Introduction

Reflection on morality, on what is or is not acceptable thought or behavior, lies at the heart of every human society. For philosophers, theologians, and, more recently, for psychologists and anthropologists, morality constitutes a key concept for understanding human judgement and conduct. However, there is no unanimity about the definition of morality, as it is used in two distinct senses.

The first, descriptive, sense refers to certain codes of conduct applied by a society or a group, or accepted by an individual for her own behavior. The second, normative, sense refers to a code of conduct that, given specified conditions, would be applied by all rational people (Gert & Gert, 2020). According to the former, morality refers to ‘right’ and ‘wrong’ judgements and conducts. Moral issues are considered on the level of particular social groups (societies, communities) and systems of ideas (politics, religion) in history, sociology, and anthropology. In its normative sense, morality concerns ‘good’ and ‘bad’ judgements and conducts. In (especially developmental) psychology, where the normative sense prevails, morality is considered principally in terms of “universal” values, judgements, and behaviors assumed to be shared by people from any culture (Schwartz, 1994).Footnote 1 While some have argued that the two senses of morality are unbridgeable (Churchland, 2011), psychologists like Haidt (2011) hold that different members of a society can and do take different features of morality to be superior to others.

Moral education has been primarily influenced by Kohlberg’s (1984) theory of moral development, which relies on a normative sense of morality. Accordingly, moral education has favored values such as justice, care, and respect. The moral development field has also been heavily influenced by Martin Hoffman, who focused on the affective – rather than the cognitive – dimension, especially emphasizing the development of empathy (Hoffman, 2000). Both Kohlberg’s and Hofmann’s universal and linear developmental trends have been subject to critique (e.g., Jia & Krettenauer, 2017; Narváez, 2013). A major attack on Kohlberg concerns his focus on moral reasoning skills that have not been shown to predict moral action (e.g., Blasi, 1983; Talwar, 2011). Such findings are problematic for moral education.

In the last decades, moral education has undergone a shift from a ratio-oriented to a socially-oriented morality (e.g., Noddings, 2013). This reality favors stressing the difference between ‘good’ and ‘bad’ in judgement and conduct, as agreed upon by a certain community or social setting, rather than the difference between universal ‘right’ and ‘wrong’ in judgement and conduct. The terms ‘morality’ and ‘ethics’ have been used interchangeably for a long time. The new sense given to ‘ethics’ – the standards of ‘good’ and ‘bad’ distinguished by a certain community or social setting – allows to distinguish between the two senses of morality, as the term ‘ethics’ refers to a descriptive sense, and the term ‘morality’ solely refers to a normative sense.Footnote 2 This terminology enables to describe the above shift as a shift from moral to ethical education. Accordingly, ethical learning can be defined as any advancement in, or coordination between, ethical thinking and ethical conduct that are concerned with socially-oriented values. As we will see in the next section, this definition bridges between an Aristotelian approach to ethics, according to which virtues stem from actions, and a Kohlbergian approach to morality, which focuses on moral thinking, especially moral reasoning. While this definition of ethical learning may sound narrow, no other definition has been agreed upon so far. Moreover, this definition reflects the abovementioned general shift in education.

In this paper, we study the role of an educational program in coordinating ethical thinking and conduct. The program reflects the shift from a ratio-oriented to a socially-oriented morality. We focus on three socially-oriented values: tolerance, empathy, and inclusion. Granted, there are many other values that could be considered (e.g., social justice or equity), but the specific values we focus on were decided upon by the EU in the program in terms of which this study was conducted, as will be elaborated upon below. It is telling that tolerance, empathy, and inclusion are commonly used in official texts in educational organizations (especially in Europe), contrary to the terms ‘justice’ and ‘respect’, which were omnipresent in the past, but have grown less frequent. This change echoes the shift from moral to ethical education.

Indeed, Kairė et al. (2021) find that the use of the terms ‘tolerance’, ‘empathy’, and ‘inclusion’ in official texts in national educational institutions in seven European countries is extremely common. Tolerance is central to the 2014 UNESCO report on Learning to Live Together, and more than merely ‘putting up with’ alternative views, we view it as “an absence of prejudice, racism or ethnocentrism” (Rapp & Freitag, 2015, p.1033), and a capacity “to maintain ongoing relationships of negotiation, compromise, and mutuality” (Creppell, 2008, p.351). We define empathy as the ability to understand and experience another person’s emotions (Evans, 2001), and, relying on Buber (1957), we view inclusion as valuing diversity, fostering collaboration, respecting others, and enabling their active participation in shared events.

According to Shady and Larson (2010), the use of these terms in education indicates philosophical influences, especially the interpretation of Buber’s work: Tolerance might still present more of an ‘I-It’ perspective (Buber, 1947) – namely, tolerance does not simply imply holding back the (opinion of the) other from oneself. Empathy and inclusion are considered as deeper commitments to living together: Buber (1957) argues that in empathizing with other viewpoints, it is crucial that one’s own position is not lost, and he turns to the concept of inclusion to promote ‘genuine dialogue’ “where each of the participants really has in mind the other… and turns to them with the intention of establishing a living mutual relation” (Buber, 1947, p.22).

In this paper, we hypothesize that new general pedagogies have the potential to foster coordination between ethical thinking and behavior. Dialogic pedagogies are leading representatives of the educational change occurring in many schools worldwide. They emphasize the expression of voices in the public sphere, alongside the importance of listening to them. The development of the dialogue intertwines these voices as every utterance is related to another, and conversational norms are followed to create a cohesive flow where individuals respond to one another (Bakhtin, 1981). Dialogic pedagogies which were first mostly teacher-centered (Alexander, 2005), now favor collaboration in small groups, as well as critical discussions (Schwarz & Baker, 2016).

