Abstract
Trust in nonpartisan news is essential to civil society—but is declining in the United States. However, language that demonstrates active engagement with opposing views may build trust. One way to demonstrate such active engagement is conversational receptiveness: the use of linguistic features such as agreement, acknowledgment, subjectivity, and positive emotion, among others. A review of prior work on conversational receptiveness suggests its usefulness in interpersonal conflict. This toolkit might effectively apply also to the challenge of restoring trust in nonpartisan media. A demonstration study illustrates proof of concept: In 600 opinion articles from prominent news sources, more receptive language was associated with reader trust. Pending programmatic research will address limitations, feasibility constraints, open questions, and future empirical directions—including causal tests in applied settings. At a minimum, extrapolating conversational receptivity from its role in interpersonal conflict suggests a role in building trust in nonpartisan media. Conversational receptiveness might present a cost-effective, scalable approach for media producers to bridge political divides and rebuild trust—without alienating existing audiences.
Social Media
To restore trust in nonpartisan news media, engage with opposing views. Conversational receptiveness might present a cost-effective, scalable approach for media producers to bridge political divides and rebuild trust—without alienating existing audiences.
Highlights
Trust in nonpartisan news is essential to civil society—but is declining in the United States. Perhaps it could be restored by signaling openness.
Conversational receptiveness (i.e., using language to demonstrate willingness to engage with opposing views) is a potential tool.
Reviewing the evidence from interpersonal conflict and from leadership suggests emerging support for bridging partisan divides.
A proof-of-concept study shows greater receptiveness in 600 opinion articles from prominent news sources is associated with trust from readers
Conversational receptiveness could present a cost-effective, scalable approach for media producers to bridge political divides without alienating their current audience.
The news media plays a central role in civil society. Sixty-one percent of adults in the United States get their news from national news organizations (Liedke & Gottfried, 2022). Such information is crucial for making informed choices, from learning the logistics of voting (DellaVigna & Kaplan, 2006; Gerber et al., 2009), to deciding which economic and social policies merit support (DellaVigna & La Ferrera, 2015). Trust in the news provides a foundation for common understanding (Arias, 2019) and ensures accountability of those in power because trusted media can effectively scrutinize and challenge the actions of elites (American Press Institute, 2019). Without trust in a common information source, opportunities for collaboration across political divides erode (Finkel et al., 2020).
And yet, doubts about the trustworthiness of the news media have grown. Younger people report that they currently trust social media sources more than traditional news organizations (Liedke & Gottfried, 2022). In multiple polls, consumers point to biased reporting and the spread of false information as having made them question whether they can trust what they see and read in the news (Arias, 2019; Gallup & Knight Foundation, 2023). Scholars and practitioners alike have argued that rebuilding trust in the media is essential to ensuring that a society has a reliable means of staying informed and engaged in the democratic process (Gallup & Knight Foundation, 2023; Iyengar & Kinder, 2010).
Efforts to rebuild trust have often focused on structural approaches. For example, many newsrooms have adopted measures to increase transparency by posting extended bylines of reporters and explaining their reporting process more clearly (Chen et al., 2019). Some rely on greater direct engagement with readers, whether by inviting them to submit content, expanding in-person events, or holding listening sessions (American Press Institute, 2014, 2023; Trusting News, n.d.). Significant attention has also been devoted to promoting Solutions Journalism: journalism that focuses on responses to social problems, aiming to rigorously evaluate problems and solutions and drive democratic engagement (Bauder, 2023; McIntyre, 2015, 2019). Relatedly, organizations like the Journalism Trust Initiative have developed tools and frameworks to allow both readers and media creators to evaluate the trustworthiness of content (Journalism Trust Initiative, 2018).
