They are coded separately from the simple act of requesting and adding new ..... [The experimenter gives them the excerpt and they read aloud]. Difficultâ¦.I think ...
EVALUATING THE QUALITY OF SCRIPTED DIALOGS REGARDING CONCEPTUAL LEARNING: BETWEEN THE COMPREHENSIBLE AND THE INCOMPREHENSIBLE Baruch B. Schwarz and Christa S. C. Asterhan The Hebrew University Research on concept learning in educational settings has traditionally been conducted in the Piaget-inspired cognitive conflict paradigm in which students’ naïve conceptualizations are confronted with anomalous data or contradicting views or are paired with peers who have different views (socio-cognitive conflict). Inducing conceptual change has been extensively shown to be a hard goal to achieve, especially in individual settings (Limon, 2001): Students have to become aware of their initial (mis)conceptions, understand the new information presented to them, become aware of the contradictions, compare and evaluate the different ideas, and adapt their misconceptions. We propose that the engagement in dialectical argumentation on one’s own and another person’s views by producing reasoned arguments in favour and against these ideas will facilitate conceptual change processes. We base this claim on the fact that dialectical argumentation intertwines a number of social and cognitive processes that are considered to promote concept learning: First of all, similar to the ‘selfexplanation effect’, the epistemic examination of one’s personal theories and the reasons behind them promotes understanding and knowledge construction processes (Chi, deLeeuw, Chiu & Lavancher, 1994; Kuhn, 1991). Furthermore, in dialectical argumentation participants are exposed to a multiplicity of ideas and encouraged to explore each other's validity. This implies that they have to consider objections to their personal theories and assumptions, to attempt to understand alternative positions and to formulate objections and/or counter-objections. The unique structure of argumentation that links premises, conclusions, conditions, and rebuttals is also thought to considerably improve and extend the organization of knowledge, leading to better recall and understanding on subsequent test occasions (Means & Voss, 1996). In theory, the engagement in solitary argumentative reasoning may be expected to yield similar results. However, argumentation is basically a social process that presupposes the presence of an audience (Leitao, 2000). The act of solitary argumentation on scientific concepts is likely to be cognitively too demanding. The presence of a dialog partner, on the other hand, may promote reflection and awareness to one's own beliefs (Amigues, 1988), cause learners to engage in explanatory activities (Okada & Simon, 1997) and reduce cognitive load through the personification of different alternatives. In a recent study we investigated the effects of argumentative dialog on concept learning in evolutionary theory in an experimental design (Asterhan & Schwarz, 2005). Undergraduates were assigned to dyads and collaboratively tried to solve an evolutionary phenomenon (i.e., the evolution of webbed feet of ducks). Half of these dyads were instructed to engage in dialectical argumentative dialog on their respective solutions and received some examples of argumentative moves within a dialog; the other half was merely instructed to collaborate. Individual evolutionary understanding was assessed on three test occasions: Prior to, immediately after and a week following
the intervention. When controlled for pretest performance, delayed posttest explanations of students in the argumentative condition were found to be of a higher quality, compared to those of control students. Furthermore, the pattern through which this advantage was attained revealed that students in both conditions improved their conceptual understanding immediately following the intervention. However, students who were merely instructed to collaborate lost this temporary gain, whereas students in the argumentative condition retained the same level of performance at the delayed posttest. In spite of these mean effects, not all experimental subjects attained conceptual change and not all experimental dyads engaged in a dialectical argumentative discussion as they were instructed to. This difficulty was expected since arguing about scientific issues is excessively difficult to sustain (Baker 2003). On the other hand the results of this experimental study indicate that proper instruction given as scripts to fulfill during discussion often yield "productive" dialogs. In this paper we compare the dialogs of successful and unsuccessful dyads and attempt to identify dialog features that distinguished between them. The results of these protocol analyses then formed the basis for designing a dialog script which was tested in a second experimental study. The script was based on those dialog features that were identified as crucial to concept learning in the first study. ANALYSES OF DIALOGS The sixteen dialogs analyzed in this study were conducted by dyads that were assigned to participate in the experimental condition of a study on the effects of dialogical argumentation on concept learning in evolutionary theory and whose dialogs could be transcribed (see Asterhan & Schwarz, submitted, for further details). Procedure in a nutshell All students participated in the following sequence of activities: (1) Individual pretest to assess prior evolutionary understanding; (2) Instructional intervention: screening of instructional movie excerpt on evolutionary theory; (3) Collaborative intervention during which dyads solved two evolutionary problems, one warming-up and one transfer item; (4) Individual delayed post-test administered a week later. Pre- and posttests were administered to individuals in group format as paper-and-pencil tests. The dialogs analyzed are those on the transfer item in which subjects were asked to explain the evolution of webbed feet of Ducks. All experimental dyads received written instructions to engage in dialectical argumentative dialog on their respective solutions and some examples of argumentative moves. Experimental dyads also received a short excerpt of a critical discussion of four turns between two (hypothetical) subjects which, they were told, had participated in the experiment a year earlier: A: Then the ducks had to change their feet so that they could swim. The area was flooded with water, and because of the new environment webbed feet developed. B: What do you mean “developed”? How did that happen? A: Hmmmm. In the beginning they did not know how to swim. But slowly they learned to do it and that caused some sort of development in their feet. I mean, webs developed between their fingers. And that’s how it was passed on to the next generation. B: Well if that were true, then Olympic swimmers should also develop webbed feet, since they also swim all day long?!
