Conversational Polarization: Eight Key Considerations for Communication Scientists

Abstract

Previous research has largely overlooked how polarization is constituted through conversations. We introduce the concept of “conversational polarization,” highlighting how polarization is socially constructed in conversations about contentious issues. We present eight considerations for communication scientists to study polarization. Theoretical considerations include: (a) positioning research within the interactional paradigm, (b) considering interlocutors’ responses to speakers’ conversational contributions, (c) uncovering moralities, and (d) identifying conversational turning points. Methodological considerations involve (e) focusing on everyday conversations, (f) linking micro- to macro-level analyses, and (g) noting pitch, facial expressions, and arousal. Finally, concerning societal considerations, we recommend (h) eliciting participant feedback.

Introduction

Polarization research on contentious issues, such as COVID-19 (Hart et al., 2020) and climate change (Van Eck, 2024), has surged in recent years. Although this research has greatly enhanced our understanding of polarization, it often lacks an interactional approach to studying how polarizing communication unfolds in real time. Instead, research focuses on individuals’ cognitive and affective states, treating them as information processors. For example, studies analyze isolated statements (e.g., tweets, hashtags, and frames), individuals’ retrospective accounts (through interviews or surveys), or the effects of experimental exposure on cognitive or emotional responses (e.g., Hart et al., 2020; Huang & Liu, 2022).

This research often treats communication as a static process, rather than an interactive one where interlocutors jointly negotiate meaning. A key oversight is how polarization develops and is sustained in everyday conversations—whether in family discussions, debates between politicians and scientists, or public interactions with differing views. Current approaches miss how polarization unfolds moment-to-moment in real-life (online) conversations, identifying conversational turning points where interactions may shift toward polarization or depolarization. These dynamics often play out in everyday conversations, subtly reinforcing divisions (e.g., Van Burgsteden & Te Molder, 2022; Raymond et al., 2023; Reynolds, 2011). Ignoring the interactive nature of polarization risks overlooking communicative dynamics that solidify divides.

We argue that there is no polarization without conversations. Our main point is that more interactional approaches are needed to study polarization dynamics. By focusing on how polarization is constructed, managed, and maintained in both online and face-to-face interactions, we can better understand why discussions on scientifically contentious issues (e.g., climate change, vaccination, and immigration) become polarized and how they can be depolarized. We introduce the concept of “conversational polarization,” which sees conversations as constitutive of polarization. Conversational polarization is defined as real-life, everyday interactions in which individuals construct the meaning of issues, identities, and relationships in a conflictual way (cf. Van Burgsteden, 2022; Dewulf et al., 2009). Specifically, conversations are polarized when individuals treat each other’s contributions on contentious issues as offensive, leading to defensiveness or attacks that escalate hostility and hinder productive discussion.

We do not assume the common cognitive classification of ideological polarization (opposing positions) and affective polarization (out-group dislike; Van Eck, 2024). Instead, conversational polarization examines how these dimensions are actively managed in real-life exchanges, contrasting with studies that focus on ideological and affective polarization through individual attitudes and emotions. For instance, in ideological polarization, how do conversational partners respond to cues suggesting the other holds an incompatible stance? In affective polarization, how do listeners interpret and navigate expressions of disdain or distrust during the conversation?

In this commentary, we explore the concept of conversational polarization through eight key theoretical (#1-4), methodological (#5-7), and societal (#8) considerations for communication researchers. Importantly, we do not suggest that other polarization research is irrelevant or inadequate. Rather, we aim to emphasize that investigating conversational dynamics adds a crucial layer to understanding why societies become polarized around controversial, science-based issues.

Eight Key Considerations for Communication Scientists

1. Position the Research Within the Interactional Instead of a Static Paradigm

In the communication discipline, most research on polarization is positioned within a static paradigm, conceptualizing polarization as a state instead of a process (DiMaggio et al., 1996; Van Eck, 2024). These studies often treat people’s utterances as representations of their inner cognitive and affective states. In this way, meaning-creation is ultimately treated as a private activity of individuals, who use language to represent this inner world (Dewulf et al., 2009). For example, many studies have analyzed individual hashtags, tweets, and keywords (e.g., Haunschild et al., 2019; Pearce et al., 2014) or have investigated how specific statements trigger negative reactions in controlled settings (e.g., Reiman & Killoran, 2023). The results are used to explain how positions (e.g., opinions, attitudes, and beliefs) are opposed, interactions are divided (e.g., networked echo chambers), and speech is hostile (Van Eck, 2024).

