News sources and the epistemic authority to represent reality: How journalism constructs an order of authorized knowers in local television news

Abstract

Distinguishing two types of epistemic authority, narrative and eyewitness, illuminates their distinct contributions to journalistic authority and reveals how journalism constructs an order of authorized knowers in local television news. Quantitative content analysis of local television news shows how different sources, such as public officials, civil authorities and citizens, are assigned epistemic authority both in terms of how frequently they appear in news and what journalists empower them to say. The basis of sources’ epistemic authority and the type of story they appear in shapes the knowledge about social reality they are discursively permitted to create. Citizens uniquely contribute to journalistic authority by imparting eyewitness authority to news. They take more speaking turns than other sources, but the epistemic authority their eyewitnessing creates is commonly coopted by elite sources whose status-derived narrative authority lets them characterize social reality without claiming to have witnessed it. Citizens appear in few government stories and take relatively few speaking turns in them, although most of their turns characterize social reality. Citizens take more speaking turns than authorities in crime stories, but their relatively even mix of characterizing and witnessing undermines their narrative authority relative to officials who take few witnessing turns. Demographics matter: White women citizens take more witnessing turns than all other citizen race and gender categories combined. Connecting social status with the authority to characterize social reality maintains structures of inequality, and authorities telling average people how to make sense of their experiences can be seen as dismissive of citizens’ judgment.

Keywords

Sources television news content analysis local journalism journalism credibility race/ethnicity gender

News gives shape to social reality, assigning order and meaning to real world events. As Matt Carlson observes, “The news composes a vision of the social order through aggregating the visions of its sources” (2009, p. 527). Mediating between sources is an evidentiary strategy that contributes to journalistic authority (Carlson, 2017) and creates what Fishman calls a “normative order of authorized knowers” (1980, p. 144-145). Journalists turn to official sources in positions of political or social power as authorized knowers (Carlson, 2017; Cook, 1998), and in so doing confer upon those sources the epistemic authority to characterize social reality. As Carlson notes, “the reliance on official sources … is one of the most reproduced findings in studies of journalism” (Carlson, 2009, p. 529). Eyewitnessing is another form of epistemic authority, invoked to nullify the gap between experience and its representation through the claim of having “been there” (Zelizer, 2007). Where the criterion is personal experience rather than social status, citizens become authorized knowers, imparting authority to journalistic representations of social reality by virtue of their status as witnesses.

This study examines how the basis of a source’s epistemic authority – having status or bearing witness – shapes the sort of knowledge about social reality they are discursively permitted to create. Traditional critiques of U.S. news sourcing practices (e.g., Bennett, 1990; Hallin, 1986; Sigal, 1973) have decried the reliance on political elites in the news. European scholars observe that citizens appear in the news quite commonly but that “they do not propose, initiate, debate or engage, they are like ‘extras’ in the news drama, providing background and mood, with little agency” (Lewis, 2006: p. 312; see also Kleemans et al., 2017). Building upon these observations, this study examines the nature of journalists, officials, civil authorities and citizens’ epistemic authority in the context of different types of news stories. It uses news content data to reveal journalism’s professional practices for presenting news sources to audiences, althought it cannot illuminate individual journalists’ decision-making processes about how to report particular stories. The study reveals both how frequently different types of sources appear in news and what journalists empower them to say. It differentiates the epistemic authority of witnessing from the epistemic authority to characterize situations in order to better theorize the “normative order of authorized knowers” (Fishman, 1980: pp. 144-145) that in turn establishes journalistic authority.

To better contrast between forms of epistemic authority, this analysis examines local rather than national news. National-level stories inherently imply a level of abstraction in which individual experiences and evaluations may be eclipsed by elite activities and representations. In contrast, local news commonly features specific events of limited scope where witnesses play a more obvious role – crimes, accidents, public meetings, and human-interest stories, among others. Under these conditions, more favorable than those of national or international news coverage, citizen sources appear frequently in the news and perform the unique role of bearing witness to lived experiences. However, the narrative authority to which citizens’ witnessing authority contributes is commonly realized by more elite sources and by journalists, who characterize social reality without claiming to have witnessed it.

Literature review

As Carlson observes, “In promoting patterns of particular sources as authoritative, news discourse provides a symbolic map of society’s authorized knowers” (2017, p. 186). Yet the idea that the basis of a source’s epistemic authority – bearing witness or having status – shapes what sort of knowledge they are discursively permitted to create has not been explored in the literature on journalistic authority. To the extent that journalists distinguish between the types of epistemic authority held by social actors who appear in the news as sources, they are not only engaged in boundary work that distinguishes journalists from non-journalists (Lewis, 2012). Beyond including or excluding particular social actors as sources, what journalists allow those sources to testify about sustains the “contingent relationship in which certain actors come to possess a right to create legitimate discursive knowledge about events in the world for others” (Carlson, 2017: p. 13), what Carlson calls journalistic authority. While journalistic authority is supplemented by layers of editors, coders, algorithms, and others features that define modern digital journalism (Carlson, 2020), epistemic authority originates in journalistic practice and source selection (Godler and Reich, 2013).