The excerpt was handed to them while they had been working on the target question (the “Duck” question) for at least 30 seconds, so as to allow them to at least articulate their own solutions to the Duck item. They were told that the experimenter ‘forgot’ to give them the item in time and that they had to read it and try to relate to it in their discussion. The discussion in the excerpt modeled a critical discussion on the Ducks item without actually revealing or hinting at the correct solution. The comprehensible: Identifying and quantifying dialog features The dialogs were analyzed on two different level of granularity: Identification of dialogical moves (micro) and characterization of the interaction as a whole according to a number of dimensions (macro). We will shortly present the categories here, but further details can be found in another publication (Schwarz & Asterhan, in press). Micro-analysis: Dialogical moves . It is theoretically important to distinguish between instances of explanation development and between discussing the epistemic status of these explanations (Baker, 2003). Whereas both processes may be important in knowledge building, only the latter involves argumentation. The micro-analysis scheme does not consider the canonical validity or strength of a contribution, only its function within the conversational context and speaker's intent. It comprises a number of non-exhaustive, but mutually exclusive categories that assess dialogical moves involved in the two above-mentioned processes: argumentation and explanation development which are summarized in Table 1. Table 1 gathers argumentative and explanation development (or elaborative) moves: 1) Argumentative moves: The aim of this group of categories is to assess the elements of dyadic (collaborative) argumentative reasoning. Whereas in argumentative discussions on societal and/or moral issues the level of reference is a speaker's position on a certain issue, in a scientific learning activity such as ours, the object that is discussed and disputed is the most appropriate explanation to the phenomena. We assessed the following argumentative moves: Claims (which in this case is a proposed explanation of the Duck phenomenon), Request for claims, Simple Agreements, Supports, Challenges, Rebuttals, Simple Oppositions and Concessions. 2) Elaborative moves: A contribution is assigned to this group when it contains an addition to or expanding of the information at disposal, provided that this addition cannot be labeled as argumentative in nature. It includes both instances of elaboration and of providing and requesting information. Elaborations are contributions that build upon the content of a preceding contribution to the discussion in a significant way, for example, by extending it, making the idea more explicit, developing it, interpreting it, providing an implication based on it. They are coded separately from the simple act of requesting and adding new information that does not build upon the content of preceding contributions, such as when one of the discussants asks of the other to clarify a previous contribution. Acts of simple repetition or correction of the content that appeared in preceding contributions without significantly adapting it or adding new information to it (such as, for example, paraphrases, corrections of mistakes) are disregarded.
Category (Code)
Description of category
Claim (Cl) Req. claim (ReqCl)
a proposed explanation x (or part of it) to the ducks phenomena request for a solution / explanation, or request for an evaluation of a proposed explanation (or part of it) or to take a stance towards an explanation that has been proposed any verbal, reasoned utterance that is intended to strengthen the epistemic status of an explanation x. overt verbal utterances of unreasoned agreement, simple reconfirmation of the correctness of (part of) a certain explanation x, provided that it is embedded in a noncritical preceding constellation (as opposed to a concession) any verbal, reasoned utterance intended to weaken the epistemic status of a solution x.