Our general goal is to explore the study of ethical learning in a context of dialogic pedagogy. Research on moral education and dialogue is abundant, but there is no literature on moral (or ethical) learning within dialogue, which emerges within social interactions. Nevertheless, important theoretical advancements have been made to suggest that conversations may boost moral development (e.g., Nucci, 2014). Also, some researchers have shown that transactive patterns of interaction in dialogues foster moral development (e.g., Berkowitz et al., 2008). However, the above studies have considered moral development as an outcome of participation in dialogues, not as a process occurring in the course of dialogues. The fine-grained focus we suggest is new and necessitates innovations in the methodological realm. We thus elaborated an analytical framework – the Dialogue on Ethics/Ethics of Dialogue (DoE/EoD) framework. Dialogue on Ethics reflects thinking (conceptualization, reasoning, judgement) about ethical concepts expressed in dialogue, while Ethics of Dialogue reflects behaviors on the level of interpersonal communication. In the current study, we put this framework to the test to accomplish two goals: First, examining the reliability of the DoE/EoD framework vis-à-vis appraising ethical thinking and conduct, and second, examining the relation between children’s DoE (ethical thinking) and their EoD (ethical behavior).

Theoretical background

The issue of moral/ethical reasoning vs. conduct in education

The abovementioned disjunction between moral reasoning and action (Blasi, 1983; Talwar, 2011) is highly problematic for education. Accordingly, researchers have driven away from an exclusive focus on reasoning, and instead stress strengths and virtues that teachers are to foster, or even instill, in students (Killen & Smetana, 2014). The most notable aspect of this approach is its emphasis on a set of dispositions and behaviors defining a person – a more global approach than Kohlberg’s and Hoffman’s (Lapsley & Yeager, 2013). This approach attempts to bridge between an Aristotelian theory of virtues and moral psychology: Habits of action lead to epistemic virtues that help to exercise reason and thus fulfill the human function. Other researchers focusing on the link between moral judgement and action have defined moral identity as “the degree to which being a moral person is important to an individual’s identity” (Hardy & Carlo, 2005, p.212). Some conjectured that people form moral identities and that internalizing one’s moral identity can influence their moral actions (e.g., Krettenauer et al., 2016). Krettenauer and Jia (2013) hypothesized that moral identities are cultural constructs, and that a culturally-inclusive approach should boost moral conduct towards others – demonstrating an ethical approach.

In this paper, we check whether an educational program designed to favor such an ethical approach can unveil a relation – heretofore unexamined – between ethical thinking and conduct. In the next section, we conjecture that positive correlations between ethical thinking and conduct can surface in the context of Dialogic Education, in contrast with the lack of correlation between moral reasoning and action found by Blasi (1983) and reviewed by Talwar (2011).

On dialogic education

Dialogic Education aims at impinging on classroom talk, for it to include accountability to the other, to reasoning, and to knowledge, as seen in different models of classroom talk, such as Accountable Talk (Michaels et al., 2008), Exploratory Talk (Mercer & Littleton, 2007), or Deliberative Argumentation (Asterhan & Schwarz, 2016; Schwarz & Baker, 2016). The deontological and socio-emotional perspectives of Deliberative Argumentation have been theorized (Schwarz & Baker, 2016), showing that Exploratory Talk or Deliberative Argumentation provide a potential for ethical learning and development. For example, in an educational program called Thinking Together (e.g., Mercer & Littleton, 2007), students progressively complied with ground rules of conversation pointing at the adoption of certain ethical conducts (see also Yackel & Cobb, 1996). Despite this potentiality, researchers in Dialogic Education have generally focused on epistemological gains, not on ethical learning.

Additional reasons for fostering ethical learning lie in the inherent difficulties in peer-learning settings in Dialogic Education: Students may avoid disagreements and critiques, perceiving them as conflicting with the desire to maintain pleasant relationships with disagreeing peers (Asterhan, 2013). In a self-report study, Nussbaum and Bendixen (2003) found that avoiding argumentation was predicted by an individual’s need to maintain warm interpersonal relations. Moreover, van der Puil et al. (2004) reported that prolonged sequences of argumentation were often followed by actions meant to repair the relationship, as if argumentation itself has affected it negatively.

This conflict between epistemologically- and socially-driven talk in peer learning provides opportunities for ethical learning in the context of Dialogic Education. Pedagogies simultaneously favoring critical discussions and kinship (aforementioned as dialogic pedagogies) may invite the integration of thinking and behavior that such learning necessitates. Dialogic pedagogies may thus help enacting specific habits of actions leading to epistemic virtues that help exercise reason and thus fulfill the human function. Some experiments point at this possibility. Lin et al.’s (2015) study on fourth-grade small-group argumentation showed that more popular children were more active and cogent discussants, supporting and challenging their peers’ arguments to a greater extent. Furthermore, children who were friends also proved more likely to support each other’s ideas. This combination of critique and challenge with openness and respect would arguably make for a potentially productive learning interaction.

Appropriation and re-use of ethical patterns of interaction have been reported in the literature. For example, using an ethnographic approach, Vossoughi et al. (2020) showed how ethical values such as showing solidarity and responsibility for the learning of others were embedded in embodied interaction patterns that were appropriated from others, including the teacher, then shared as interactive resources. However, the current study looks for a relation between ethical thinking and behavior, focusing on the role of interaction in revealing this connection. Such a relation would be more likely to attest to the internalization of ethical values in one’s thoughts and conduct, rather than the mere imitation of ethical behavior (as seen in Vossoughi et al., 2020, for example). We conjecture that when ethical values are posed for discussion, a program favoring deliberative argumentation may lead to the integration of behavior and thinking. Our conjecture relies on pioneering research already showing the positive impact of dialogic learning on value-loaded critical thinking (Frijters et al., 2020), or of collaborative learning on group work behavior (Veldman et al., 2020). However, no study has focused on the coordination between ethical thinking and behavior.