Existing research focuses on the presence or absence of particular social actors as news sources, and being present in the news is a necessary precondition to being an “authorized knower” in any capacity. Critiques of journalists’ choice of news sources observe a heavy reliance upon elite sources (Bennett, 1990; Hall et al., 1978; Hallin, 1986; Sigal, 1973). According to these critiques, reliance upon elite sources for news reinforces existing authority relationships and limits the scope of debate to elite perspectives. Reporters have also been shown to select elite sources based on their ability to affect the outcome of an ongoing event (Entman and Page, 1994) or based on what sort of story the journalist perceives (Cook, 1998; Edy and Daradanova, 2006). The epistemic authority of these sources, drawn on and reinforced through their repeated appearance in news discourses, is subject expertise grounded in their social status (Ekstrom, 2002). As elected officials, bureaucrats, diplomats or business leaders, they have expert knowledge about the government, the world order, elections, or the economy that permits them to characterize what is and speculate about what might be.

Citizens, in contrast, are generally less appealing news sources. They are harder to locate and contact than elite sources, and their information is not treated as implicitly trustworthy (Andén-Papadopoulos and Pantti, 2013; Lewis, 2012), so extra work is required to verify it. Appearing in ad hoc or spot news (Reich, 2015), citizen sources do not accrue credibility over time as elite sources may, nor are they adept at the concise quotability of practiced elites. Research suggests journalists value citizen sources less and use them less (Reich), although their likelihood of using citizen sources varies by story type (Hopmann and Shehata, 2011). Local news tends to feature more stories in which citizens commonly appear (e.g. crime or human interest), and Kaniss (1991, p. 120) argues that local news tends to emphasize the human dimensions of news. Mariska Kleemans and colleagues (2017) find that over time, citizens have come to appear more frequently in news. However, existing literature supports the prediction that citizens appear less frequently than other sources.

H1: Citizens appear less frequently as sources in the news than do government authorities or civil authorities.

The presence of sources in the news does not necessarily indicate their role in representing social reality. There is also the question of what their epistemic authority allows them to speak about (Kleemans et al., 2017; Lewis, 2006). Barbie Zelizer (2007) documents how witnessing generates a distinctive form of epistemic authority valued in journalism. Journalists use “eyewitnessing to underscore, establish, and maintain their authority for reporting,” treating it as “not only a mark of journalism but a synonym for good journalistic practice” (Zelizer, 2007: p. 408). Stuart Allan argues, “This attitude revolves around the journalist’s personal presence as a reporter on the scene, serving as an on-the-spot witness rather than a commentator from afar” (2013, p. 38). Witnessing ties into authority claims, for, “This valorization of tangible experience appears as a guarantee of the ‘truth’ of whatever is recounted” (Muhlman, 2010: p. 34). Tangible experiences may have even greater epistemic authority today as trust in political authorities, social authorities, and media has declined (Quandt, 2012) while testimonies of personal experience are largely unquestioned (Van Zoonen et al., 2007). Carlson (2017) notes that witnessing is a form of “distant presence” arising from journalists’ greater access to events in the social world relative to that of citizens. However, for many iconic news events as well as during crisis and conflict, it has been citizens who generated eyewitness authority, from Abraham Zapruder who filmed Kennedy’s assassination to George Holliday who videotaped Rodney King’s beating to Darnella Frazier who captured George Floyd’s murder on her cell phone (see also Chouliakari and Al-Ghazzi, 2022).

In news contexts, tension emerges when the objectivity norm is counterpoised to the essential subjectivity of witnessing (Chouliakari and Al-Ghazzi, 2022; Parks, 2019; Zelizer, 2007). Eyewitnessing can include a moral imperative (Chouliakari and Al-Ghazzi), and its emotionality contrasts with the rational forms of “hard news” (Parks). Where once its subjectivity was embraced and its unreliability downplayed (Zelizer), contemporary journalists commonly mediate this tension between subjectivity and impartiality by contrasting their own professional objectivity with the subjectivity of witnessing by non-professionals. For example, Karin Wahl-Jorgensen (2013) observes that journalists “outsource” the emotionality of their reporting to their sources even as they endeavor to stay true to the ideals of objectivity. Relatedly, Sue Robinson (2009) found citizens were used as “exemplars” in journalists’ commemorative coverage of Hurricane Katrina, becoming “symbols and metaphors advancing [a] narrative plot” created by journalists (pp. 804-805).

Historian Michael Frisch (1986) illuminates two distinct forms of epistemic authority by separating witnessing authority from what might be termed “narrative authority,” the authority to characterize reality. He observes that in the documentary film Vietnam: A Television History, Vietnamese peasants and U.S. ground troops:

Tend to be quoted as offering personal experiences, direct observations, and recalled feelings. The higher or more important the position of the subject, however, the more likely he or she is to be seen as offering historical judgments of a broader nature, sweeping evaluations of what an event meant, what caused it, or what the public felt. (p. 13)

Carlson draws a similar distinction in news, arguing “The ability of witnesses located in the right place at the right time to gather and disseminate information or of experts to analyze events from an informed yet subjective viewpoint signals the occasional fluid transference of authority” (2017, p. 194; Godler et al., 2020, make a similar distinction). Consequently, it is appropriate to distinguish between witnessing authority and the narrative authority by which social reality is characterized and thus acquires meaning beyond the personal. Witnessing and characterizing are functions, not identities, since any news source can characterize beyond personal experience to draw conclusions about its broader meaning. Conversely, social authorities can describe their personal experiences of public events as well as ascribing meaning to them (Edy and Austin, 2022).

H2: Citizens make more statements bearing witness than government authorities, civil authorities, or journalists. Government authorities, civil authorities, and journalists make more statements characterizing than do citizens.

The interplay of these forms of epistemic authority, together with the finding that the frequency of citizens’ appearance in news varies by story type (Hopmann and Shehata, 2011), raises a research question about how sources’ epistemic authority is constructed in particular types of news stories. Narrative and witnessing authority may be distributed differently, for example, in crime stories than they are in human interest stories.