Support (Su) Agreement (Ag) Challenge (Ch) Opposition (Op) Rebuttal (Re) Concession (Co) Req. info (ReqIn) Information (Info) Elaboration (El) Repetition (Rep)
Table 1.
overt verbal utterances of unreasoned disagreement, simple opposition to (part of) a certain solution x without providing any further justifications / reasons of why they think so. response to a "explanation(x)-challenge(to x)" chain that is intended to strengthen the epistemic status of explanation x by weakening the challenge to x any overt verbal, unreasoned expression of agreement in a critical constellation, that is: when the content the discussant agrees to was previously opposed by him/her request for further information or clarification, non-critical when pure information is provided (usually in response to a question ) so as to clarify something that is not clear, or provide information about a subject unknown to the partner one of the discussants builds upon the content of own or other's previous turn in a coconstructive way, he is transacting on own or the partner's idea in the previous turn (s) or immediate preceding discussion, develops it, continues it the speech repeats the content of previous turns, does not include any new information or inferences compared to preceding discussion content
Categories of different dialogical moves
Macro-analyses: Interpersonal and cognitive properties of the discourse. The aim of this coding scheme was to examine certain macro-properties of students' dialogs that could be accountable for differences in subsequent individual learning gains. Operationally, we relied on the identification of these features in the dialog as a whole and segmentation was, therefore, unnecessary. We categorized each dyadic interaction on a number of cognitive and interpersonal features: 1) Cognitive properties of the dialog: Argumentative structure of the dialog: This category assesses the argumentative nature of the interaction as an activity. Two types of argumentation were identified (see Asterhan & Schwarz, submitted, for more details). a) Dialectical argumentation : The dialog contains more than one solution which Ss feel obliged to choose from, or the dialogue contains a single proposed solution that is both contested as well as defended (i.e., Ss consider both the pros and the cons of that solution). Note that this interaction may have been adversarial in nature, or not.
b) One-sided argumentation: The dialog contains only one proposed solution and Ss provide justifications and explanations for the reason why they think that this solution is correct 1 . Darwinian solution: Whether a Darwinian explanatory schema (which refers at least to intra-species variation and natural selection) is mentioned at least once during the interaction. Key issues of change: Whether students discussed the key issue of how the ducks' feet could have changed from "feet like those of pigeons" to "webbed feet". 2) Interpersonal properties of dialog: Interpersonal distribution of solutions: It has been theorized that one of the advantages of argumentative group discussions is that it enables the objectification of perspectives and their representation by actual persons defending them (Baker, 2003). We, therefore, included a coding dimension that assesses whether the different solutions were proposed and represented by different persons, or whether they were not distributed among the interlocutors. Note that a dyad may conduct a dialectical argumentative discussion without their views being distributed among the dyadic partners. For example, they may collaboratively explore the pros and cons of a number of collaboratively proposed solutions, or the same interlocutor that proposed a solution may also be the one to attack that solution. We, furthermore, distinguished between two different types of distributed dialogs: Those in which interlocutors overtly confront the different solutions and relate to the differences between them (Distributed: confrontational), and those in which they do not overtly discuss these differences (Distributed: not confrontational). For example, the dyadic partners may each mention a different solution at some point in the dialog, without actually discussing and comparing the two. In such a situation, the different views are personified by different persons, but no attempt is made to resolve this incongruence. Whereas the former may be important in its own right, the collaborative attempt of resolving may prove to be crucial for conceptual change. Closure: Whether the dyad agreed upon a certain solution at the end of the dialog or not. Symmetry: The extent to which the dyadic partners contributed evenly to the discussion, or not (Baker, 2002). Operationally, symmetry was defined as the number of significant contributions that were proposed by the most contributive partner in the dyad, divided by the total number of such statement units in the dialog. Significant contributions are those statement units that contain newly asserted information or, in other words, any of the contributions categorized as claims, supports, challenges, rebuttals, information and elaborations.
1
Note that in the current design a student's rebuttal that is proposed in reaction to the challenge that appeared in the written excerpt they were given does not necessarily change the nature of the argumentative discussion to a dialectical one, even though the move by itself is a dialectical move. If Ss proposed only one solution, provided supports in favor of that solution and responded to the excerpt with rebuttals that were intended to strengthen the epistemic status of that same solution, then the dialog is still one-sided in nature.