Measuring moral/ethical thinking and conduct

The novelty of the current study does not consist only in its object – the coordination between ethical thinking and behavior – but also concerns the methodological realm. Measuring moral development has traditionally relied on individually-administered interviews (like Gibbs et al.,’s 1982 Moral Judgment Interview) or paper-and-pencil tests (like Gibbs, Basinger, et al.’s 1992 Sociomoral Reflection Measure or Rest’s, 1975 Defining Issues test). Researchers in Dialogic Education have elaborated methodologies that trace learning in social interaction (mostly in small groups) and compare interactions across activities. Such methodologies have hitherto been applied for the study of learning concepts (e.g., Asterhan & Schwarz, 2009) or skills (Zohar & Nemet, 2002). This paper applies a new methodology to study the relation between ethical thinking and conduct in discussions among children.

Our general methodology is that of Interaction Analysis (Bakeman & Gottman, 1997). This involves elaborating determinate codes for interaction segments, on the basis of theory, research objectives, and previous research, that can be applied to the interaction data in a reliable manner, on the basis of operational definitions and behavioral indicators. In this case, the requirement for scientific validity may occasion some reduction of the subtility of concepts (in this case, concerning values).

We analyze interactions in accordance with the two main dimensions of the methodology proposed in Cedar et al. (2021): DoE and EoD. The DoE dimension bears on the processes by which student-participants co-construct understandings, in relation to the three values under focus: tolerance, inclusion, and empathy. This dimension is defined in relation to research on the way in which students engage with literary texts in educational situations, in relation to the ethical dimension (Rouvière, 2018). Through a process of “fictional empathy” (Larrivé, 2015), the act of reading stories involves envisioning a dialogue on values with fictional ‘others’, with empathetic engagement, axiological reading (making judgements about characters’ actions), and “pre-philosophical” conceptualization of values. This also applies to visual narratives, and ‘reading’ of multiple sign-systems (Hassett, 2010). With respect to stories that have been explicitly written to have ethical import, the processes by which students engage with these texts, possibly moving towards co-constructing “the moral of the story” have been described in terms of three main dimensions (Baker et al., 2023): interpretation, judgement, and conceptualization. Note that these processes do not necessarily occur in a particular linear order (students may ‘leap’ towards conceptualizing the ethical dimension, then return to interpretation and judgement, or the inverse).

The EoD dimension bears on the ethical conduct of students in their deliberations. We restrict the EoD dimension of analysis to the consideration of the same three values: tolerance, inclusion, empathy, but with respect to the interpersonal relations between students, in their hic et nunc dialogue on them (in line with Buber, 1947, 1957, 1958). This enables us to study the relations between DoE and EoD more directly – for example, how does students’ discussion bearing on the value of tolerance relate to the degree of tolerance that they show towards each other in their dialogue?

Together, the DoE and EoD dimensions fit the aforementioned general effort (Killen & Smetana, 2014) that bridges between an Aristotelian theory of virtues and moral psychology: Habits of action lead to epistemic virtues that help to exercise reason and thus fulfill the human function. The DoE and EoD dimensions enable us to simultaneously measure ethical thinking and behavior using the same methodology (i.e., that of Interaction Analysis) and concentrating on the same values in both ethical dimensions. This methodological innovation thus facilitates the investigation of the object of our study: the ways in which ethical thinking and conduct can coordinate.

The educational context

The educational context of this study was a three-year project (2018–2021) that focused on teaching schoolchildren cultural literacy, concentrating on three main values: tolerance, empathy, and inclusion. The project prioritized intercultural dialogue and emphasized mutual understanding (Maine & Vrikki, 2021). The children were guided to use the dialogue and argumentation skills needed for discussions where differing perspectives are discussed, the children’s own cultural identities are recognized, and the children become sensitive to and can empathize with others’ cultural identities. This pledge conveys an ethical – rather than moral – approach.

The learning program aimed to develop an understanding of children’s cultural literacy via teaching dialogue and argumentation in order to understand European cultural identities. European identities encompass cultural, political, moral, pragmatic, and cosmopolitan meanings. They are also shaped by political, legal, historical, and heritage dimensions, as well as by cultural values (Antonsich, 2008; Beck & Grande, 2007; Bruter, 2005; Pichler, 2008). European identity evolves alongside other significant layers of one’s identity. Risse (2003, 2004) elucidates the complexity of individual identities using the “marble cake model”. This model illustrates how various individual identities interconnect and intertwine in intricate and reciprocal ways within one’s personal experience and self-understanding (Lähdesmäki et al., 2018).

Cultural identities encompass shared experiences and cultural codes, perpetuated within communities through diverse cultural narratives and symbols (Hall, 1990). Such identities are created in constant dialogue, negotiation, and contest of similarity and difference, sameness and distinction. Therefore, diverse cultural phenomena can be understood as both manifestations of cultural identities and spaces of negotiations and contests where their contents and meanings are formed (Lähdesmäki, 2012; Lähdesmäki et al., 2018). Cultural identities manifest in diverse forms over time, place, and discourse (Hall, 1990). Thus, European cultural identity is characterized as a complex, fluid, and unsettled concept (Jansen, 1999; Lähdesmäki, 2012; Schunz, 2012; Stråth, 2002). Accordingly, its development is a dynamic process influenced by dialogue with others in the relational I-Thou sense (Buber, 1958), implying that young people’s perceptions of themselves as Europeans may change through discursive and dialogic practices. Thus, the educational context and learning program within which our study was situated recognized European cultural identities as diverse and multi-layered. These identities were perceived as constructed situationally, incorporating various spatial, social, and cultural elements, with some elements being more substantial and others more nuanced.