RQ1: How do news sources’ frequency of speaking, witnessing, and characterizing differ from story type to story type?

Finally, Frisch’s (1986) observations highlight how social power and hence social inequities may shape the distribution of epistemic authority in news. Previous research finds women appear less frequently as news sources (Ross, 2007; Zoch and Turk 1998) and as political experts (Searles et al., 2023) than men but has not specifically examined what women were allowed to say. Historically, the statements of both women and people of color have been undervalued in public discourse. Both groups are stereotyped as emotional rather than rational (Harris-Perry, 2011; Kent, 2004), making citizens in these groups especially useful as exemplars and illustrators who invoke the unique blend of subjectivity and realism particular to eyewitness authority. This may mean that citizens who are women and people of color are more likely to make statements bearing witness to their lived experience compared to citizens who are white and male.

H3: Citizens’ gender and race affect the frequency with which their speaking turns bear witness or characterize.

Method

Research on news sources is typically conducted on national news (e.g., Bennett, 1990; Hopmann and Shehata, 2011; Kleemans et al., 2017; Sigal, 1973). This study, in contrast, considers local news, which offers richer opportunities than national news for sources whose epistemic authority derives from witnessing as well as including sources whose epistemic authority derives from their social status. The goal is not to theorize about the nature of local news but rather to theorize about how journalism constructs sources as authorized knowers. Since the data take the form of news content, no speculation can be made about individual journalists’ motivations for their choices, including how they themselves perceive the nature of their sources’ authority. Rather, the study treats the observed patterns as produced by trained professionals applying established newsmaking practices and as constrained but not determined by an underlying reality. While journalists cannot create witnesses, their decisions about how to cover a story affect who is perceived as “witnessing,” and they can impart narrative authority to characterize to any type of source.

Unlike national news, local news is not centrally archived, so randomly sampling newscasts across geographic areas over long periods of time is not feasible. Thus, this study employs a case study approach. Chain ownership of local television stations as well as standards of news professionalism increase the likelihood that findings are transferable to local news in other television markets. Although there is no way to determine whether the time frame when data were gathered represents a typical local news agenda, the agenda was not dominated by one or a few multi-day stories.

Context: Oklahoma city

Data were collected from the Oklahoma City television market. When data were collected, Oklahoma City ranked 45th of 213 U.S. television markets with 676,720 television households (MediaTracks Communications, n.d.). The market includes most of central Oklahoma but excludes the Tulsa area. Each major U.S. national broadcast network has an affiliate in Oklahoma City. At the time of data collection, Hearst Communications owned the local ABC affiliate, KOCO. Nexstar Media group owned the NBC affiliate, KFOR, and locally based Griffin Communications owned the CBS affiliate, KWTV.

Oklahoma City is the state capitol and a main population center in what is otherwise a relatively rural state. Politically, Oklahoma is conservative, with more than 65% of voters casting a ballot for Donald Trump for president in 2020 (Oklahoma State Election Board, n.d.). The U.S. Census, 2019 data reports the Oklahoma City area is approximately 54% white/non-Hispanic, 19% Hispanic, and 15% Black or African American.

Materials: Late local news

The 10:00 p.m. weekday local newscasts for the Oklahoma City ABC, NBC and CBS affiliates were recorded from February 12, 2020, through March 4, 2020. The dataset is comprised of 48 newscasts, 16 from each station. News stories from the start of the broadcast until the main weather segment were eligible for coding (most newscasts start the weather report about 17 minutes into the program). In addition to typical spot news reports, the data collection period included the weeks leading up to (and the day after) the Oklahoma presidential primary and early local coverage of the COVID-19 pandemic. Local campuses experienced several racial incidents that garnered media attention, and the Oklahoma state legislature (which is only allowed to meet for 5 months a year) was in session during the data collection period. Data collection ended before the onset of severe weather season, which can have an outsized effect on the local news agenda.

Coding process

Since the study focuses on how sources are incorporated into news, news stories or packages were coded when either (a) they included a source other than a journalist or (b) they included some form field reporting. For example, a story featuring only a news anchor reading script or narrating over video from the studio would not be coded, but if the studio anchor “threw” to a reporter in the field who narrated footage, the story would be coded. A story which included a quote from a source, even if the source was quoted as a visual element (text on screen) is included, even if the quote was read by the anchor. Each broadcast provided an average of 8.2 min of codable content. Excluding reports that comprised only journalists narrating from the studio focused coders’ time and attention on the role of sources in finished packages. However, one should note that results will understate the dominance of journalists’ voices in newscasts.

Stories were coded by network, date, and type (crime, accidents/disasters, government, business, health/environment, race, human interest, or other). Each story was coded into a single category based upon which topic predominated. All stories about COVID-19 were coded as health/environment. Since the “other” category accounted for less than 1% of coded story types, the categories are exhaustive.

Within stories, every speaker turn was coded, so every turn generated a new unit of observation. In other words, when a journalist stopped speaking and the person they were interviewing (or a source on camera) began speaking, a new observational unit started. When an anchor stopped relaying information and started reading a quote from a source that appeared on screen, a new unit of observation was also created. When the anchor finished the quote and returned to narrating, another observational unit was created. Therefore, stories are parsed not by speakers but by turns. In many stories, the same speaker has multiple turns.