Results In nine of the sixteen dyads at least one of the students attained conceptual change (see Schwarz & Asterhan, in press, for details on how conceptual change was defined). The dialogical features of these nine dyads were then compared to those of the seven remaining dyads in which none of the dyadic partners attained conceptual change. Analyses on features of the dialogs' microstructure were conducted with the help of Mann-Whitney tests for non-parametric samples and are presented in Table 2. Table 2 shows that the dialogs of dyads in which at least one dyadic partner attained conceptual change were characterized by a larger number of (requests for) claims, challenges, rebuttals, concessions and provision of information. However, of these differences only the number of challenges was significant. More importantly, the dialogs of the two different types of dyads did not differ on number of elaborations or on number of supports. Since the average length of dialogs was not significantly larger for any of the two types of dyad (see Table 2), analyses that were conducted on the relative (instead of the raw) frequencies of the dialogical moves yielded similar outcomes. They are, therefore, not presented here.
Dialogical moves Claim Request claim Agreement Support Challenge Rebuttal Opposition Concession
Type of dyad Gaining dyads Non-gaining dyads (n=9) (n=7) 2.00 (.87) 1.43 (1.33) 2.22 (1.71) .71 (.76) 5.89 (3.89) 5.57 (3.78) 4.11 (3.51) 3.86 (2.85) 5.33 (1.33) 1.43 (2.51) 3.56 (1.33) 2.85 (3.86) 1.00 (1.12) .57 (.98) 1.00 (1.32) .29 (.49)
Elaboration 7.33 (3.84) 8.43 (5.47) Request information 1.11 (1.27) .86 (1.07) Information 1.33 (1.87) .29 (.49) Total moves that testify of 10.89 (5.84) 4.86 (3.44) dialectical reasoning a Total moves that testify of non17.33 (8.87) 17.86 (8.63) dialectical reasoning b Total number of turns 2.44 (2.96) 1.14 (1.46) a Total number of oppositions, challenges, rebuttals and concessions in a dialog. b Total number of supports, agreements and elaborations in a dialog
p value ns ns ns ns p < .05 ns ns ns ns ns ns p < .05 ns ns
Table 2. Mean number of different dialogical moves (and SD) for dyads of which either at least one or none of the partners attained conceptual change
In light of these findings, we then compared dyads on two aggregative scores, each considered to capture moves that testify of different types of reasoning: (1) Total number of dialogical moves that testify of dialectical, or critical, reasoning (consisting of challenges, oppositions, rebuttals and concessions); (2) Total number of dialogical moves that testify of non-dialectical reasoning, or the construction and solidification of an explanation in a consensual, non-dialectical manner (consisting of supports, agreements and elaborations). Whereas non-gaining and gaining dyads equally
engaged in non-dialectical reasoning, gaining dyads were found to show a larger number of dialogical moves that testify of dialectical or critical reasoning. Analysis of macro-properties of dialog All sixteen protocols were analyzed according to the macro-analysis coding scheme. Table 3 shows comparison of the two types of dyads on each of these macrodimensions. Table 3 shows that interactions that resulted in conceptual change for at least one of the dyadic partners were more likely to be characterized by interpersonal distribution of explanatory schemas among discussants, were more likely to be of a dialectical nature and its discussants were more likely to have mentioned crucial elements of Darwinism at some point in the discussion. However, the dialogs of gaining and non-gaining dyads did not significantly differ on whether the partners reached closure at the end the discussion and whether they discussed the critical issue of how the duck's feet could have changed. In addition, a Mann-Whitney test for nonparametric samples showed that significant contributions were more evenly distributed among peers in dialogs of gaining dyads, than in non-gaining dyads.