In parallel to the top-down approaches focused on structural changes, behavioral scientists have been proactively developing consumer-level interventions, primarily focused on combating disinformation (for review, see Kozyreva et al., 2024). These psychologically informed strategies include “inoculation” and “prebunking,” both of which are designed to build consumer capacity to recognize and dismiss false information (Lewandowsky & van der Linden, 2021; van der Linden et al., 2017). Relatedly, work on accuracy nudges has shown that priming readers to think about the veracity of content can enhance critical scrutiny and reduce sharing of misinformation (Arechar et al., 2023; Pennycook et al., 2020, 2021), perhaps especially among those on the political right (Martel et al., 2024). Although these approaches do not change the trustworthiness of the content itself, they are designed to limit the spread and influence of untrustworthy content, thus increasing the quality of the overall ecosystem.
The present article considers a complementary approach: conversational receptiveness. Conversational receptiveness is the use of language to demonstrate engagement with opposing views (Yeomans et al., 2020) 1 . Emerging research literature suggests that news organizations might be able to use conversational receptiveness to build trust across divides. This approach expands the available toolkit of interventions by offering news leaders what is becoming an empirically validated method to build trust without requiring major structural changes or vigilance on the part of news consumers. Instead, conversational receptiveness could be deployed at will by individual communicators who may be concerned about the perceived trustworthiness of their content, especially for disagreeing consumers.
The article begins by reviewing prior work on conversational receptiveness and related approaches in interpersonal conflict. Next, this article reports a proof-of-concept study testing the relationship between conversational receptiveness and trust in opinion articles. While conversational receptiveness has traditionally been assessed within the context of one-on-one dialogue, one-to-many communication is inherently more difficult for a communicator to signal receptiveness to any given reader's concerns—given that they will never meet with their audience. Initial evidence suggests that this approach has merit and should be tested more broadly.
Conversational Receptiveness in Interpersonal Conflict
Prior research has operationalized conversational receptiveness in terms of specific words and phrases that lead individuals to conclude that their counterpart is thoughtfully engaging with their point of view (Minson et al., 2024; Yeomans et al., 2020). These words and phrases are extracted from text using natural language processing and fall into interpretable categories or “features.” Some features of conversational receptiveness increase the overall receptiveness score assigned to a piece of text, while others decrease it. For instance, phrases that emphasize subjectivity (e.g., “I think” or “in my opinion”) and phrases used to hedge one's claims (e.g., “sometimes” or “maybe”) are positive features, while phrases that denote negation (e.g., “did not” or “never”) or the use of negative emotion words (e.g., “terrible”) are negative. The full set of features and examples of each feature are summarized in Table 1.
List of Conversational Receptiveness Features, Including Examples.
Conversational receptiveness is conceptually related to a variety of other constructs that seek to capture an individual's engagement with opposing perspectives including intellectual humility (Porter & Schumann, 2018), actively open-minded thinking (Stanovich & West, 1997), and need for cognitive closure (Webster & Kruglanski, 1994). However, conversational receptiveness is distinct in that it is defined in terms of behavior (i.e., words and phrases uttered by communicators), not in terms of internal mental states. As such, conversational receptiveness can be readily observed and evaluated by counterparts and has a direct and documented impact on counterpart experience of the interaction.
Conversational receptiveness is also unique in the level of precision with which it is defined. For example, constructs such as active listening (Itzchakov et al., 2024) or felt understanding (Livingstone et al., 2020) similarly seek to improve communication outcomes by helping parties in disagreement feel heard and understood by the other side. However, prior research has not offered a precise and well-tested operationalization of these approaches, leaving different researchers and practitioners to operationalize the construct differently. Because conversational receptiveness is defined and measured algorithmically, both practitioners and future scholars can readily assess whether it is being correctly implemented.
Previous studies have demonstrated that conversational receptiveness improves trust between counterparts discussing controversial policy issues. High levels of receptiveness in written text (as measured both algorithmically and by human raters) are a strong predictor of interpersonal conflict outcomes including evaluations of objectivity and the desire for future contact (Minson & Chen, 2022; Minson et al., 2020, 2023; Yeomans et al., 2020). Arguments using receptive language are often more persuasive (Hussein & Tormala, 2021), and individuals who use more receptive language are seen as more reasonable, agreeable, and trustworthy (Minson et al., 2024, Yeomans et al., 2020). These positive evaluations can even extend to the group to which the counterpart belongs, potentially mitigating intergroup conflict and fostering cross-group collaboration (Minson et al., 2024; but also see Mousa, 2020).