Turns are coded according to how long they last (in five second intervals) and the characteristics of the speaker. Coded characteristics for each speaker include type of source (journalist, public official, law enforcement, emergency services, celebrity/sports figure, advocacy group leader, businessperson, expert, citizen), apparent biological sex (male, female, cannot identify); apparent race/ethnicity (Hispanic; white, not Hispanic; Black or African American, not Hispanic; Asian, not Hispanic; cannot identify). Speakers who appear non-white and Hispanic (e.g., Afro-Latino), are coded as Hispanic. Less than 2% of source types were coded in the “other” category, indicating exhaustiveness. Less than 2% of sources could not be classified as appearing biologically male or female (this included cases where someone was quoted on screen but not named). Almost 20% of speakers were not classified by racial or ethnic identity, and intercoder reliability is somewhat lower for this variable than it is for gender. Some speakers could not be classified by race or ethnicity because they did not appear on screen. Ethnic ambiguity, such as white-presenting journalists with Hispanic surnames or Native American-presenting journalists with Anglo-Saxon surnames, also produced coding challenges. These racially ambiguous cases could result from the U.S.’s changing demographics, with more multi-racial people, but may also be the result of television stations intentionally making such hires to appear more diverse (Tapia et al., 2024).

To operationalize the two forms of epistemic authority (witnessing authority and narrative authority), each turn is coded for whether it bears witness to personal experience or characterizes a situation. Witnessing comprises relating one’s personal experience or describing what one has witnessed. Characterizing involves assigning meaning to a situation or speaking beyond one’s personal experience. For example, a person relating what happened to them during their arrest is witnessing. A person charging police misconduct is characterizing. Similarly, a speaker relating his or her personal feelings of loss upon the death of a friend is witnessing; a speaker explaining what sort of person the friend was is characterizing. Journalists explaining what happened are classified as characterizing unless they were eyewitnesses. The greater degree of coder judgment involved in classifying the substance of sources’ quotes means intercoder reliability is not as high as for straightforward variables such as gender or source type.

Intercoder reliability was established using a random sample of newscasts (two from each television station). Coders agreed on unitization of turns in 283 of 301 cases, or 94% of the time. All variables had intercoder reliability within acceptable limits (see Table 1).

Table 1.

Intercoder reliability for all variables.

Variable	Percent agreement	Krippendorff’s α
Network	100	1
Date	100	1
Story type	89.7	0.87
Source (speaker)	94.4	0.92
Apparent gender	98.3	0.97
Race/ethnicity of source	85.7	0.70
Witness/Characterize	91.4	0.74
Time speaking (ordinal brackets)	91.4	0.82

Note. Values calculated for two coders.

Analytic approach

The coding permits two possible measures of both story types and speaker turns: count and time. Story type can be measured with the number of stories falling into each category or the amount of time devoted to each category. Similarly, total number of turns or total speaking time could be used as measures. Chi-square tests of independence indicate no significant difference between measurement modes (time, turns, and stories) X² (14, N = 10,617) = 18.49, p = .19. Since it does not make a difference which measure is used, the following analysis uses story counts and turn counts as measures to simplify interpretation.

Although the data are coded in greater detail, this analysis groups sources into four types following the approach of Kleemans and colleagues (2017). The categories of “journalists” and “citizens” appear as coded. Public officials, members of law enforcement, and emergency services workers are collapsed into a category labeled “government authorities.” Celebrity and sports figures, advocacy group leaders, businesspeople, and experts are collapsed into a category labeled “civil authorities.” The analysis also collapses categories of story types into crime, government, and human interest, which are the three most common story types, and a catchall category of all other story types. This configuration provides increased statistical power, creating more interpretable results.

Chi-square analyses of story type and source type were conducted to evaluate the consequences of pooling data across all three channels. For story type, a significant relationship emerged X² (6, N = 281) = 14.68, p = .02. Chi-square pair-wise comparisons were conducted to find the differences between channels revealed in the initial test. Results indicate no significant difference between the ABC affiliate (KOCO) and the CBS affiliate (KWTV) X² (3, N = 201) = 3.85, p = .28 or between the NBC affiliate (KFOR) and the CBS affiliate X² (3, N = 170) = 3.74, p = .29. Differences between the NBC affiliate and the ABC affiliate are statistically significant and show that differences in story distribution are concentrated here X² (3, N = 191) = 14.23, p = .003. Story counts reveal these differences are focused in two categories, government stories, where the ABC affiliate has 35 and the NBC affiliate has 14, and other stories, where the ABC affiliate has 33 and the NBC affiliate has 13. Chi-Square indicated there were also significant differences in source prevalence X² (8, N = 668) = 19.70, p = .01. Pairwise comparisons between the three channels (NBC/ABC X² (4, N = 460) = 15.43, p = .004; NBC/CBS X² (4, N = 414) = 11.24, p = .02; ABC/CBS X² (4, N = 462) = 1.15, p = .89) suggests that NBC is significantly different than the other two networks in terms of local news source prevalence. Raw counts show the largest differences are the presence of more citizens on NBC and fewer of every other type of source. Data were pooled with these differences in mind since chi-square can be sensitive and the additional citizen data provides more data for citizen-focused analysis.

Findings

The first hypothesis was that citizens would appear less frequently as sources than government officials or civil authorities. This was tested in two ways: in terms of how many stories include citizen sources and in terms of the number of speaking turns by citizens. Table 2 depicts both the number and the proportion of stories that include at least one turn by journalists and by each type of source for all 281 stories in the data set. It also presents the proportion of stories that include these source types for each of the three most common story types and for the catchall category. The table shows that nearly half (47.3%) of all stories aired on the local newscasts included at least one citizen source, a proportion slightly lower than government sources, who appear in just over half (55.5%) of all stories, and ahead of civil authorities, who appear in less than one third (29.2%) of all stories. The table also shows that story type matters a great deal. Citizen sources appear in over two thirds of human-interest stories and in over 60% of crime stories, and for both these story types, they take at least one speaking turn in more stories than any other type of source. However, they appear relatively rarely in government stories. Across the less common story types, sources are relatively equal in the proportion of stories where they appear.

