotional language” in mother-child exchanges about story characters.  So called “theory theory” proposes that “cognitive development is analogous to processes of theory formation and change in science” (Gopnik, 2001, p. 45).  The theory assumes that language learning in certain contexts has positive effects on cognitive development.  In this view, cognitive development involves rules and representations that are similar to those used in scientific development and progress. 

Theory formation and revision are concerned with accounts of the “underlying causal structure of the world” (Gopnik, 2001, p. 46).  Keil and Silberstein (1996) point out that it is commonplace for us to have theories about people and their dispositions to behave.  “Humans spend endless amounts of time on . . . after-the-fact explanations about why people acted as they did, not so much to predict their future behavior as to know how to evaluate and react when people act in certain ways” (p. 626). 
As theories are communicated, they may involve particular selections and misinterpretations of evidence.  Revisions of children’s theories of causal structure occur when they compete with counter-evidence to their theories.  The driving force for theory development, according to Gopnik (2001), is the search for truth.  Because they have greater explanatory power, more veridical theories, such as those of adults, are internalized by children and replace their own theories.

In the Oxford study (Beck, 2007) it was found that in tutorial practice, there is an integrated relationship between discussion and students’ narrative essays. The purpose of the discussion is to critique the students’ oral essays as read to the tutor. Lessons learned from discussion, whether about points of content or argument, are intended to be applied to future essays, and discussions. Thus, in this study our intent was to inquire into the role of discussions in contributing to writing (on the same subject), just as the mother-child discussions in Beck & Clarke-Stewart were assessed for their contribution to children’s subsequent storytelling.

Two theories were used in analyzing the Oxford tutorial:  Knowledge Building; the Tutorial as Dialogic Education: Questioning, Response, and Feedback Exchanges

Knowledge building: writing as an “improvable object”

A written presentation that is orally performed in a tutorial perfectly resembles our image of the teacher. Thus far we have described the social context of learning in the tutorial, but little of the content of the discourse. Without exception, the dons heralded writing as an essential feature of the tutorial. As writing is such an implicit part of the tutorial, no tutor felt the need to justify writing as a practice to increase students’ abilities to think for themselves. However, Ryan’s observation that students themselves use writing to understand what they know about a text is some evidence of the crucial role of writing: “knowing that he will not know what he thinks until he sees what he has written” (p. 79). Therefore, the very idea of “thinking for yourself” is embodied in writing. Not only is writing, obviously, the supreme method of communicating thought, but just as obviously it lays down a record of thought’s progress.

In his historical and linguistic study of the transition of oral discourse to writing, Olson (1994) analyzed several distinctive properties of writing, among numerous possible, that demonstrate its quintessential importance in education. “Literate thought is premised on a self-consciousness about language, for it is modern writing that provides a relatively explicit model for the intentional aspects of our language and so renders them conscious” (Olson, 277).  Because writing creates representations of thought, that are more precise and reliable than oral discourse, such as concepts, evidence, and arguments, it allows these forms to become self-consciously the object of further reflection, analysis and design (p. 266) and affords further discourse (p. 51). Writing is thus the unique pathway to attain the ability to think for oneself.
According to Scardamalia, Bereiter, and Lamon (1994), researchers at the Ontario Institute for the Study of Education (OISE), knowledge building requires an “improvable object” as the focus of the activity. The improvable object may be orally communicated ideas or an artifact, such as an essay prepared for a tutorial. The goal is to transform the “object.” We think that any scientific model of tutorial pedagogy will inevitably require concepts and methods that provide evidence of the serial transformation of the object.  Marjorie Reeve’s observation in this regard is apt: “it is to set the student the task of expressing his thought articulately, and then to assist him in subjecting his creation to critical examination and reconstructing it” (cited in Palfreyman, 2000, 7).
This assumption is elaborated by Wells (2001):

“…for the students, the goal of inquiry is making not learning, or as I put it above, working on an improvable object….Learning is an outcome that occurs because the making requires the student to extend his or her understanding in action—whether the artifact constructed is a material object, a demonstration, explanation, or theoretical formulation” (Wells, 2001, 190).

Because in tutorials, the student artifact, the essay, is the starting point, the outcome of the “making” is a refined essay, the object reconstructed. Nevertheless the “improvable object” is not limited to writing but also to speech. In tutorials, students are expected to improve steadily their oral arguments in the face of criticism. Therefore, a model of tutorial pedagogy would need to explain how both the written essay and/or its oral presentation are transformed during and across a number of tutorials.

The Tutorial as Dialogic Education: Questioning, Response, and Feedback Exchanges

If writing is the focal object to be improved, dialogue during tutorials is the method for improving writing and oral argument. In this section we emphasize that dialogue is more than questioning. Scardamalia, Bereiter and Lamon (1994) and Wells (1999), another (former) OISE educational researcher, use the concept of progressive discourse or dialogic inquiry, respectively, to characterize knowledge building during discussions (p. 7). From an analysis of a large corpus of classroom dialogues, Wells (1999) concluded that dialogic inquiry consisted of inquiry-response-evaluation (IRE) exchanges (also modeled as Initiation-Response-Feedback (IRF) exchanges by the Birmingham school of discourse processes), that offer evidence of a ubiquitous conversational structure, with variations, that contributes to knowledge-building in a wide range of educational settings. Because of the flexibility of turn taking in dialogues, either tutor or tutee may initiate and sustain several exchanges of IRE/IRFs. This may afford researchers a means of measuring the extent to which students are directing or co-directing tutorials. The proportion of direction over instruction taken by students or tutors may indicate the extent to which the equality principle is being followed and whether there is a co-teaching approach down in the trenches of the tutorial discussion.