Dialog characteristic Interpersonal distribution of schemas Key issue of change Darwinian answer mentioned Closure at end of discussion Argumentative macrostructure
Not distributed Distributed - not confrontational Distributed - confrontational Not discussed Discussed Not mentioned Mentioned No closure Closure One-sided Dialectical
Type of dyad Non-gaining Gaining dyads dyads (n=9) (n=7) 0 6 4 1 5 0 0 3 9 4 2 7 7 0 1 1 8 6 0 5
9
2
χ2 (N=16)
12.75*** 4.75 9.68*** .04 9.35**
Table 3. Macro-properties of dialogs among 'gaining' and 'non-gaining' dyads
Discussion In one-sided argumentative dialog, students not only propose and develop a solution, but also provide justifications and explanations for the reason why they think that this solution is correct. In the literature, such processes have been found to lead to substantial learning gains by themselves, similar to the self-explanation effect, for example. The micro- and macro analyses in our study, however, revealed that onesided argumentation and dialogical moves that testify of non-dialectical reasoning were by themselves not found to relate with learning gains. None of the dyads that engaged in one-sided argumentation attained conceptual change. On the other hand, dialectical moves were found to relate with conceptual change and the macrostructure of gaining dyads’ dialogs were all characterized by dialectical argumentation. The reason for this striking finding may lie in the paradigm according to which the learning task was designed, namely the socio-cognitive conflict paradigm. Learning within the context of such tasks may particularly benefit from dialectical argumentation, especially when the desired outcome concerns a big leap
from one conceptual schema to another. For this to occur, students will have to consider more than one solution, compare them and evaluate their respective weaknesses and strengths. The features that we identified in productive dialogs, for example the fact that they are characterized by dialectical argumentation, and other characteristics identified in another publication (Schwarz & Asterhan, in press) should not suggest that such characteristics explain the emergence of new understandings concerning evolutionary theory. Table 3 suggests different kinds of emergence that do not necessarily fit a socio-cognitive conflict paradigm in the sense that they are not necessarily characterized by overt inter-personal conflict. This in contrast with argumentation on moral/societal issues which is often characterized by a certain amount of interpersonal conflict, not in the least because participants usually represent and defend personal opinions instead of ‘neutral’ scientific explanations. In the following section we depart from the search for characteristics to understand through the protocol of one particular dyad, how evolutionary concepts could develop through the dialogue. We will see that the identification of the characteristics of the dialogs of productive dyads did not render the magic and somehow the incomprehensible that often accompanies conceptual learning in dyadic interaction. The Incomprehensible: A Qualitative Exploration We present here the protocol of one dyad, A and B who both showed conceptual gains in the post-tests. The last column indicates the types of their dialogical moves. 1 2
B A
3 4 5 6
B A B A B B
7 8 9 10
A B
11
12
A
13 14
B A
15 16
B A
17 18
B A
Okay. Do you have an idea? Ah, yes. Because the area was flooded with water and those with legs that suited the water actually survived. Humhum. And then those that survived, developed and continued themselves. That is, that ducks developed webbed feet especially to survive-Yes-in areas that once were dry and become flooded with water? Yes, that sounds reasonable.
ClA(1) ElA(1) ClB(2) AgA(2) ElB(2) AgB(2)
What can we say against this? [The experimenter gives them the excerpt and they read aloud] Difficult….I think that-If it is a matter of survival, then, ehhh, evolution will not occur, because swimmers, they do not need it for their survival.
ChB(2)
It is something that nature feels that something has to happen, or the creature feels that something has to change for it to survive, only then will the change take place. That's amazing! The question is, ehh, whether the change is biological or not? I mean, one survives and the rest does not, so one develops- You mean like a mutation? Cause there are all kinds of animals. By chance one kind is well adapted to the new situation and that kind survives and continues itself. Yes I do not understand how he understood that- suddenly he says that they developed webbed feet So how do you think it happened, that it happened overnight? That one by chance had something similar to webs. He survived, and…the
ElB(2)
ReB(2)
ChA(2) EloB(2) ElA(1) AgB(1) ChA(2) ReqCl(1) El(1)
19 20 21 22
B A B
23 24 25
B A
26 27 28 29
B A B A
33 34 35
B B A B A B A B A
36 37 38 39
B A B B
40
A
41
B
42 43
A B
44 45
A B
46 47 48
A B A
49 50
B A
51
B
30 31 32
52
webs just developed [evolved], become more sophisticated. And what if-Not something out of nothing! And what if no one had it? Then they would not have survived.