Table 2.

Source story appearance by story type.

	Combined	Crime	Government	Human interest	Other
	n (% of story type)
Journalist	281 (100)	98 (100)	73 (100)	45 (100)	65 (100)
Government	156 (55.5)	55 (56.1)	56 (76.7)	10 (22.2)	35 (53.8)
Civil authority	82 (29.2)	18 (18.4)	21 (28.8)	20 (44.4)	23 (35.4)
Citizen	133 (47.3)	60 (61.2)	14 (19.2)	30 (66.7)	29 (44.6)
Other	17 (6.0)	3 (3.1)	2 (2.7)	5 (11.1)	7 (10.8)

Note. Percentages reflect the proportion of stories of a given type that feature that source. Journalists appear in every story X² (12, N = 669) = 50.03, p < .000.

Comparing the number of stories in which a particular source type appears shows significant difference across story types, X² (12, N = 669) = 50.03, p < .000. Even when journalists are removed from the analysis, given that they are present in every story in the dataset, differences are still significant X² (9, N = 388) = 49.92, p < .000.

Measuring in turns produces a similar pattern, X² (12, N = 2561) = 256.73, p < .000. Table 3 depicts the number of speaking turns journalists and each source type have across the data set as well as the percent of those turns that are witnessing or characterizing. Turning first to the number of speaking turns taken by each type of source, the table and accompanying chi-square test reveal the dominance of journalists’ voices in the news. Recall that the data set includes only news items containing field reporting and/or sources other than journalists. It does not include studio-based news stories in which only journalists speak. Consequently, it underrepresents the preeminence of the journalistic voice. Still, the table reveals that over half of all speaking turns in the coded news stories are taken by journalists. Citizens are the second most frequent speakers in the news, taking more turns than either government or civil authorities. Government and civil authorities combined account for about 24% of speaking turns while citizens alone account for over 22% of turns.

Table 3.

Source turn function.

The results reported in these tables do not support H1. Rather, they reveal that citizens have more speaking turns than any other type of news source and that the type of story strongly affects the presence or absence of citizen sources.

Table 3 also tests H2, revealing the unique role citizens play in the news. H2 posited that citizens make more statements bearing witness than government authorities, civil authorities, or journalists and that government authorities, civil authorities, and journalists make more statements characterizing than do citizens. The table shows that sources like journalists clearly engage in one function on local news, while citizens clearly engage in another X² (4, N = 2561) = 814.00, p < .000. While all sources characterize situations, only citizens take a substantial number of turns bearing witness to personal experiences. The overrepresentation identified by the standard residuals in Appendix 1 reveals that citizens bearing witness contributes the most of all theorized variables to the statistical significance. Citizens’ witnessing in local news is the primary driver of difference in the distributions. Citizens take more turns bearing witness (315) than all other types of sources and the journalists combined (186). These results support H2. The results suggest that although citizens speak frequently in the news, their narrative authority, that is, the number and proportion of their speaking turns characterizing, is low relative to that of other sources who characterize in most of their speaking turns. Although government officials have fewer speaking turns than citizens (381 vs 565), they make more characterizing statements than citizens do (309 vs 250).

Together, the results so far suggest an interplay between how many speaking turns a source is given and whether those turns involve characterizing or witnessing, which likely affects sources’ ability to shape meaning in the news. Figure 1 illuminates how frequency and type of turn varies across story types and so offers insight into the research question: How do news sources’ frequency of speaking, witnessing, and characterizing differ from story type to story type? Like the earlier table, this figure breaks out the three most common story types in the newscasts (crime, government and human interest), and collapses the less common story types into an “other” category. Results clearly show that in every story type, journalists have the most narrative authority as gauged by both number and type of turns. However, variations appear among news sources, suggesting that citizens’ narrative authority relative to other sources may be undercut differently in different kinds of news stories. Consider, for example, the differences between citizens and government authorities in the context of crime stories and government stories. In crime stories, citizens have more speaking turns than government authorities, but well over half their speaking turns involve witnessing authority, describing their experiences, whereas over 80% of the turns by government authorities assert narrative authority, characterizing the situation. In contrast, there is relatively little difference between government authorities’ and citizens’ proportion of characterizing statements in the context of government stories (84% vs 73%). However, government authorities take three times more speaking turns in government stories than citizens do. The figure suggests that citizens have the greatest narrative authority in human interest stories, but even here, the number of characterizing statements they make is roughly equal to those of government and civil authorities (87 vs 68).

These results also suggest civil and government authorities may dominate the narrative in different areas of public life. Government authorities dominate in crime and government stories compared to civil authorities, with far higher numbers of turns in both types of stories and a greater proportion of characterizing turns in crime stories. Civil authorities play a larger role in human interest stories and in less frequently appearing story types. They speak more frequently than government authorities in human interest stories and are roughly on par with government authorities in how many turns they take and their relative frequency of characterization in other story types.