Using Conversational Receptiveness to Improve Trust in the Media
Research on conversational receptiveness, which has been developed for and tested on dyadic conversations between peers, could also be fruitfully applied by authors in major news outlets. However, arguing for an extension requires a thoughtful consideration of the potential challenges that might arise. It is one thing for a conversation counterpart to demonstrate receptiveness toward you in response to your own points and arguments. It is quite another for the professional writer in a major news outlet with an audience of thousands (or millions) to do so.
One possible issue is that when an author lacks the ability to respond to a specific reader's individual concerns, conversational receptiveness might lose its effectiveness. People have a deep-seated desire to feel heard—and experience more positive conflict outcomes as a result (Bruneau & Saxe, 2012). It may be difficult to make an entire group of people who endorse a particular perspective, but to different degrees and for different reasons, feel heard. Fortunately, recent research on the phenomenon of felt understanding (Livingstone et al., 2020) suggests that this is possible. In this work, relatively “light touch” manipulations convinced members of groups in conflict that their opponents understood them—and this belief improved intergroup relationships.
The case of one-to-many communication is further complicated by the fact that conversational receptiveness might produce heterogeneous effects depending on the composition of the readership. For example, a liberal-leaning outlet being receptive to conservative viewpoints might appeal to conservative readers but alienate the existing liberal readership base. If increasing trust with one constituency erodes trust with the other, the net effects of incorporating conversational receptiveness into journalistic writing may vary across contexts and could even be negative on average.
This is of particular concern in the current context where readers are highly polarized. For example, engaging with opposing-political-party views can carry negative interpersonal consequences (Hussein & Wheeler, 2024). Same-party observers judged targets more negatively when those targets signaled the desire to engage with ideas from the opposing political party, perceiving those targets as less moral. This tendency may be driven by the fact that engaging with an opposing perspective can be interpreted as signaling agreement with that perspective (Heltzel & Laurin, 2021).
Finally, yet another set of concerns surrounds the implications of conversational receptiveness and related approaches for people in positions of leadership or those seeking to influence beliefs and behavior based on their expertise. Both prior research and popular culture suggest that leaders ought to stand firm on their convictions (Kreps et al., 2017). However, more recent research paints a more nuanced picture that may make the use of conversational receptiveness particularly attractive to public figures hoping to manage their reputations. For example, leadership incivility is costly with constituents (Frimer & Skitka, 2018, 2020). Surprising reputational benefits accrue for decision-makers who change their mind in the face of new evidence (John et al., 2019). This work, combined with new research on the persuasive power of two-sided arguments (Xu & Petty, 2022), points to the possibility that using a warmer and more nuanced communication style, such as conversational receptiveness, can have benefits even from the perspective of leadership and influence (for review, see Hussein & Tormala, 2021).
Evaluating the Relationship Between Conversational Receptiveness and Trust in Opinion Journalism
Given the potential constraints described above, we recruited 5,574 participants from an online participant pool to conduct an initial test of the relationship between conversational receptiveness in the text of published articles in major news outlets and readers’ evaluations of those articles and their authors. For interested readers, greater analytic details are presented in the Appendix, and full methodological and analytic details are reported in Supplemental Materials 1–3. 2
To draw generalizable inferences, selected articles should represent a wide range of ideological viewpoints and reach wide audiences. One common approach is to rely on the classification used by the Pew Research Center (Jurkowitz et al., 2020), which uses the typical political ideology of the audience members of any given media outlet to classify that outlet as liberal- or conservative-leaning. We selected three prominent conservative and three prominent liberal sources based on the size of the outlet's readership. We also focused on political opinion editorials to identify a sample of articles where conversational receptiveness could vary from very low to very high, in a way that we would be less likely to observe in news articles. We then collected the 100 most recent political opinion articles from CNN, Fox News, the New York Post, the New York Times, the Wall Street Journal, and the Washington Post, resulting in 600 articles in total. To conserve participant effort, we presented participants with the first 100–200 words of each article, with the exact length of the excerpt being determined by the placement of the paragraph break. Table 2 presents sample excerpts that are high and low in receptiveness.