rep Ch(1) Re(1)
How could they have survived? Maybe they just developed it somehow? The question is, development of the type "something out of nothing"…. or something that was already there. Humhum. I do not know whether they ever proved that- It says here that it is a change that occurred in their feet. There was no-The question is whether some already existed. Maybe there was something similar to it that-to webbed feet. Well, it is possible that…for a small number-yes!-of individuals that-had- something-a bit more developed. They survived And then they continued the next generations. Yes, because it does not seem reasonable to me that an animal suddenly develops webs. That was not there. Yes. That really does not sound reasonable. Difficult. Yes, but, hmmmmmm. Okay. That sound logical to me. Do you think that if there was a person that had a bit of webs between his fingers, a human creature, then , ehhmm, and if we were to let this human being reproduce, then he would continue this or that it would come off in the next generations? No! Of course, when all the rest of the people would not have it and he would be the only one to exist, he is the basis for the reproduction, so obviously it, ehhh, his offspring would be ;like him, he is made of the same materials in fact. But maybe these things, when they are at the start of their development, say something like theoretically a man with webs between his fingers, maybe they will develop only when they are needed for survival. And when they are not needed then they will go extinct with the following generations. Development [evolution] just like that-By accident, by accident, mutation. Like, haven't you ever seen someone with a problem, say something a bit different-yes-with his hand. I think that even if it is useful in some way, for the swimmer it could be useful, it will not survive and it will not survive and it will not pass to the next generations. Ahhh, but you see mutation, you see it as a flaw. No, mutation in a positive sense in this case. But something completely different. So that's it, what is "completely different"? It is something a bit different, but not… not something that is one of a kind. -hummmmI mean, some part like this and another part like that, some sort of gap, not necessarily something really exceptional. That is really, that is really not that likely. But do you agree with me on the issue that-…
Ch(2) Re(2) Ch(2)
that is somehow, that there are some individuals within a group that have a slightly different development, like, for example, those ducks that you imagined to be slightly closer to a flipper, ehhhh, they had feet with some webs,
rep
rep cont Con(1) Ag(1) cont cont El(3) El(3) rep Ag(3) Con(3) ReqCl(3)
Su(3) Ch(3)/ El(3) rep Su(3) El(3) Ch(3) Re(3) El(3) cont ReqCl(3)
53
A
54 55 56 57 58 59
B A B A B A
60 61
B A B
62 63
B A
some initial webs somehow, do you agree with me that that would not have developed if they would not have needed it for their survival? The question is, if, say, if they, that is, if all of the others would have died and they would have survived, then it would have developed [evolved]. Because of genetic transfer? Yes, genetic transfer, it would be passed on. Humhum. If they would have been the only ones that would have survived somehow… Yes. Okay. Good. Because not- the mind does not decide whether it has- I mean why, I want to, since[unclear] -things that happen in nature. Yes [both laugh] Okay. So let's write that down? We'll try.
Ag(3) El(3) Ag(3) rep Ag(3) Su(3) rep
This protocol shows that the discussants begin their discussion by complying with the script proposed: They present their reasoned opinion concerning why ducks have webbed feet. The argument is first presented by discussant A (Turn 2), and then elaborated (Turn 4). The argument of discussant B in Turn 5 is presented as an elaboration (although it is a different idea) and is agreed upon, like the first argument by his peer (Turn 7). In turn 8, B continues complying with the rules suggested in his script as he declares "What can you say against this?". The experimenter proposes at the same moment the protocol which is intended to suggest new ideas. However, this "scaffold" and the script are used by B to express a rebuttal of his own (turn 10) and to elaborate it. This rebuttal is continued as a challenge to the second idea, the fact that the change (webbed feet) is biological and not environmental, which is beautifully expressed in Turn 12 "one survives and the rest does not, so one develops" and B's elaboration of this challenge by "You mean like a mutation?" Like for many other ideas in this protocol, this idea does not result from the confrontation of two ideas but is a real creation. What is striking in the rest of the protocol is that although the peers have different initial views, they help each other to find a more satisfactory solution: In turns 30-34, they collaboratively develop the basic principles of evolutionary theory, variability in population, and selection. The dialog is of course dialectical, not controversial, and distributed. But apart from that, we really wondered how these students created such ideas. For example the challenge/elaboration in Turn 41, "But maybe these things, when they are at the start of their development, say something like theoretically a man with webs between his fingers, maybe they will develop only when they are needed for survival. And when they are not needed then they will go extinct with the following generations" brings the new idea that evolution depends on needs of the entire population. We wondered how B created such a challenge to the idea that special humans would progressively (in further generations) have webbed feet. In summary, we showed here the limitations of a characterization of productive dialogues to show that for dyadic interaction to lead to conceptual learning, creativity is needed. Somehow the design of the task proves to be a fertile soil for this creativity to burst: The script we proposed engaged many dyads in dialogical dialectical argumentation, and led some to create new arguments. In the second experiment we