Learning sessions revolved around wordless short films and picture books reflecting a diversity of European cultural identities. The materials were wordless to overcome the language barrier deriving from the project being an international one. Seven European or Europe-affiliated countries implemented the program. This paper only deals with data collected in one country, Israel.Footnote 3

The program offered three levels of participation: whole-class teacher-mediated (hereinafter whole-class) interaction, small-group teacher-mediated (hereinafter small-group) interaction, and small-group unmediated interaction. In mediated interactions, teachers guided participation within a framework of adaptive intervention. Mediation included asking students open questions, stressing the ground rules for talk (listening to others, justifying ideas, building on others’ ideas, etc.), modeling the language to be used, and encouraging verbally-passive students to participate. The different interaction levels (whole-class, small-group, and unmediated small-group) were all performed within context, thus enabling the DoE/EoD analysis (to be elaborated in §Transcription_and_coding), which is expected to bridge between ethical behavior and thinking, both interwoven into the dialogue.

Objectives and research questions

Our study is both qualitative and quantitative in nature, and focuses on learning ethical values through teacher-mediated discussions around wordless short films. The project provides the educational framework for this general aim by generating opportunities for rich social in-class interactions. We split this aim into three objectives: Firstly and secondly, elaborating tools for measuring both ethical thinking and ethical behavior as they are manifest during in-class interaction; and thirdly, inspecting the relation between ethical thinking and behavior. To achieve the former, we propose the Dialogue on Ethics (DoE) dimension of our framework, which considers children’s dialogue concerning ethical issues, spanning children’s ethical judgements and conceptions. To meet the second objective, we propose the Ethics of Dialogue (EoD) dimension, which considers children’s ethical behavior towards each other during in-class interactions. As for the third objective, we propose to measure the relation between DoE and EoD. Our framework takes as reference points students’ understanding of three ethical concepts: tolerance, empathy, and inclusion, as demonstrated in both their DoE and EoD (Lähdesmäki et al., 2020).

These research objectives must be situated in the context of the specific task students engage in, given that its structure is a major determinant of the structure of (epistemic) dialogue (Grosz & Sidner, 1986). Here, the task involves elaborating a variety of narratives in successive sessions, which prompt ethical questions and constructive discussion concerning them, while both are mediated by the teacher when necessary.

Our first research question asks whether ethical thinking and behavior in in-class interactions can be reliably measured using the DoE and EoD tools, respectively. We propose to measure inter-rater agreement to address this question, as well as to test the usability of the two tools in terms of our second research question, which inquires what relation holds between children’s dialogue on ethical issues (DoE) and their ethical behavior towards each other (EoD) in different in-class interactions. We propose to inspect the interrelation between DoE and EoD in different sessions to answer this query.

Our review of pioneering research in dialogic learning (e.g., the positive impact of dialogic education on value-loaded critical thinking and group work behavior seen, for example, in Mercer & Littleton, 2007) suggests that dialogic education can help boost ethical thinking and behavior. We conjecture that the dialogical context will help coordinate ethical thinking and behavior, an aspect of ethical learning that has not been researched. We base our conjecture on Bakhtin’s (1981) theory of dialogism, which confers to dialogues the power to coordinate thought and action.

Methodology

Participants

The study involved 172 fourth graders (aged 9–10) from 7 classes recruited from 6 schools situated in a variety of geographical areas in Israel. Classes consisted of 24.6 participants on average. In small-group interactions, classes were divided into groups generally consisting of 4–6 students. Table 1 summarizes the details for each class.

Table 1 Details of participants
Full size table

Materials and stimuli

Wordless short films

Two sessions were pre-selected for analysis: Session 3, which was early though not preliminary, to ensure children’s familiarity with the work procedure, and session 8, which was the latest session held by all classes. These two sessions were selected due to the differences in the nature of the ethical question they concerned, and in the ethical values they elicited. These differences enabled us to explore the effect of different ethical topics upon students’ ethical thinking and behavior.

Session 3 concerned the animated short film “Papa’s Boy” (3:03 min), which unfolds the story of a young boy mouse who, much to his boxer father’s chagrin, wishes to be a ballerina. The boy later proves his worth by using the ballet moves to bewilder a cat attacking his father, subsequently saving the father’s life. Only then does the father come to terms with his son’s aspirations and accepts him. The subject of gender roles (a ballerina-boy) brought about a lively discussion in all participating classes.

The DoE involved in this session concerned the fictional father-son story and the children’s real life through the ethical question: ‘Should I follow my dreams or do what society tells me?’. This is perhaps the most fundamental question to be asked in ethics, since it raises the question of the relevance of socially-defined moral rules per se for individual conduct. The prominent values elicited were tolerance and inclusion.

Session 8 concerned the animated short film “Baboon on the Moon” (6:04 min), which presents a baboon’s daily life on the moon. The sad and lonely baboon wakes up, has breakfast, performs his chores (notably, of lighting up the moon), plays the trumpet, and sheds tears while looking at planet Earth, his home.

The DoE involved in this session concerned the ethical question of belonging and the concept of home. This task appears to clash with the dilemma-based tasks that Kohlberg (1984) suggested could trigger moral reasoning, but it did invoke ethical thinking in the children. Many of them focused on the baboon’s feelings and on possible reasons behind his arrival on the moon. The prominent value elicited was empathy.

In both sessions, participants were encouraged to interact via dialogue, and thus EoD encompasses the inclusion, tolerance, and empathy (or lack thereof) that participants expressed towards one another.

