Do Reconstructive and Attributive Quotes in News Narratives Influence Engagement, Credibility and Realism?

ABSTRACT

Current trends of declining newspaper circulation on the one hand and the public distrust in the news media on the other stress the need for journalistic texts that the audience finds to be both engaging and credible. This study therefore tests the effects of two quotation types—(1) reconstructive quotes presenting what was said and thought by news actors during the news events, and (2) attributive quotes attributing information to news sources after the events took place—on readers’ engagement with and perceived credibility and realism of news narratives. In an experiment (N = 123), participants read a crime news narrative that included either only reconstructive quotes, only attributive quotes, both reconstructive and attributive quotes, or no quotes at all. Results indicated no differences between the four story versions in levels of engagement, perceived credibility and perceived realism. These findings challenge results from previous research as well as textbook recommendations, suggesting that readers’ experience and critical evaluation of news narratives are not necessarily influenced by quotations.

KEYWORDS:

• Newspaper
• narrative
• quotes
• credibility
• engagement
• realism

Introduction

Quotes are essential aspects of news stories (Zelizer 1995; Nylund 2003). Much research on quotation in journalism is concerned with linguistic and functional differences between direct quotes, in which a news source’s speech is rendered verbatim, and indirect quotes, in which a news source’s speech is paraphrased (Waugh 1995; Semino, Short, and Culpeper 1997; Vis, Sanders, and Spooren 2015; Van Krieken and Sanders 2018). In line with research on quotation in non-narrative discourse genres, a general assumption is that direct quotes increase both the liveliness and faithfulness of stories to a greater extent than indirect quotes (Clark and Gerrig 1990; Thompson 1996; Short, Semino, and Wynne 2002). This assumption is based on the conventional notion that direct quotes are fully authentic, allowing for more expressivity than indirect quotes while inhibiting the possibility to twist and distort a person’s original words (Toolan 2006). The exact accuracy and faithfulness of direct quotes is questionable (Lehrer 1989; López Pan 2010), however, but their implied faithfulness is nevertheless greater in comparison with indirect paraphrases (Short, Semino, and Wynne 2002).

Direct quotations may function to confirm a news story’s news values, to evaluate problems, to express subjective experiences, or to strengthen a news story’s truth-value (Nylund 2003; Grunwald 2005). Similarly, Zelizer (1989) lists a number of citation functions that can be divided into two central functions: enhancing the engagement of the reader with the news text and enhancing the credibility of the news text. First, quotes may positively influence readers’ attention and engagement by humanizing a news story and adding liveliness to it. Second, quotes may increase the credibility of a news text by adding precision and authority to it and emphasizing its neutrality, factuality, and truthfulness (Zelizer 1989). Both functions are of crucial importance in the current media climate, which is characterized by declining newspaper circulation on the one hand (Shim 2014) and public distrust in the media on the other (Lewis 2019). These trends call for journalistic products that the audience finds to be both captivating and trustworthy.

However, much remains unknown about which quotations can enhance a news text’s potential to be credible as well as engaging. The present study advances and tests the idea that the impact of direct quotes in news articles might be dependent upon the functional nature of the quote. The paper is structured as follows. First, research on the effects of quotations in the news will be reviewed as well as the inconsistencies between the results found in these studies. It will then be argued that a possible explanation for these inconsistencies might lie in a lack of differentiation between quotation types. Next, it will be discussed how the difference between reconstructive quotes (i.e., quotes reconstructing what was said or thought by news actors at the time the news events took place) and attributive quotes (i.e., verifiable quotes attributing information to news sources at a point in time after the news events took place) might relate to differences in the effects of news texts on engagement, credibility and realism. This discussion results in the formulation of hypotheses about the effects of reconstructive versus attributive quotes on the audience. These hypotheses are subsequently tested in an experiment of which the details are provided in the method section. The study’s findings are reported in the results section and interpreted and related to previous studies in the final section of the paper.

Functions and Effects of Quotes in News Discourse

So far, the effects of quotes in news articles on the audience have been tested in five studies. Gibson and Zillmann (1993) made a comparison between direct quotes and indirect paraphrases and found that direct quotes affect the opinions of audience members more strongly than paraphrases. News items with direct quotes also influenced opinions more strongly than news accounts without any quotes or paraphrases. These effects were found for print news, but not for radio news. A follow-up study manipulated direct quotes and indirect paraphrases in news texts expressing two opposing views (Gibson and Zillmann 1998). The results showed that readers base their perception and evaluation of social issues on statements expressed by persons who are quoted in the direct mode rather than on statements expressed by persons whose words are paraphrased in an indirect mode. These findings lead Gibson and Zillmann (1998, 173) to conclude that “direct quotation is a powerful journalistic tool that can be used to influences news consumers’ perceptions of issues.”