Linguistic theorists (Halliday, 1993) and educational discourse processes analysts (Coulthard & Montgomery, 1981) also assert that exchanges of dialogue are the meaningful units to analyze in conversations involving information transfer. The interpretation of exchanges implies that any particular dialogic move be analyzed in terms of the preceding and succeeding moves. Only exchanges could adequately show how knowledge was co-constructed by participants in dialogue. We assume that this ubiquitous conversational structure, with complex variations, is used during Oxford tutorials and that analyzing these patterns could lead to a better understanding of the precise differences in the pedagogic principles we seek to document.

We have previously discussed the tutor’s use of questions, whether or not questions are designed to form a “web” or are simply methods for inducing students to think for themselves. Student responses to tutor questions, of course, provide material for evaluation and other forms of feedback. Consistent with the IRE/IRF model, Mash proposes that tutorials are exceptional settings for providing extensive feedback: “Tutorials should offer excellent opportunities for feedback that is positive (while always being honest), and the frequency of feedback should help the process whereby students settle in mentally and feel that, subject to the required effort, they can be successful” (p. 91). The only caution here is to recall that students probably experience feedback as assessment, which is one of their anxieties in the “no place to hide” syndrome. Mash’s statement indicates sensitivity to this issue. The extent and contingency of feedback in tutorials is surely one of its distinctive pedagogic features, particularly when contrasted with the less frequent and less expert feedback afforded by peers in small-group seminars. In a dialogic educational environment, the study of tutors’ variations in feedback/assessment should prove just as interesting as variations in questioning.

In the electronic discussion studies (Beck, Fitzgerald & Pauksztat, 2003b; Boyd, Fitzgerald & Beck, 2005) our intent was to investigate how the social organization of the group affected the discussions. In computer mediated communications, 8-person groups of teachers were asked to write behavioral observations and then respond asynchronously to two or more others with feedback. In these studies rubrics were used to rate the quality of discussions. Because we theorized that group hierarchy influenced discussions, we used models of hierarchy formation to study how different groups performed in their discussions.

Models of Hierarchy Formation
Several studies have demonstrated that, in unassigned, initially leaderless groups, a stable influence hierarchy and role system will emerge and persist after initial instability (Homans, 1950; Bales, 1970). What factors contribute to the development of status hierarchies? Strangers initially can use observed status characteristics such as age, gender, and race to establish leadership and influence hierarchies and, subsequently, these may be reinforced or altered through other processes. Burt (1999) contrasted the human capital explanation (people do better and hence attain higher positions in a status hierarchy because of their personal attributes such as knowledge, skill, and charisma) with the social capital explanation (people do better because they are better connected). Gould (2002) used a comparable dichotomy in summarizing groups of factors that lead to interaction hierarchies. He specified an individualist or market framework in which outcomes are unequal because individuals vary in qualities that have locally meaningful importance, such as talkativeness or confidence. Differentiation occurs because people make different contributions. This was contrasted with a social structural framework in which outcomes are unequal because of the quality of social positions one occupies, largely independent of personal qualities.
Gould (2002) has not been content to accept one or the other of these factors alone in explaining why hierarchies develop. He defined a hierarchy as “an emergent social process” (p. 1145), without assuming that it is a reflection of underlying qualities (individual or social). Outcomes result from a more decentralized, less purposive process. In this view, status rankings are stable due to the self-validating character of social judgments. For example, “if one individual attracts slightly more positive judgments than others because of some intrinsic quality, then the social influence process will set off a cascade in which this small difference is inflated as people react to one another’s reactions to its existence” (p. 1149). Thus, “collective adherence to socially provided assessments reproduces and thereby validates those very assessments” (p. 1148). It has also been argued that group members develop strong ties to each other because they recognize each other as well connected and central in the communication network (White & Houseman, 2002).

Which common status characteristics do group members use to form hierarchies? Sev’er (1991) reviewed three conceptual positions on how status is achieved in groups: dominance theorists (Mazur, 1985) argue that initial power contests produce a ranking of the interactants; performance theorists (Berger, Fisek, Norman, & Zelditch, 1977; Ridgeway, Berger, & Smith, 1985) argue that status is achieved based on beliefs of how well an interactant will perform a task on the basis of personal characteristics such as sex; and behavioral styles theorists (Sev’er, 1991) consider that cues that lead to estimates concerning task performance are influential. This position is consistent with Fisek, Berger, and Norman’s (1991) concept of a behavior interchange pattern. This concept was more recently described as “a set of interaction cycles . . . between two or more actors that are consistent in their power and prestige significance” (Wagner & Berger, 2002, p. 61). A limited amount of research has shown the influence of behavioral characteristics on hierarchy formation (Lee & Ofshe, 1981; Shelly & Troyer, 2001a, 2001b).

In the electronic portfolio study (Beck, Livne & Bear, 2005) our intent was limited to showing that teachers’ discussions contributed significantly to teacher portfolio development when compared with non-discussed portfolios. In fact, the teachers’ reflective narratives of their classroom experiences in teaching were enhanced by those teachers who shared and discussed their portfolios with other teachers in electronic dialogues. Portfolios critiqued through peer discussion were rated as more reflective.