Lesson plans

The lesson plans (LPs) for both sessions involved activities related to cultural literacy as well as dialogue and argumentation skills. The LP for session 3 (LP3) included the following activities:

  1. (1)

    Presenting lesson goals: respecting others’ contributions and arguments, and contemplating issues of family tolerance;

  2. (2)

    Introducing video and discussing diversity post-video;

  3. (3)

    Discussing the father’s feelings/emotions throughout the story;

  4. (4)

    Discussing individualism versus social norms/stereotypes and preparing posters summarizing the discussion (talking point: We should not suppress our needs and dreams to follow social stereotypes but stay true to ourselves.);

  5. (5)

    Preparing comic strips showing father’s and son’s emotions throughout the story;

  6. (6)

    Reflection;Footnote 4

The LP for session 8 (LP8) comprised the following activities:

  1. (1)

    Presenting lesson goals: building on others’ ideas to reach consensus, and conceptualizing home/belonging;

  2. (2)

    Introducing video and discussing the title (“Baboon on the Moon”) pre-video;

  3. (3)

    Discussing the baboon’s situation post-video;

  4. (4)

    Discussing the dialogic goal of the lesson and the need to justify opinions;

  5. (5)

    Discussing the baboon’s home and the concept of home/belonging;

  6. (6)

    Discussing the baboon’s feelings;

  7. (7)

    Reflection;

  8. (8)

    Preparing a collage of students’ homes

Procedure

Teaching sessions

Participants were taught over the course of 8 consecutive 45–90-minute weekly sessionsFootnote 5. Sessions were delivered by each class’s homeroom teacher (or a designated teacher) and included 3 interaction types: whole-class, small-group, and unmediated small-group. The sessions under study – 3 and 8 – took place during November–December 2019 and January–February 2020, respectively. The research was authorized by the ethics committees of the university and of the Chief Scientist of the Israeli Ministry of Education.

Classes and sessions varied both within and across classes in several respects: (1) total number of turns per discussion, (2) percentage of teacher and student turns out of the entire class dialogue, and (3) percentage of each interaction type out of the entire class dialogue. In session 8, three classes involved only whole-class interactions, and one class involved only teacher-mediated interactions (both whole-class and small-group). Tables 2 and 3 provide the respective details for each session.

Table 2 Details of session 3
Full size table
Table 3 Details of session 8
Full size table

Transcription and coding

The sessions pre-selected for analysis were recorded, transcribed, and translated into English as part of a multilingual database (Rapanta et al., 2021). Transcribed interactions were qualitatively coded in accordance with the two main dimensions of the methodology proposed in Cedar et al. (2021): DoE and EoD. These dimensions consist of several aspects, to be detailed below (see Fig. 1). Some of the aspects involve several indicators, to be presented below too, followed by brief excerpts demonstrating our analysis. Most of the dialogue could be coded according to DoE or EoD, or both, but some of it was constrained by the LP and involved neither. Inter-coder reliability was established based on assignment of the different aspects to dialogue units and was duly adjusted during the process (DoE Cohen’s Kappa = 0.733, p < 0.001; EoD Cohen’s Kappa = 0.846, p < 0.001; overall Cohen’s Kappa = 0.784, p < 0.001). When disagreements arose, they were discussed and reconciled, and the guidelines for coding were updated accordingly.

Fig. 1
figure 1

The different aspects comprising the two main dimensions indicating coordination between ethical thinking and conduct – DoE and EoD

Full size image

Dialogue on Ethics (DoE)

The DoE dimension of analysis bears on the processes by which student-participants co-construct understandings of the short films, in relation to the three values on which we focus here: tolerance, inclusion, and empathy. As described above, the first film, “Papa’s Boy”, was chosen for its potential to stimulate students to engage in meaning-making of the values of ‘tolerance’ and ‘inclusion’ (a boy mouse behaves like a girl mouse) and the second, for its potential to stimulate discussion bearing on ‘empathy’ (for the feelings of the baboon on the moon, who misses his earthly home).

Interpretation concerns attributing beliefs, desires, and intentions to the characters and reconstructing causality between events. This interpretation refers to ethical thinking towards characters as it fits the “Theory of mind” (Flavell, 2004), indicating awareness that others — including fictional ones — may have different perspectives from one’s own. The following excerpt exemplifies interpretation, as student 4 (S4) assigns attributes and intentions to the characters, and those help her reconstruct the events of the story:

S4: […] in the beginning of the movie, the dad of the, boy mouse, he didn’t want him to dance. In the second part of the movie, there was a cat who came to eat up his dad, and then eh the mouse saved him and then he thought it was really good that the mouse would dance.

Judgement involves a move from the narrative towards ethical personal positioning. For instance, S1 makes an ethical judgement about the young mouse’s decision to become a ballerina:

S1: “Yes, I know but I think you’ve got, like, a whole world ahead of you. You can’t just do girly stuff.”

Conceptualization of values underlies the above aspects, and concerns explicit/implicit discussion of ethical concepts. For example, students may explicitly discuss the meaning of values, or their conceptualizations of them may underlie their judgements concerning characters’ actions. Thus, following the teacher’s (T) question concerning the meaning of respecting someone who is different, S7 explicitly defines the concept of respect:

T: “To respect the different. What does it mean ‘to respect’? Does it mean you should offer him biscuits, pretzels, and the like?”Footnote 6

S7: “To- to treat him as we treat a person who is no different from us.”

Similarly, after T asks how listening to others helped the students converge on a joint idea, S2 implicitly discusses the concepts of tolerance and individualism:

T: “From the- from listening to each other and from accepting each other’s words, you created something together. S2?”

S7: “Ehm we added lots of ideas and we actually made out of it something that’s really uh, lovely in my opinion because each one wrote what they really thought and really, this says that everyone is, like, everyone is different from one another.”

In terms of inter-coder agreement, disagreements in the DoE aspects initially arose mainly concerning implicit conceptualizations, which required the coders to make an interpretation which is less closely aligned with the actual words used by participants. For example, when students shared what ‘home’ meant for them, it was coded as conceptualization of ‘belonging’, even though the word ‘belonging’ was not mentioned by the students.