Sample Excerpts.
Each participant read one excerpt. They then made 16 evaluations about the article and its author, including assessments of trustworthiness, reasonableness, accuracy, fairness, and truthfulness. Each evaluation was made on a 5-point Likert scale anchored at “Not at all” and “Extremely.” These evaluations were highly correlated (Cronbach's α = 0.97 for the combined scale), so we created a single trust index by averaging them. We also asked participants a series of additional questions, including their beliefs about the ideological stance of the article, whether they’d like to continue reading it, and several additional demographic questions. These are reported in the Supplemental Materials.
Our main independent variable in the ensuing analyses is the conversational receptiveness of the opinion articles. We calculated the receptiveness of each article using a natural language processing algorithm, specifically employing the politeness package available in R (Minson et al., 2024; Yeomans et al., 2018, 2020). This algorithm assigns receptiveness scores to text by evaluating the presence and frequency of positive versus negative linguistic features and weighting each feature in proportion to how it impacts receptiveness evaluations offered by human coders. Using this algorithm, we obtained a receptiveness score for each excerpt participants read.
Our main finding was that higher levels of conversational receptiveness were associated with greater reader trust. Articles higher in receptiveness were rated as more trustworthy; articles lower in receptiveness were rated as less trustworthy. We next assessed whether the association of receptiveness and trust differed across levels of ideological alignment between reader and source. It is possible, for example, that whereas on average receptiveness is associated with greater trust, it may be associated with less trust when a source with which a reader typically aligns demonstrates receptiveness to the opposing political perspective. For example, a liberal reader might appreciate when Fox News displays receptiveness, but may be displeased when CNN does so. We did not find evidence to support this hypothesis: receptiveness showed a consistent and positive relationship with trust regardless of whether the reader was ideologically aligned with the source; if anything, the positive relationship between receptiveness and trust appeared strongest for ideologically neutral participants (i.e., independents). Most importantly, receptiveness did not appear to alienate ideologically aligned readers (e.g., a liberal reading CNN)—even though it could be construed as showing openness to an opposing political perspective.
Policy Implications
Generating, rebuilding, and sustaining trust in news media is essential for the maintenance of civil society and democracy itself. Prior work has provided a variety of strategies for how to repair the current trust deficit. These strategies often focus on structural changes, such as greater transparency about the reporting process, more direct engagement with audiences, and a focus on positive, explanatory, and solutions-focused coverage (American Press Institute, 2014, 2023; Chen et al., 2019; Trusting News, n.d.). More recently, psychologists have developed several effective interventions focused on consumers of news media (Kozyreva et al., 2024). As a complement to these approaches, we focus on a behavioral intervention that can be readily deployed by the producers of news media. Conversational receptiveness could present a cost-effective, scalable approach for media creators to bridge political divides and rebuild trust without alienating their current audience.
Critically, prior research provides evidence that conversational receptiveness is trainable at a relatively low marginal cost. Prior experimental research on interpersonal conflict has found that even minimal training sessions can effectively teach people to use more receptive (and less unreceptive) language—in turn producing benefits for bridging ideological conflict (Minson et al., 2024; Yeomans et al., 2020). These prior results suggest that expert communicators will easily be able to incorporate this toolkit into their writing and our study suggests that they may garner similar benefits.