In contrast, polarization can also be treated as the product of interactional alignments or co-constructions, where interlocutors jointly negotiate the meaning of issues, their identities and relationships in their interactions. In this interactional paradigm, individuals are treated as conversationalists rather than mere information processors (Dewulf et al., 2009). For example, previous research has shown how commenters on climate change blogs often shift their interaction strategy depending on their interlocutors’ responses (Van Eck et al., 2020). We advocate moving beyond common conceptualizations of polarization as a state and consider it as an emergent property of interactions.

2. Mind Interlocutors’ Responses to Others’ Utterances

The majority of research that conducted content analyses to investigate polarization has focused on individual/singular messages, such as tweets (e.g., Anderson & Huntington, 2017; Jang & Hart, 2015; Sanford et al., 2021). Alternatively, in the interactional paradigm, they could have explored the dynamic unfolding of polarization by analyzing how users’ interpretations of each other’s responses shaped the course of the ongoing conversation (cf. Van Eck et al., 2020; Van Burgsteden & Te Molder, 2022). Analyzing isolated or separate utterances risks misinterpreting polarization dynamics, as these utterances may seem polarizing or neutral out of context. Only by examining how recipients interpret and respond can researchers accurately identify the (de)polarizing character of an exchange.

In the field of conversation analysis, this interpretation-centered approach, known as the “next-turn-proof procedure,” directly reveals how participants’ responses to each other can escalate or mitigate polarization. It involves examining conversational turns to understand how a prior turn is interpreted (Sacks et al., 1974). Next turns, thus, serve a dual purpose: (a) they provide interlocutors with evidence of how their turns are understood by others, and (b) they offer analysts insight into the understanding of the prior turn. Thus, a conversation about a contentious issue is considered polarized if interlocutors treat each other’s responses as offensive. Seemingly neutral contributions can be interpreted in various ways; one interlocutor might interpret an utterance as offensive, while another might treat it as constructive (Van Burgsteden & Te Molder, 2022). We therefore argue that analyzing individual or separate utterances is problematic, as it is unclear whether they are in fact taken up as polarizing, depolarizing, or neutral. Instead, researchers should look at how speakers’ utterances are treated by recipients, to determine the (de)polarizing character of certain utterances.

3. Uncover Interlocutors’ Moralities

In debates over contentious issues involving scientific questions (e.g., climate change, vaccination), the stakes extend beyond knowledge to include the values, ideologies, and worldviews associated with these issues (Jasanoff & Simmet, 2017). These moral dimensions often go unnoticed in conversations. That is, interlocutors frequently embed their values within the facts they present without explicitly articulating them (e.g., Van Burgsteden & Te Molder, 2021, 2022; Clark & Botterill, 2018; Prettner et al., 2024). For instance, vaccine-hesitant parents use scientific arguments to support their perspective rather than explicitly saying that a “good parent” avoids blindly trusting medical advice (Te Molder, 2012).

By studying conversational polarization, our focus shifts from individuals’ cognitive and affective schemas to how their values, ideologies, and worldviews are embedded in their conversations. Fine-grained analysis of real-life conversations enables communication research to bring these moralities to the surface (Prettner et al., 2021), revealing how individuals position themselves in everyday life and for what purposes. For example, one study revealed how citizens who reject scientific facts about nitrogen emissions use this morality to build their identity as “good citizens” (Van Burgsteden & Te Molder, 2021). We propose specifically focusing on the moralities embedded in conversations when taking an interactional approach.

4. Identify Conversational Turning Points

People often remember conversations with a family member (e.g., “the cranky uncle” at Christmas; Cook, 2020) or a friend about a contentious issue that led to conversational hostility or a deadlock. We advocate for more research on ‘conversational turning points’; the moments that trigger conversational escalation and potentially a conversational impasse, or conversely, depolarization and mutual understanding. For example, one study found that interactions between ideologically opposed users on Reddit were more likely to end if users made negative references to the other’s political in-group (Marchal, 2022). Similarly, another study showed that climate skeptics and climate acceptors started using polarizing interaction strategies when they felt that their identity was being attacked (Van Eck et al., 2020). These studies identified specific conversational cues—such as the use of negative references to in-group identities or perceived identity attacks—that signal a turning point toward escalation and conversational impasse.