Ethics of Dialogue (EoD)

As discussed above, we analyze the students’ discussions firstly in terms of Dialogue on Ethics (DoE), i.e. their engagement with the wordless texts, seen through the lens of three processes and the three values on which we focus here, with respect to which the texts were initially chosen.

To operationalize these concepts in interaction analysis, we restrict their definitions to particular intersubjective ‘planes’ and define sets of indicators for both the positive and negative facets of each ethical value.Footnote 7

We operationalize ± tolerance on the plane of ideas. Accordingly, example indicators are acceptance (+ tolerance) or rejection (–tolerance) of others’ diverging ideas. For example, in the following excerpt, S3 manifests tolerant behavior towards S1’s opinion, though it contradicts S3’s beliefs.

S1: “[…] later ehm the child ehm saved him and the father wanted him to dance ballet.”

S3: “I think that I differ in opinion from S1, because I think that the father saw that the mouse-boy’s ballet is like, it’s like eh, it’s like, kind of boxing […]”

In contrast, S115 instantiates an intolerant reaction to S113’s question.

S113: “Did you write an example too?”

S115: “I’m not supposed to write an example, leave me alone!”

In their review of over forty definitions of empathy in the research literature, Cuff et al. (2016) define the concept in terms of an emotional (affective) response that is automatically elicited, resulting in an emotion in the perceiver that is similar to a stimulus emotion (p. 16). Accordingly, whilst we operationalize ± empathy on the plane of emotions, we redefine the indicators of + empathy operationally, in the context of social interaction, to involve tangible responses to such perceived and felt emotions, such as regulating negative group emotions or showing positive support. Thus, the excerpt below shows S131 empathizing with children who might be embarrassed by their home:

S131: “S14 said they have to draw their home, but if some kids are embarrassed, or don’t wanna show their home, or live in a home they don’t really like, then [...] they don’t have to do it”.

By contrast, indicators of –empathy include creating negative group emotions and verbally attacking others. Thus, S121’s reaction to S122’s announcement that he does not have enough writing space shows S121’s lack of empathy:

S122: “I have no place to write.”

S121: “Do I care?”

We operationalize ± inclusion in terms of interactive participation. Accordingly, indicators might be including/excluding others’ contributions, building on/ignoring others’ ideas, or allowing/blocking others’ intervention. For instance, in this example, S3 includes S7’s statement despite her disagreement with it.

S7: “[…] sometimes it’s useful to listen to others and do what they tell you.”

S3: “But eh I want to add to what S7 said […] sometimes I also don’t want to do that and sometimes I can also trust myself […]”

However, S6 demonstrates –inclusion, blocking another student’s participation.

S1: “Yes and the dad, and also the dad-”

S6: “So, wait! And then at the end, the son showed him that he can dance ballet.”

In terms of inter-coder agreement, disagreements in the EoD aspects initially arose mainly concerning empathy. While cases of –empathy were more easily identified by both coders, positive manifestations of this ethical aspect were initially more elusive. For example, one of the coders interpreted the expression of agreement with another discussant (e.g., “Yes, you’re right”) as indicative of positive support, but, following discussion, such cases were eventually agreed not to be coded as ethical, being neutral in terms of ethics, while cases of acceptance of ideas in a positive/supportive manner (e.g., “Your idea is really good/nice/smart”) were interpreted as + empathy.

Coded turns were counted and summed according to the relevant dimension (EoD/DoE) and aspect (empathy/tolerance/inclusion and interpretation/judgement/conceptualization, respectively). These sums were used in the quantitative analysis, as detailed in §Results. Table 4 provides a code summary.Footnote 8

Table 4 Summary of dimensions, aspects, and indicators comprising the coding scheme
Full size table

Data analysis

To address the first research question, i.e., whether the DoE and EoD tools can reliably measure ethical thinking and behavior in in-class interactions, we relied on the Kappa statistic to establish inter-coder reliability, as reported above (see §Transcription_and_coding). To inspect the second research question, namely, what relation holds between students’ DoE and ± EoD in different in-class interactions, we looked for correlations between the two ethical dimensions in each of the sessions under analysis. We further attempted to establish inter-session differences using paired difference tests.

We analyzed the data from all teacher-mediated interactions (both whole-class and small-group), since not all classes involved data from unmediated small-group interactions (see §Teaching_sessions). EoD was split into two facets – + EoD and –EoD – as the two were not necessarily expected to correlate (i.e., more instantiations of positive behavior do not guarantee less manifestations of negative behavior, and vice versa). The split was first performed at a general level, as we determined a behavior as ethical ( +) or unethical (–), and was followed by the identification of the relevant aspect in which the behavior proved (un)ethical (e.g., (in)tolerance).

The sample size was determined based on the necessary statistical tests: correlations and paired difference tests. Since we were looking for pairwise comparisons, a large sample was required (at least 100 data pairs in each class, see e.g. Brysbaert, 2019). Observations were thus collected from as many (coded) turns as possible in each class, given the limitations imposed by resource constraints. Since the number of coded turns in each class did not always involve normally distributed data, we used non-parametric tests: Kendall’s tau-b correlation and Wilcoxon signed-ranks test. Calculations were based on percentages (number of turns involving manifestations of DoE/EoD divided by total number of turns). In what follows, only significant effects (p < 0.05) are reported.Footnote 9

Results

Figures 2, 3, 4, 5, 6, and 7 present the different aspects of students’ ± EoD (± tolerance, ± empathy, ± inclusion) and DoE (interpretation, judgement, conceptualization) in teacher-mediated interactions (both whole-class and small-group) in sessions 3 and 8, respectively.Footnote 10 These data indicate that in both sessions, students’ manifestations of ± EoD involved mainly (in)tolerance and inclusion/exclusion, while manifestations of empathy (or lack thereof) were scarce. Concerning DoE, in both sessions, students demonstrated mostly interpretations and conceptualizations, but not judgements.