This work offers policy implications not only for media professionals, but also for other mass communicators concerned with maintaining and building trust across diverse audiences. From a purely financial standpoint, at a time when many news organizations are ever more dependent on subscriber revenue (e.g., Guaglione, 2024), the opportunity to engage with more diverse or disengaged audiences represents a rare opportunity for growth. From a policy perspective, a columnist who sincerely weighs the arguments of opponents is more likely to appeal to wavering readers by addressing their doubts or concerns. To the extent that conversational receptiveness allows authors to not alienate likeminded readers when they reach out to opposing ones, we offer media professionals a potential strategy to enhance reader trust, even in today's highly charged media environment.
Several news organizations already have dedicated units and initiatives focused on rebuilding reader trust. For example, in 2021 the New York Times established a cross-functional “trust team” to address “the challenge of developing innovative ways of deepening our audience's trust in our mission and in the credibility of our journalism, no matter where it is encountered” (The New York Times Company, 2021). The paper has designed new author pages that offer more background on reporters and their beats; provided details on ethical standards; and made datelines more explicit (Peck, 2024). Such initiatives could be enhanced by incorporating training or editorial policies designed to increase conversational receptiveness into their ongoing activities.
Constraints and Open Questions
Conversational receptiveness is a potential behavioral strategy to address a significant policy issue. However, it has several limitations that must be fully considered before adoption or implementation at scale. The most important limitation of the analyses presented in this article is the correlational nature of the data. While conversational receptiveness was systematically associated with greater trust in the data we collected (and this relationship appeared linear; see Supplemental Material 3 for details), it could be the case that a third variable drove this association. And although correlational and causal analyses have provided similar patterns of results in the context of interpersonal conflict (Yeomans et al., 2020), future studies should involve experimental manipulations of conversational receptiveness in journalism specifically. Potential methods could include randomly assigning journalists to receive receptiveness training (Yeomans et al., 2020) to systematically vary article receptiveness levels. Relatedly, conversational receptiveness could be effectively implemented in a large language model such as ChatGPT. Prior research suggests that conversational receptiveness is relatively easy to train, and thus we expect experiments to yield similar results to the ones we obtained in the correlational studies. This, however, remains an open question requiring testing, both in the lab and at scale.
Another question raised by the receptivity hypothesis is why journalists (in particular opinion journalists) have not adopted this approach intuitively, given their vast experience communicating with audiences and the general concern about trust in the media. One possibility is that writers (and editors) have incorrect lay theories about the consequences of conversational receptiveness and related approaches. For example, they may believe that conversational receptiveness will alienate their ideological supporters without appealing to their ideological opponents, or signal agreement with perspectives they find objectionable. The present work contradicts this intuition. It is generally considered harder to attract a new subscriber than retain an existing one—hence the emphasis on not alienating readers. But as audiences shrink, newsrooms increasingly need to pay attention to the readers they’re losing or never had.
Another possibility is that even if intuitions are well calibrated, it could be that conversational receptiveness undermines other goals. That is, while trust might be a valued outcome, it is not the only outcome authors pursue. Advancing a particular ideology or strongly appealing to a dedicated group of paying readers might supersede any benefits of conversational receptiveness for trust. This may be especially true for individual columnists who are not professional journalists, and who often have other priorities: capturing attention, activating allies, fundraising, or gaining notoriety—some of which may be at odds with appearing more receptive and/or generating trust across political divides. While we found that conversational receptiveness was systematically associated with greater trust, we did not find evidence that it was associated with a greater desire to continue reading the article (see Supplemental Material 3 for greater detail). Future studies are needed to assess how conversational receptiveness drives other behavioral metrics commonly used to gauge the impact of online opinion articles, including likes, shares, and comments. It is only with this expanded scope of analysis, considering the full set of goals that media producers hold, that researchers and practitioners alike can design comprehensive policy recommendations. For now, we note that trust stands as one such goal that requires additional attention.