The discussed experimental studies focused on the capacity of direct quotes to influence beliefs and opinions of audience members, in short: their persuasive potential. However, persuasion is generally not considered to be the main communicative function of citations. Rather, key functions of quotations are to enhance the reader’s engagement and the news text’s credibility (Zelizer 1989). To date, however, only few studies have experimentally tested the effects of quotations on engagement and credibility. An early study manipulated newspaper articles on the inclusion of direct quotes versus indirect paraphrases (Weaver et al. 1974). The participants, consisting of journalism students, rated the articles on evaluative attributes such as accuracy, objectivity and believability as well as readability, interestingness and colorfulness. They furthermore rated the persons quoted in the articles on attributes such as believability, preciseness, and emotionality versus rationality. Results of the experiment showed no differences between direct quotes and paraphrases in readers’ evaluation of the news articles. Quotation type did affect readers’ evaluation of the quoted person, but only on two aspects: persons who were quoted in a direct way were rated as more emotional and dramatic compared to persons who were quoted in a paraphrasing way. From these results, Weaver et al. (1974, 404) conclude that “quotation marks simply do not make that much difference.”

Another study manipulated the presence versus absence of direct quotations in online news articles (Sundar 1998). Participants in this study read six news articles, half of which featured direct quotes and half of which featured no quotes at all. Their evaluation of each article was measured in terms of article credibility, article liking, article quality, and article representativeness. Results indicated that articles with direct quotes were rated as of higher quality and credibility than articles without quotes. The inclusion of quotes did not affect readers’ liking of the articles nor their perceived representativeness of the articles. These results lead to the conclusion that readers “do notice quotes” and “actively use them in evaluating story credibility and quality” (Sundar 1998, 63).

A third study compared a news text without direct quotes with a news text including quotes that were attributed either to government sources, security sources, or police sources (Matthews 2012). No differences were found between the various source types in perceived credibility of the news text. Moreover, and in contradiction with Sundar’s (1998) findings, the credibility of the news text was unaffected by the inclusion of direct quotations, indicating that readers find articles without quotes equally credible as article with quotes.

In sum, research has not yet produced conclusive evidence about the effect of quotes in news texts on text credibility and evaluation. It could be possible that these effects differ between news genres. Another possibility, which would also account for the inconsistent findings of previous studies, could be that these studies did not distinguish between different types of direct quotes. The present study advances the idea that the impact of direct quotes in news articles might be dependent upon the functional nature of the quotes. It examines this possibility by testing the effects of two functionally different direct quotation types—reconstructive quotes representing what the people involved said and thought at the time the news events took place versus attributive quotes representing the information provided by news sources at a point in time after the events took place—on engagement, credibility and realism in the journalistic genre of news narratives. Research on news narratives offers suggestions that reconstructive quotes primarily enhance a news text’s liveliness and, by implication, readers’ engagement with the news text, whereas attributive quotes primarily enhance a news text’s credibility. This will be explained in the following.

News Narratives: Reconstructive and Attributive Quotes

News narratives constitute a specific journalistic genre that has distinctive characteristics in terms of stylistic form, communicative function, and audience reception. Journalists writing news narratives employ “rhetorical techniques commonly associated with the “realistic” novel and short story” (Hartsock 1998, 62). These techniques include point-of-view writing, vivid details, thought reports, dialogues, and chronological (rather than inverted pyramid) structures (Johnston and Graham 2012; Vanoost 2013; Van Krieken 2019). A narrative style is often assumed to have advantages over more traditional hard news styles in catching readers’ interest in the harsh competition with 24/7 online news (Neveu 2014) and, as such, in revitalizing newspapers (Shim 2014; Groot Kormelink and Costera Meijer 2015).

The function of news narratives is to simultaneously inform as well as engage the audience. Experimental studies have shown that news narratives are indeed more engaging than non-narrative news reports (Oliver et al. 2012; Shen, Ahern, and Baker 2014; Van Krieken, Hoeken, and Sanders 2015). Under some circumstances, news narratives—written as well as broadcasted—have furthermore been shown to positively influence information comprehension (Machill, Köhler, and Waldhauser 2007; Emde, Klimmt, and Schluetz 2016; Kleemans, Schaap, and Suijkerbuijk 2018). These findings are in line with theories of Narrative Engagement (Busselle and Bilandzic 2009) and Narrative Persuasion (Green and Brock 2002; Slater and Rouner 2002) which posit that narratives are more engaging and, by consequence, more persuasive than non-narratives. Narrative engagement is considered to be a multidimensional experience which covers readers’ sense of being present in the world of the narrative, their identification with the narrative characters, their emotional engagement with the narrative, their focus on the narrative, and their understanding of the narrative (Busselle and Bilandzic 2009).

The degree to which readers become engaged with a narrative is influenced by specific textual characteristics, such as choices in narrative perspective (Hoeken, Kolthoff, and Sanders 2016) and referential expressions (Van Krieken and Sanders 2017). Quotations might constitute yet another text characteristic that influences narrative engagement; in addition, and specifically in journalistic contexts, they can be expected to influence a story’s credibility. These expectations follow from research showing that quotations fulfill one of two distinctive functions in news narratives (Van Krieken and Sanders 2016) that map onto the functions of enhancing engagement and credibility (Zelizer 1989). The formal difference between these functions lies in the temporal relation between the occurrence of the news events and the moment of utterance. First of all, quotations may reconstruct speech and dialogues that have been uttered during the news events, such that the events and the utterances overlap temporally. These quotes primarily serve to enhance the liveliness of news narratives and thereby readers’ engagement with these stories (Van Krieken and Sanders 2016). Excerpt 1 provides an example of such reconstructive quotes. This excerpt is taken from a Dutch newspaper story¹ about a man murdering his wife. The story provides a detailed narrative reconstruction of the run-up to the fatal event, the event itself, and its aftermath.