Fig. 2
figure 2

Percentages of students’ manifestations of the ethical aspects of + EoD (+ tolerance, + empathy, + inclusion) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 3

Full size image
Fig. 3
figure 3

Percentages of students’ manifestations of the ethical aspects of –EoD (–tolerance, –empathy, –inclusion) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 3

Full size image
Fig. 4
figure 4

Percentages of students’ manifestations of the ethical aspects of DoE (interpretation, judgement, conceptualization) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 3

Full size image
Fig. 5
figure 5

Percentages of students’ manifestations of the ethical aspects of + EoD (+ tolerance, + empathy, + inclusion) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 8

Full size image
Fig. 6
figure 6

Percentages of students’ manifestations of the ethical aspects of –EoD (–tolerance, –empathy, –inclusion) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 8

Full size image
Fig. 7
figure 7

Percentages of students’ manifestations of the ethical aspects of DoE (interpretation, judgement, conceptualization) out of overall number of student turns in teacher-mediated interactions of 7 classes in session 8

Full size image

A Kendall’s tau-b correlation was run on the teacher-mediated data in each session to determine the relationship between students’ DoE and ± EoD during teacher-mediated interactions in the 7 classes. Both sessions revealed strong correlations. Session 3 revealed a strong, positive, and marginally-significant correlation between students’ DoE and + EoD [τb = 0.619, p = 0.051; see Fig. 8], alongside a strong, negative, and statistically-significant correlation between students’ DoE and –EoD [τb = –0.683, p = 0.033; see Fig. 9], indicating that in session 3, the more students discussed ethical issues, the more ethical their behavior was.Footnote 11 In contrast, session 8 revealed a strong, negative, and statistically-significant correlation between students’ + EoD and DoE [τb = –0.714, p = 0.024; see Fig. 10], indicating that in session 8, the more students discussed ethical concepts, the fewer manifestations of ethical behavior they made.

Fig. 8
figure 8

The relationship between students’ manifestations of + EoD and DoE in teacher-mediated interactions of 7 classes in session 3

Full size image
Fig. 9
figure 9

The relationship between students’ manifestations of –EoD and DoE in teacher-mediated interactions of 7 classes in session 3

Full size image
Fig. 10
figure 10

The relationship between students’ manifestations of + EoD and DoE in teacher-mediated interactions of 7 classes in session 8

Full size image

To detect inter-session differences in teacher involvement, we compared the percentage of teacher turns out of the entire class dialogue in each session. A Wilcoxon signed-ranks test revealed that teacher turns constituted a significantly-larger volume of the class dialogue in session 8 (Mdn = 40.7%), compared to session 3 (Mdn = 26.9%) [z = -2.197, p = 0.028, r = 0.59].

Finally, we ran a Kendall’s tau-b correlation on the teacher-mediated data collected from each session to determine the relationship between students’ EoD and the percentage of teacher turns in teacher-mediated interactions in the 7 classes. Only session 8 revealed a strong, negative, and statistically-significant correlation between teacher involvement and students’ + EoD [τb = –0.781, p = 0.015; see Fig. 11], indicating that in session 8, the more involved the teacher was, the fewer manifestations of ethical behavior her students made.

Fig. 11
figure 11

The relationship between students’ manifestations of + EoD and teacher turns out of entire class dialogue in teacher-mediated interactions of 7 classes in session 8. EoD ethics of dialogue

Full size image

Discussion

The present exploratory study opens a new path in ethical education. It focuses on the coordination between ethical thinking and behavior in the context of Dialogic Education. The learning was limited to the coordination between ethical thinking and conduct, which can be observed concurrently in the context of educational dialogues. To do so, we elaborated the DoE/EoD (Dialogue on Ethics/Ethics of Dialogue) methodology, which encompasses ethical thinking and behavior.

Our first research question was methodological. It inquired whether ethical thinking and behavior in in-class interactions can be reliably measured using the DoE and EoD tools, respectively, as operationalized under our proposed methodology (see §Transcription_and_coding). Although we relied on a small sample (14 discussions in total), and compared the coding of only two raters, our results suggest that the answer to this question is positive. Inter-rater agreement was substantial in the case of DoE (0.733) and near-perfect in the case of EoD (0.846).

Granted, reliability constitutes only one aspect of the validity of the proposed tools. To fully validate the DoE and EoD tools, it is necessary both to compare our results with those produced using customary methods (e.g., interviews) and to test the sensitivity of our tools to in-class interaction in different settings. The latter will be addressed in future research. As for the former, while we acknowledge the need for such validation, we propose the new tools precisely because existing tools used to measure morality cannot appraise ethical thinking and conduct as both surface during interaction. Moreover, existing tools rely on self-reports. Those have been argued to be weakly correlated with behavioral measures, which are more aligned with our proposed methodology (see e.g. Dang et al., 2020). A comparison of our results with those obtained from customary tools thus seems problematic.

Our second research question concerned the relation between children’s DoE and EoD, which was indeed attested. At first glance, our results seem to be erratic: In session 3, the more students discussed ethical issues (higher DoE rates), the more ethically they behaved (higher + EoD rates and lower –EoD rates). In session 8, the more students discussed ethical issues (higher DoE rates), the fewer manifestations of ethical behavior they made (lower + EoD rates). However, we argue that this supposed contradiction can be reconciled.