Table 4 uses a chi-square to test H3, that citizens’ gender and race affect the frequency with which their speaking turns bear witness or characterize. Since the number of turns by Hispanic and Asian citizen sources in the news is very small, the variable for race and ethnicity is collapsed into white and non-white to make analysis feasible. The relationship between race/ethnicity and frequency of witnessing or characterizing turns rises to conventional levels of statistical significance X² (1, N = 468) = 4.65, p = .03, but the relationship for gender X² (1, N = 555) = 3.51, p = .06 does not. A chi-square including both variables shows significant results X² (3, N = 468) = 9.96, p = .02. The table, and its underlying standard residuals, reveals that white women citizens are the key category, for they witness much more frequently than they characterize. Male citizens of any race tend to characterize about as frequently as they witness, and citizen women of color characterize somewhat more frequently than they witness. While these results must be read against the backdrop of Table 3, which shows that the number of witnessing turns taken by other sources is vanishingly small, these results nevertheless support H3: gender and race affect how frequently citizen sources witness. White women, in particular, have more witnessing authority and less narrative authority than other citizens.

Table 4.

Citizen demographic turn function.

	Witness		Characterize
	n (source %)	std. residual	n (source %)	std. residual
Male white	84 (51.2)	−0.71	80 (48.8)	0.79
Male black	16 (50.0)	−0.41	16 (50.0)	0.45
Female white	142 (62.0)	1.36	87 (38.0)	−1.51
Female black	17 (39.5)	−1.39	26 (60.5)	1.55

Note. X² (3, N = 468) = 9.96, p = .02.

Discussion

This analysis illuminates how journalists create a “normative order of authorized knowers” (Fishman, 1980: pp. 144-5) by demonstrating how different forms of epistemic authority are distributed among news sources. Journalists give all news sources the epistemic authority to characterize social reality by making claims that extend beyond their personal experience. However, citizen sources are distinctive in their witnessing authority to describe their experiences and emotions and so make unique contributions to the construction of journalistic authority by supporting claims to truth that derive from tangible experience (Muhlman, 2010).

Distinguishing between forms of epistemic authority also helps unpack critiques that citizen sources appear rarely in the news and that when they do appear, their contributions lack substance (e.g. Kleemans et al., 2017; Lewis, 2006). First, the data show citizen sources have more speaking turns than any other type of source in local news. Moreover, about half of their speaking turns invoke narrative authority, characterizing social reality beyond their personal experience. However, news sources in positions of social or political power exercise their narrative authority in almost every speaking turn, whereas half of citizens’ speaking turns invoke witnessing authority. To the extent narrative authority can be equated with substance, citizens’ employment of two kinds of epistemic authority makes their contributions to news less substantive than those of officials and civil leaders who almost uniformly employ only one. Citizens take more speaking turns in the news, but they take fewer turns than elite sources characterizing social reality.

The results also show the normative order of authorized knowers varies by story type, often in ways that coopt the epistemic authority of citizen sources. Citizens have the largest number of turns and, relative to other sources, the most narrative authority in human-interest stories, taking 87 characterizing turns compared to 68 for government officials and civil authorities. From a journalistic perspective, however, human interest stories have low news value (Parks, 2019). Citizens appear in a low percentage of government stories and take relatively few speaking turns in those stories even though most of their speaking turns in government stories employ narrative authority. Results for crime news, the most frequently appearing story type in local news, are intriguingly mixed, with citizens appearing in almost as many stories as government authorities and taking more speaking turns than government authorities. Here, it is the relatively even mix of characterizing and witnessing turns citizens take that undermines their narrative authority relative to that of government authorities, who take very few turns witnessing.

There is also evidence that the normative order of authorized knowers is affected by demographics. While the number of turns in which white women citizens employ narrative authority is roughly equal to that of white men citizens, white women contribute a disproportionate number of speaking turns bearing witness to personal experience, more than all the other race and gender categories combined. The small number of turns by citizens of color and their relatively high frequency of characterizing statements suggest a form of tokenism in the news, wherein people of color are expected to speak for their group. This implies a kind of toxic narrative authority in which the experiences of the few characterize the group as a whole (e.g. Eliasoph, 1998; Gilens, 1999). However, these differences are not statistically significant.

Previous scholarship suggests journalists have concerns about the trustworthiness of citizen sources (Andén-Papadopoulos and Pantti, 2013; Lewis, 2012), but the data presented here suggest a complex construction of source credibility. Citizens’ descriptions of their lived experiences, of events they personally witnessed, and of their emotional reaction to events establish them as witnessing authorities in news, making them part of the “symbolic map of society’s authorized knowers” (Carlson, 2017: p. 186). The realism and authenticity they impart to news contributes to journalistic authority. However, witnessing authority does not necessarily substantiate the narrative authority news requires to characterize the broader meanings of one’s personal experiences. Contrastingly, government authorities and social authorities rarely substantiate their narrative authority by appeals to witnessing authority, drawing instead upon their social status as a means of legitimating their characterizations of social reality.

One could argue that these outcomes are reflective of social reality rather than constructive of it – journalists really do have greater access (Carlson, 2017), experts and officials really do know more, and citizens really are less reliable than other sources (Andén-Papadopoulos and Pantti, 2013; Lewis, 2012). Journalism’s professional practices exist for a reason. However, professional practices can have an unacknowledged dark side (Carlson et al., 2021). In this case, citizen and minority sources may describe an event, but a speaker in a higher power role like a civil authority, government official, or journalist may characterize its meaning. In an example from the dataset, an elderly Black woman has a check stolen from her. While she describes discovering the crime and her subsequent frustration, the journalist and an expert characterize this as identity theft and explain what the woman (and the viewer) should do when this happens. In this and similar cases, individuals in authority roles speak for citizens and marginalized people, negating some of their agency (Edy and Austin, 2022). This act of ventriloquism, whether it results from material constraint or professional practice, produces at least two problematic outcomes. First, connecting social status with the authority to characterize social reality supports structures of inequality. To the extent that women and people of color are less likely to have the relevant social status, they are less able to characterize the social reality constructed in news. Second, these practices feed into populist discourses to the extent that they create a clear divide between citizens and experts. Telling average people how to make sense of their own experiences dismisses their judgment as unworthy, which may generate resentment of news media, as Sue Robinson (2009) found in her study of Hurricane Katrina and Arlie Hochshild (2016) found in her study of Tea Party activists. Populist discourse capitalizes on this disrespect and turns it to resentment of liberal democratic values (Peck, 2018).