Contrastingly, previous research on contentious issues has also focused on indicators of conversational turning points leading to depolarization and mutual understanding. For instance, research on government–citizen meetings about livestock farming found that when officials invited citizens to elaborate on their concerns and citizens reciprocated, both parties showed signs of having reached mutual understanding (Van Burgsteden et al., 2022a, 2022b). Although research on turning points in conversations about polarizing issues is growing, it remains limited. Nevertheless, this area of study holds significant potential for informing depolarizing interventions aimed at fostering constructive dialogues.

5. Focus on Everyday Real-Life Conversations

Focusing on conversational polarization means studying real-life, authentic conversations that occur naturally in everyday life, without researcher intervention. This is because retrospective accounts gathered through interviews or surveys are incomplete, and individuals’ intuitions about their behavior in experimental settings often differ from reality (Hoey & Kendrick, 2018). While naturalistic data offer rich insights, experimental settings can be important for testing conversational depolarization interventions in a controlled setting (e.g., moderation’s impact on depolarization in group discussions). Therefore, naturalistic and experimental data should complement each other, with natural conversations serving as a foundation for more controlled studies.

Face-to-face conversations can be audio/video recorded on-site with participants’ permission. To capture naturalistic data, researchers should avoid intervening to preserve the spontaneity and context of interactions without introducing artificial constraints. For example, real-life police–civilian encounters recorded via dashcam (Raymond et al., 2023), conversations during protests (Reynolds, 2011), or doctor–parent discussions on child vaccination (Prettner et al., 2024) can be collected with careful ethical consideration. When gathering online conversational data, it is crucial to capture the conversation sequence (see consideration #2).

6. Complement Micro With Macro

Analyzing conversational polarization benefits from both qualitative and quantitative approaches, which complement each other for a full understanding. Qualitative methods uncover dynamics like conversational turning points and moralities, which can then be translated into deductive coding for broader quantitative analysis. This allows for assessing the prevalence of these dynamics across topics (e.g., vaccination and immigration), communities (e.g., scientists and policymakers), platforms (e.g., X and WhatsApp), online versus offline interactions, and over time. Regardless of methodology, it must allow analysis of interlocutors’ immediate responses, making methods like content analysis, interviews, and surveys less suitable. Qualitative methods like conversation analysis (Sidnell & Stivers, 2013), discursive psychology (Wiggins & Potter, 2017), and interactional framing analysis (Dewulf & Bouwen, 2012) are more appropriate, as are quantitative methods such as manual content analysis and computational tools (e.g., BERT classifiers and sentiment analysis) with a sequential focus (e.g., Marchal, 2022).

7. Also Pay Attention to Pitch, Facial Expressions, and Arousal

To understand the polarizing or depolarizing nature of real-life conversations, it is crucial to consider both content and delivery, as communication includes verbal, vocal, and nonverbal cues. Approaches that overlook these elements miss key aspects of polarization in real conversations. For example, a statement delivered with stress or a high pitch can intensify polarization, while the same statement spoken calmly can reduce conflict (cf. Van Burgsteden & Te Molder, 2022). Acknowledging these nuances provides a deeper analysis of conversational polarization. However, capturing these details requires careful methodology, beyond the scope of this paper. Yet, we present three examples: first, conversation analysts use the Jefferson transcription system to record interaction details (Jefferson, 2004); second, skin conductance measures arousal levels, offering insights into why people may use polarizing strategies (Bakker et al., 2021); and third, technologies like MEXCA analyze emotional expressions through facial movements, vocalizations, and speech, generating data for large-scale analysis (Lüken et al., 2024).

8. Elicit Feedback and Provide Conversational Techniques

We recommend presenting research findings to participants, such as through workshops, when possible. This serves as member-checking, where researchers informally assess participants’ reactions to the reconstruction of conversational polarization (Candela, 2019; Lincoln & Guba, 1986). Feedback helps researchers better understand the data without altering conclusions, offering insight into how participants experienced their interactions. These workshops not only enhance data trustworthiness but also provide an opportunity for public engagement. By analyzing real-life episodes, workshops offer participants valuable insights into their own behavior and the polarizing or depolarizing effects of their contributions (cf. Stokoe, 2013). This allows participants to practice new conversational techniques for engaging in productive discussions on contentious issues.

Concluding Remarks

We have introduced the concept of “conversational polarization” by outlining eight key considerations for communication researchers in studying the underlying processes of (de)polarization. While this list is not exhaustive, it provides a starting point for exploring conversational polarization. To date, communication researchers have largely overlooked how conversations constitute polarization. Hence, we hope that future research will further develop this concept theoretically and empirically by elucidating the specific conversational dynamics involved in various polarized contexts and by exploring how these insights can be applied to depolarize contentious debates.