The reverse trend observed in session 8 can be attributed to the heavier moderation of the teacher, as indicated by the fact that teacher turns constituted a significantly-larger volume of class dialogue in session 8, compared to session 3, and by the negative correlation between teacher involvement and students’ manifestations of + EoD, unearthed only in session 8. An additional factor that could have contributed to the negative correlation between DoE and + EoD in session 8 concerns the topic posed for discussion. While session 3 revolved around an ethical dilemma presented as a question juxtaposing two stances: following social stereotypes versus staying true to oneself, session 8 concerned the conceptualization of home/belonging. The former seems likelier to provide opportunities for richer interaction than the latter, as was indeed the case.

Thus, the non-dilemmic topic posed for discussion in session 8 and its presentation, alongside the heavier involvement of the teacher, appear to have diminished inter-student interaction, and were consequently likely to decrease instantiations of (+)EoD. The diminished inter-student interaction, in turn, seems to necessitate heavier teacher involvement, which apparently inhibits inter-student interaction even more. Adopting this interpretation, we suggest that the nature of the interaction appears to hinge upon the topic posed for discussion, its presentation, and the extent of the teacher’s involvement in the discussion. If this effect is accurate, it stresses the importance of ensuring true inter-student interaction when designing learning environments meant to induce students’ ethical thinking and behavior. At any rate, the findings show that, in certain conditions, the program, which was based on dialogic pedagogy, boosts coordination between ethical thinking and conduct, and in other conditions, it does not.

The correlations (both positive and negative) we found between ethical thinking and conduct in the two sessions contrast with the lack of correlation found between moral reasoning and action (Blasi, 1983; Talwar, 2011). We found that educational dialogues provide a suitable context for correlating ethical thinking and action. These findings confirm some preliminary insights derived from a case-study conducted on one of the discussions, in which we showed how students are led to discuss and understand ethical implications of a particular narrative, and how this relates to the quality of their collaboration (Baker et al., 2023). The inverse correlations we found in sessions 3 and 8 also seem to indicate the central role of instructional design in boosting/inhibiting coordination between ethical thinking and conduct in educational dialogues.

We based our interpretation of the findings on inferential statistics, conducting our analysis at the class level, without directly analyzing the deployment of dialogues and their precise characteristics, or recognizing the specific students that manifested DoE, EoD, or both. Such analyses exceed the scope of this paper, but are currently being undertaken.

Conclusions

This exploratory study has laid the foundation for a new line of research that bears implications for the realm of ethical learning. The context of the study was a middle-term intervention program comprising successive discussions dealing with ethical issues. We have elaborated the Dialogue on Ethics/Ethics of Dialogue (DoE/EoD) methodology, which is tailored to measure ethical thinking and behavior, respectively. The methodology was used to study ethical learning, rather than development, since the latter could not be deduced due to inherent disparities between the sessions under inspection. For example, while session 3 provided an ethical dilemma, session 8 did not. We conjecture that the methodology would be suitable for studying ethical development in a setting where the successive discussions are comparable in their design (topic presentation, the level of interest it evokes, and teacher involvement in class discussion).

The proposed methodology offers an alternative to indirect methods of tests, interviews, and questionnaires that are commonly used today in the study of morality. Indeed, we have shown that, as far as the discussions under study are concerned, different coders converge when coding children’s ethical thinking and behavior during these discussions. It should nevertheless be noted that the sample inspected was small and was confined to a single setting. The methodology should thus be corroborated using a larger number of discussions and raters, and its sensitivity to different settings should be inspected. This issue will be addressed in future research.

Using the DoE/EoD methodology, we have also shown that, in the context of in-class mediated discussions revolving around ethical issues, students’ ethical thinking correlates with their ethical behavior. However, the correlation type seems to hinge upon the nature of the discussion, which is affected by the topic posed for discussion, its presentation, and the teacher’s involvement in the discussion. To detect the relation between ethical thinking and behavior, rich inter-student interaction appears to constitute a propitious setting, since students’ engagement seems important. The creation of such a setting requires meticulous design. Research on design principles meant to encourage engagement suggests a task that presents a problem and the provision of resources (Engle & Conant, 2002). The provision of wordless texts that allude to alternative opinions seems to comply with these principles. Another principle, providing autonomy to the students, was violated in Session 8, where the teacher dominated the discussion. Indeed, when the teacher intervened moderately in the discussion – when inter-student interactions revolved around ethical issues and teacher’s involvement was limited, the positive relation between students’ ethical thinking and behavior was rendered apparent, with ethical behavior demonstrated in both increased manifestations of positive behavior, and decreased manifestations of negative behavior. However, when teacher involvement in the discussions was more substantial, the positive relation between the ethical content discussed and the manner in which it was discussed, was apparently hampered. It seems then that, as far as ethical behavior is concerned, teacher involvement should be limited for natural interaction to arise and for ethical learning to occur. The present exploratory study nevertheless points at the potential of Dialogic Education in boosting a positive relation between students’ ethical thinking and behavior, which is a new venue contrasting with the aforementioned lack of relations between moral thinking and behavior.

In this paper, we have considered fluctuations in EoD and DoE as conveying dimensions of ethical thinking and behavior. Nonetheless, it is important to distinguish – whether we speak of cognitive or affective dimensions – between having a high command of rather procedural competencies (e.g., mastery of argumentation and rhetoric, good manners) and the actual evolving of these dimensions into habits of character. That is, ethical learning will be truly indicated if children not only show ethical behaviors when asked to, but if these behaviors are also ingrained in their actions. Some authors even believe that it is the context that really predicts the behavior, rather than the personality traits, or, at least, that the situation is complex and we must observe the coherence of the child’s behavior on the one hand and their personality traits and contexts on the other (Haight, 2000). Our scope here was limited, not least to the context of Educational Dialogues. Nevertheless, as mentioned above, this context is becoming more pivotal in education and may represent a more general trend, which is expected to help boost the coordination between ethical thinking and conduct.