Limitations and future directions

All data are limited by their period of collection, and this project is no different. Data were collected from the Oklahoma City media market over 16 days of local news coverage, including the beginning of the COVID-19 pandemic and the presidential primary. Expanding the study over a longer time frame and across more television markets might yield additional insights and nuances. Including weather, sports, and the final, feel-good kicker that typically ends newscasts could also enrich the findings. Operational definitions created to code content were designed for local news. Expanding the study to national news would involve revising the operational definitions to include more abstract content, including investigative reporting and data journalism which raise conceptual and operational questions about what constitutes witnessing. All of these would require substantial resources to implement.

Since the data are coded by turn rather than speaker, they give insight into epistemic authority but not to what might be termed “authorizing.” It is not clear whether one’s status as a witness authorizes one to invoke narrative authority about the meaning of what one witnessed. This study revealed nearly half the speaking turns taken by citizen sources employed narrative authority, raising the question of how their narrative authority was established by journalists, who may doubt the veracity of citizen sources. A study considering how the narrative authority of individual sources is established in news could address this question.

The results also raise new questions about how the demographic characteristics of sources relates to their narrative authority. Demographic differences presented here focused on the epistemic authority of citizen sources because they are the only ones taking a significant number of turns both characterizing and bearing witness to social reality. However, the demographic differences between men and women and between racial and ethnic groups may also extend to government officials and civil authorities. Results may be further complicated by white-dominated institutions employing women and people of color as media spokespersons.

Conclusion

Distinguishing between two types of epistemic authority, witnessing and narrative, illuminates their unique contributions to journalistic authority and reveals how journalism constructs an order of authorized knowers. Citizen sources employ narrative authority in some of their speaking turns, but they contribute uniquely to journalistic authority by lending witnessing authority to news stories. The epistemic authority created by citizen witnessing is commonly coopted by elite sources whose narrative authority is derived from their social status and who almost never make claims based upon witnessing. For example, in crime stories, eyewitness accounts that lend authenticity and realism to the story support the narrative authority of journalists and government authorities, who assign meaning to the broader social reality by characterizing the nature and significance of the crime. More broadly, citizen witnessing can lend an air of authenticity and realism to unrelated claims, as when human-interest stories containing eyewitness accounts by citizens contribute their unique witnessing authority to stories about healthcare or government that contain no citizen voices at all. Citizens may not reap the benefits in the form of narrative authority that the unique contribution of their witnessing authority generates, an inequity that shapes the normative order of authorized knowers in journalism.

Footnotes

Acknowledgements

The authors would like to thank Patrick C. Meirick for his assistance with data analysis and data visualization.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jill A Edy

Appendix

Table 5.

Chi-square test of source by narrative authority.

		Journalist	Government	Civil authority	Citizens	Other
Witness	Observed	17	72	75	315	22
	Expected	263.44	74.39	45.06	109.86	8.24
	Std. Residual	−0.94	−0.03	0.66	1.87	1.67
Characterize	Observed	1327	309	157	250	17
	Expected	1083.21	305.89	185.27	451.73	33.90
	Std. Residual	0.23	0.01	−0.15	−0.45	−0.50
Uncoded	Observed	30	7	3	8	4
	Expected	27.34	7.72	4.68	11.40	0.86
	Std. Residual	0.10	−0.09	−0.36	−0.30	3.67

Note. X² (4,N = 2561) = 814.00 p < .000.

Author biographies

Jill A Edy is Professor of Communication at the University of Oklahoma.

Chris Anderson is a doctoral candidate in Communication at the University of Oklahoma. He is Director of the Wabash Democracy and Public Discourse Initiative at Wabash College.

Margarita H. Tapia is a doctoral candidate in Strategic Communication at the University of Oklahoma.

References

Allan

(2013) Citizen Witnessing. New York: Polity.

Andén-Papadopoulos

Pantti

(2013) Re-imagining crisis reporting: professional ideology of journalists and citizen eyewitness images. Journalism 14(7): 960–977.

Bennett

(1990) Toward a theory of press-state relations. Journal of Communication 40: 103–125.

Carlson

(2009) Dueling, dancing, or dominating? Journalists and their sources. Sociology Compass 3(4): 526–542.

Carlson

(2017) Journalistic Authority: Legitimating News in the Digital Era. New York: Columbia University Press.

Carlson

(2020) Journalistic epistemology and digital news circulation: infrastructure, circulation practices, and epistemic contests. New Media & Society 22(2): 230–246.

Carlson

Lewis

Robinson

(2021) News after Trump: Journalism’s Crisis of Relevance in a Changed Media Culture. Oxford: Oxford University Press.

Chouliaraki

Al-Ghazzi

(2022) Beyond verification: flesh witnessing and the significance of embodiment in conflict news. Journalism 23(3): 649–667.

Cook

(1998) Governing with the News: The News Media as a Social Institution. Chicago: University of Chicago Press.

10.