present here, we show how a new design of the Ducks task (by changing its script) led to even better results concerning conceptual learning. An Experimental Study: Effects of Scripted Argumentative "Monolog" Our dialog analyses stressed the importance of dialectical argumentation in learning tasks based on the socio-cognitive conflict paradigm. In the first experiment, dyads were elicited to engage in dialogical dialectical argumentation through written scripts. However, not all experimental dyads engaged in dialectical argumentation, whereas some control dyads spontaneously did (Asterhan & Schwarz, submitted). In the following section we will present results from an experimental study with a similar overall design and procedure, but with some changes to further isolate the element thought to be responsible for the difference in cognitive gains: engagement in dialectical argumentation. A confederate played the role of one of the participants in both conditions. Participants in the experimental condition were prompted to engage in dialectical argumentation on their own and the confederate's solution, by answering structured questions read aloud by the confederate (who chose a note in an urn and invariantly picked up the role of the 'reader'). We call this activity monological argumentation. In the control condition, subjects and the confederate only read aloud their solutions to each other, without discussing them further. Thus, students in both conditions were withheld from conducting a dialogical discussion with each other. Procedure in a nutshell The overall procedure was similar to that in the prior study, except for the fact that Ss engaged in a scripted, monological argumentation with a confederate, instead of a dyadic discussion. Immediately after watching the movie excerpt, each one of the total of forty-four students was asked to solve the Ducks item. They were then introduced to the confederate who, they were told, had just completed the same stages of the experiment as they had. Ss in the experimental conditions were told that they would participate in a short collaborative task with clearly defined roles (reader and respondent) and instructions. The confederate chose a note in an urn and invariantly picked up the role of the 'reader'. The task itself consisted of nine steps that were read aloud by the collaborator; six of these steps prompted the participant to engage in dialectical argumentative reasoning on their own and the confederate's explanation of the Ducks item; three of these steps contained procedural instructions (see Appendix). The design of the task was based on the features that were found to be critical in the previously presented dialog analyses. The average length of scripted dialog was 8:05 minutes (ranging from 5:50 to 13:40). In the control condition, Ss were also seated with the confederate. They were first given seven minutes to complete a short, neutral task to control for time-on-task in both conditions. Their sheets were then "returned" to them and they were requested to read their respective answers to each other while refraining from any further verbal communication, before answering the Ducks item once more on a new answer sheet (T2). In sum, the conditions were identical on factors such as, social facilitation, exposure to content matter (whether provided by the dyadic partner or from the dialog excerpt in the first study) and personification of viewpoints, and only differed in the
elicitation of monological dialectical argumentation. Since the confederate and her behavior were controlled, the unit for all statistical analyses on cognitive gains was the individual.
Mean explanatory schema score
control
experimental
5 4 3 2 1 0
T1
T2 Test occasion
T3
Figure 1. Mean explanatory schema scores by test occasion and experimental condition
Results Nineteen of the twenty-two experimental subjects proposed at least one supporting reason for their own solution and an identical number of students proposed at least one objection. The number of subjects that mentioned at least one supporting argument for or at least one objecting argument against the confederate's solution was smaller (12 and 16, respectively). Two subjects only provided support for their own solution without further criticizing their own or the confederate's solution. All other subjects engaged in dialectical argumentation, as defined in the dialog analyses section. The mean number of supporting arguments for own argument (M = 1.22, SD = .75), objections to own (M = 1.27, SD = .83), supporting arguments for other (M = .59, SD = .59) and objections against other (M = .91, SD = .87) were all significantly larger than zero (p < .001). Mean explanatory schema scores were then calculated for pretest, immediate posttest and delayed posttest occasions and are presented by condition in Figure 1. A one-way analyses of variance (ANCOVA) was conducted on the delayed posttest scores, while controlling for subjects' pretest scores and formal schooling in Biology as covariates. Experimental subjects showed superior conceptual understanding, F (1, 38) = 8.87, p = .005, η² = .19. Previous schooling in Biology was not found to have an effect (F (1, 38) = 1.13, ns and F (1, 38) = 8.41), whereas pretest scores were a significant predictor of delayed posttest performance (F (1, 38) = 35.30, p < .001). Furthermore, repeated analysis of conceptual understanding at different test occasions and simple comparisons between test occasions replicated the pattern of retained conceptual gains for the experimental condition, and yielded lack of gains for the control condition. Conclusions and general discussion The second experiment yielded very positive results concerning explanatory schemata, thus concerning conceptual learning. Of course, it is impossible to compare gains between the first and the second experiment according to methods drawn from