Conclusion
Partisan politics in the United States is increasingly defined by animus and distrust (Finkel et al., 2020). In such a polarized climate, trust in news is essential for bridging divides and serves as the foundation for effective communication and wise policy design. We suggest that conversational receptiveness may present a promising tool to restore trust in news media—and in each other. Most broadly, our work contributes to a growing literature on behavioral science, communication, and leadership (Moore & Bazerman, 2022). As a complement to individual-level training of news readers, the present research points to a necessary focus on the communication strategies employed by those who have the most societal influence. Without effective communication, opportunities for policy collaboration erode.
Strength of association between receptiveness and trust at different levels of ideological alignment with the source. Numbers above the bars represent regression coefficients. Error bars represent one standard error around the mean. p < .10. *p < .05.
Supplemental Material
sj-docx-1-bbs-10.1177_23727322241268409 - Supplemental material for Can Conversational Receptiveness Build Trust in the Media?
Supplemental material, sj-docx-1-bbs-10.1177_23727322241268409 for Can Conversational Receptiveness Build Trust in the Media? by Dilan Tulan, Charles A. Dorison, Nancy Gibbs and Julia A. Minson in Policy Insights from the Behavioral and Brain Sciences
Footnotes
Declaration of Conflicting Interests
Two of the authors (Dorison, Minson) offer training workshops on conflict communication, which include teaching conversational receptiveness.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Harvard Kennedy School - Shorenstein Center on Media, Politics and Public Policy.
Supplemental Material
Supplemental material for this article is available online here: https://osf.io/2wskr/?view_only=e3c6b4470c744dbb80c70b151885118c
Notes
Appendix
This Appendix includes details about the analyses reported in the article. Additional methodological and analytic details are presented in the Supplemental Materials. To begin, we standardized receptiveness scores to aid interpretation, resulting in a mean score of 0 and a standard deviation of 1 across all articles. Second, we compared the receptiveness scores of opinion pieces from liberal vs. conservative outlets. While conservative news sources tended to exhibit slightly lower receptiveness compared to liberal sources, this difference was not statistical significant (b = 0.11, SE = 0.08, t(598) = 1.31, p = .189).
Our main analysis focused on whether conversational receptiveness was correlated with greater reader trust. We employed regression analyses to examine this relationship, regressing participant trust evaluations on the receptiveness scores of the articles. Each article’s receptiveness was adjusted for multiple reader ratings by including a fixed effect to the regression. Higher levels of receptiveness were associated with more positive ratings from participants (b = 0.05, SE = 0.02, t(594.50) = 2.44, p = .015). Importantly, this relationship remained significant even after controlling for the participant’s political ideology (b = 0.05, SE = 0.02, t(594.51) = 2.47, p = .014).
We next examined receptiveness across different levels of ideological alignment between readers and sources. While one might anticipate that receptiveness generally enhances trust, there could be concerns that it might reduce trust when a source aligned with the reader’s ideology shows openness to opposing perspectives. To assess this possibility, we regressed trust on receptiveness and the ideological alignment between reader and source (again including a fixed effect for article). Results revealed no significant interaction between receptiveness and ideological alignment in trust evaluations (b = −0.02, SE = 0.01, t(5421.67) = −1.14, p = .254). Furthermore, even after controlling for ideological alignment, we consistently observed a positive relationship between receptiveness and trust (b = 0.05, SE = 0.02, t(595.50) = 2.55, p = .011).
In our final analysis, we examined the relationship between receptiveness and trust for each category of ideological alignment. Among ideologically neutral participants, receptiveness showed the strongest positive relationship with trust (b = 0.07, SE = 0.03, t(349.73) = 2.31, p = .022). For readers whose ideology did not align with the source, receptiveness was also positively associated with trust (b = 0.05, SE = 0.03, t(548.48) = 1.95, p = .051), although this relationship was somewhat weaker. Finally, while there was no significant relationship between receptiveness and trust for participants aligned with the source (b = 0.02, SE = 0.02, t(597.55) = 1.00, p = .316), the positive direction of the relationship suggests that receptive language did not alienate these readers. Figure 1 presents these results visually.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