Edy

Austin

(2022) Generating more inclusive media memory: the limits and possibilities of news archives. Media, Culture & Society 44(4): 802–819.

11.

Edy

Daradanova

(2006) Reporting through the lens of the past: from challenger to columbia. Journalism 7: 131–151.

12.

Ekstrom

(2002) Epistemologies of TV journalism. Journalism 3(3): 259–282.

13.

Eliasoph

(1998) Avoiding Politics: How Americans Produce Apathy in Everyday Life. New York: Cambridge University Press.

14.

Entman

Page

(1994) The news before the storm: the Iraq war debate and the limits of media independence. In: Bennett

(ed) Taken by Storm: The Media, Public Opinion, and U.S. Foreign Policy in the Gulf War. Chicago: University of Chicago Press, 82–101.

15.

Fishman

(1980) Manufacturing the News. Austin: University of Texas Press.

16.

Frisch

(1986) The memory of history. In: Benson

Brier

Rosenzweig

(eds) Presenting the Past: Essays on History and the Public. Philadelphia: Temple University Press, 5–17.

17.

Gilens

(1999) Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy. Chicago: University of Chicago Press.

18.

Godler

Reich

(2013) How journalists think about facts: theorizing the social conditions behind epistemological beliefs. Journalism Studies 14(1): 94–112.

19.

Godler

Reich

Miller

(2020) Social epistemology as a new paradigm for journalism and media studies. New Media & Society 22(2): 213–229.

20.

Hall

Critcher

Jefferson

, et al. (1978) Policing the Crisis: Mugging, the State, and Law and Order. London: Macmillan.

21.

Hallin

(1986) The “Uncensored War”: The Media and Vietnam. Oxford: Oxford University Press.

22.

Harris-Perry

(2011) Sister Citizen: Shame, Stereotypes, and Black Women in America. New Haven, CT: Yale University Press.

23.

Hochshild

(2016) Strangers in Their Own Land: Anger and Mourning on the American Right. New York: The New Press.

24.

Hopmann

Shehata

(2011) The contingencies of ordinary citizen appearances in political television news. Journalism Practice 5(6): 657–671.

25.

Kaniss

(1991) Making Local News. Chicago: University of Chicago Press.

26.

Kent

(2004) Gender history: law and politics. In: Meade

Wiesner-Hanks

(eds) A Companion to Gender History. Malden, MA: Blackwell Publishing, 87–108.

27.

Kleemans

Schaap

Hermans

(2017) Citizen sources in the news: above and beyond the vox pop? Journalism 18(4): 464–481. DOI: 10.1177/1464884915620206.

28.

Lewis

(2006) News and the empowerment of citizens. European Journal of Cultural Studies 9(3): 303–319.

29.

Lewis

(2012) The tension between professional control and open participation: journalism and its boundaries. Information, Communication & Society 15(6): 836–866.

30.

MediaTracks Communications (n.d.) Nielsen DMA rankings 2019. Retrieved from: https://mediatracks.com/resources/nielsen-dma-rankings-2019/

31.

Muhlman

(2010) Journalism for Democracy. Cambridge: Polity Press.

32.

Oklahoma State Election Board (n.d.) 2020 election results. Retrieved from: https://oklahoma.gov/elections/elections-results/election-results/2020-election-results/2020-november-general-election.html

33.

Parks

(2019) An unnatural split: how ‘human interest’ sucks the life from significant news. Media, Culture & Society 41(8): 1228–1244.

34.

Peck

(2018) Fox Populism: Branding Conservatism as Working Class. Cambridge: Cambridge University Press.

35.

Quandt

(2012) What's left of trust in a network society? An evolutionary model and critical discussion of trust and societal communication. European Journal of Communication 27(1): 7–21.

36.

Reich

(2015) Why citizens still rarely serve as news sources: validating a tripartite model of circumstantial, logistical, and evaluative barriers. International Journal of Communication 9(22): 773–795.

37.

Robinson

(2009) 'If you had been with us': mainstream press and citizen journalists jockey for authority over the collective memory of Hurricane Katrina. New Media & Society 11(5): 795–814.

38.

Ross

(2007) The journalist, the housewife, the citizen and the press: women and men as sources in local news narratives. Journalism 8(4): 449–473.

39.

Searles

Krupnikov

Ryan

, et al. (2023) Constructing Political Expertise in the News. Cambridge: Cambridge University Press.

40.

Sigal

(1973) Reporters and Officials: The Organization and Politics of Newsmaking. Lexington, MA: D. C. Heath.

41.

Tapia

Anderson

Edy

(2024) Creating a place in Middle America: how Spanish-language local news serves the hispanic community. Howard Journal of Communications 35(2): 183–199.

42.

U.S. Census (2019) QuickFacts: Oklahoma City, Oklahoma. Retrieved from: https://www.census.gov/quickfacts/fact/table/oklahomacitycityoklahoma/PST045219

43.

van Zoonen

Muller

Alinejad

, et al. (2007) Dr. Phil Meets the candidates: how family life and personal experience produce political discussions. Critical Studies in Media Communication 24(4): 322–338.

44.

Wahl-Jorgensen

(2013) The strategic ritual of emotionality: a case study of Pulitzer Prize-winning articles. Journalism 14(1): 129–145.

45.

Zelizer

(2007) On “having been there”: “Eyewitnessing” as a journalistic key word. Critical Studies in Media Communication 24(5): 408–428. DOI: 10.1080/07393180701694614.

46.

Zoch

Turk

(1998) Women making news: gender as a variable in source selection and use. Journalism & Mass Communication Quarterly 75(4): 762–775.