inferential statistics, but the tendency was to lead to even stronger effects for the second experiment. As aforementioned, the script in the second experiment was designed according to findings concerning micro and macro characteristics of successful dialogs from the first experiment. Such findings have several important consequences. The first one concerns the important of scripts: (a) telling to people to discuss an issue, (b) to accommodate divergent views, (c) to convince each other, or (d) to adopt a critical stance in discussions, deeply impinges on the discussion in very different ways. The second lesson we can draw is that with proper scripts dyadic interaction can lead to conceptual learning (Of course the results were tested in few studies only and additional efforts should be invested to check different kinds of concepts, and different populations). The second was more successful but less creative. A third lesson concerns instructional models for promoting conceptual learning. The dialogical and monological models worked well in this endeavor, and we should admit that they provide two alternative approaches. As researchers engaged in education, the question is then ethical: Do we have to choose between fostering dialogical and monological settings? Should we promote dialogical settings at the detriment of monological ones? This is not the place to discuss such issues, but we should say that, based on the present findings, and our pedagogical experience, we opt for both kinds of practices. Of course, dyadic interaction often uncovers extraordinary processes that intertwine social and cognitive processes in the emergence of explanatory schemata. The collaboration between peers with different views in a distributed, non adversarial and dialectical way shown in the present protocol is only one among many examples in which dialog went far beyond a simple accommodation of ideas. This was a cocreation. But the Buberian stance we adopt in our group at the Hebrew University (e.g. Schwarz & Glassner, 2003) should not deny the importance and effectiveness of monological practices. REFERENCES Asterhan C. S. C. & Schwarz, B. B. The effects of monological and dialogical argumentation on concept learning in evolutionary theory. Submitted to the Journal of Educational Psychology. Baker, M. (2003). Computer-mediated interactions for the co-elaboration of scientific notions. In: J. Andriessen, M. Baker, & D. Suthers (Eds.), Arguing to learn: Confronting cognitions in computer-supported collaborative learning environments. Utrecht: Kluwer Academic Publishers. Chi, M. T. H., deLeeuw, N., Chiu, M., & Lavancher, C. (1994). Eliciting selfexplanations improves understanding. Cognitive Science, 18, 439-477. Kuhn, D. (1991). The skills of argument. Cambridge: Cambridge University Press. Leitao, S. (2000). The potential of argument in knowledge building. Human Development, 43, 332-360. Limon, M. (2001). On the cognitive conflict as an instructional strategy for conceptual change: A critical appraisal. Learning & Instruction, 11, 357-380. Means, M. L. & Voss, J. F. (1996). Who reasons well? Two studies of informal reasoning among children of different grade, ability, and knowledge levels. Cognition & Instruction, 14, 139-179.
Schwarz, B. B. & Glassner, A. (2003). The blind and the paralytic: Supporting argumentation in everyday and scientific issues. In: J. Andriessen, M. Baker, & D. Suthers (Eds.), Arguing to learn: Confronting cognitions in computer-supported collaborative learning environments. Utrecht: Kluwer Academic Publishers. Schwarz, B. B. & Asterhan, C. S. C. (in press). Argumentation among peers, dialogue characteristics, and conceptual change in evolutionary theory. In M. Chi (Ed.), Talk, Dialogue and Learning, Lawrence Erlbaum Associates. APPENDIX Nine argumentative and procedural steps that were read aloud by the confederate in the experimental condition 1) Please read me the answer you wrote down on the Ducks question. 2) I am not sure whether that really explains the Duck phenomena, or maybe I did not understand you well. Why do you think that this is the correct answer? Or in other words, what are your justifications, what is the strength of this solution that you proposed? Prompts: Why do you think so? Is there anything else you would like to add? 3) Okay. Now try to take a critical stance towards the solution you proposed. What could be the weak points in this explanation? What elements could be considered unresolved or illogical? Prompts: Why do you think so? Is there anything else you would like to add? 4) In the Duck item we were asked to explain the mechanism of change that occurred in the Duck's feet. Does your solution satisfy this requirement? How does it explain the process of change? Prompts: Why do you think so? Is there anything else you would like to add? 5) Now I will read you the solution I wrote. 6) What do you think about this solution? Prompts: Why do you think so? Is there anything else you would like to add? Repeat steps 2-4 if needed. 7) After all we just discussed, what is your opinion on the matter now? I mean- how did the change in the Ducks occur? 8) Do you have any doubts concerning the correct explanation, or evolution in general? Are there still any elements that remain unresolved or unclear to you? Prompts: Is there anything else you would like to add? 9) Now we will both answer the Ducks question one more time on the clean answer sheets each of us received. Please answer in a detailed, explicit and elaborated manner through relating to all the relevant ideas and concepts.”