Excerpt 1

In only a couple of weeks’ time Johan transforms into a different man. He barely sleeps, feels panicky and has anxiety attacks. He is depressed. The loss of his job has swept the ground from beneath his feet. He feels redundant. He basically thinks about death non-stop.

“Dad,” his 10-year old son asks during the Christmas holidays. “Why don’t you smile anymore?”

“I’m stepping out of it,” is all Johan can think.

(De Volkskrant, 20 December 2012)

Excerpt 1 displays a represented question as well as a represented thought. These quotations can be characterized as reconstructions because the represented words have been uttered—either externally or internally—in a private rather than a public setting and are in that sense unverifiable. It must be assumed that the quotations have been reconstructed by the journalist based on information gathered at a later point in time, for example at a public occasion such as a press conference or court hearing. The inclusion of reconstructed quotations primarily serves to engage readers by presenting the narrative scenes in a dramatic, lively and realistic way (Craig 2006; Van Krieken and Sanders 2016). A potential risk of reconstructive quotes is that they add unwitnessed details to the news story to a degree that readers may “suspect that the reporter embellished the facts for the sake of a good yarn” (Frank 1999, 147).

Second, quotations may demonstrate what news sources have said at a point in time after the newsworthy events took place, in which case there is temporal distance between the events and the utterances. These quotations typically represent information and explanations provided at press conferences, interviews, or court hearings. As such, these quotations help journalists to attribute their information to sources in a verifiable way and are, by implication, expected to increase the credibility of news narratives (Van Krieken and Sanders 2016). Excerpt 2 provides an example of an attributive quote.

Excerpt 2

This week Johan sat in front of the judge, crying. […] “It is clear to me that the medication has pushed me far away,” Johan says. “I regret it deeply,” he says.

(De Volkskrant, 20 December 2012)

Excerpt 2 features a quote that has been uttered at a moment in time after the central news event—the crime—took place. The words captured between quotation marks have been uttered publicly, during a court hearing, and are therefore verifiable. Although such attributive quotes may also add liveliness to a news narrative, their dominant function is to legitimize the narrative reconstruction of events and to underscore the narrative’s truthfulness and credibility (Van Krieken and Sanders 2016). Attributive quotes are, thus, considered essential to demonstrate truthfulness and credibility, but their use comes with the risk of disrupting readers’ engagement. Narratives create a dream world in which readers get absorbed, but an attributive quote “interrupts this dream and may push the reader away from the story” (Craig 2006, 64).

Summarizing, reconstructive quotes differ from attributive quotes in several respects. Reconstructive quotes represent speech uttered during the news events, are typically unverifiable, and primarily serve to enliven news stories. Attributive quotes represent speech uttered after the news events, are typically verifiable, and primarily serve to demonstrate the truthfulness of news stories. Thus, whereas reconstructive quotes are expected to enhance engagement but decrease credibility because of the lack of verifiability, attributive quotes are expected to enhance credibility but to decrease engagement by (temporarily) taking the reader out of the narrative world. News stories differ in their use of the two quotations types: whereas some stories include only reconstructive quotes, other stories include only attributive quotes and yet other include a combination of both reconstructive and attributive quotes (Van Krieken and Sanders 2016; Van Krieken, Sanders, and Hoeken 2016). News narratives combining both quotation types may add to their perceived realism, i.e., enhance the reader’s belief that “the narrative world is reflective of the real world” (Cho, Shen, and Wilson 2014, 830). The concept of perceived realism plays an important role in narrative engagement and persuasion theories because perceptions of realism are more important to narrative processing than the status of a narrative as being either fictional or factual (Busselle and Bilandzic 2008). Like narrative engagement, perceived realism might be affected by the use of quotations. Specifically, reconstructive quotes add details to the described scenes and portray the news actors as speaking and thinking—and hence “realistic”—persons, while attributive quotes signal the truthfulness—and hence “realistic” nature—of the news narrative.

Hypotheses

The integration of theories of narrative engagement and research on the functions and effects of quotations in journalism leads to the formulation of a set of hypotheses about the impact of reconstructive and attributive quotes in news narratives on readers’ engagement, credibility judgments, and perceptions of realism. Together, these measures represent the degree to which quotations fulfill their dual function of enhancing engagement and credibility (Zelizer 1989; Van Krieken and Sanders 2016) and, in the specific genre of news narratives, how quotations help these narratives to meet “the double constraints of aesthetic persuasiveness through concrete detail, and ethical persuasiveness through the attempt to test details against an external reference point; an ‘other’” (Greenberg 2014, 529), that is: to be both engaging and credible at the same time, in a realistic way.

Previous research indicates that attributive quotes do not necessarily disrupt narrative engagement (Van Krieken, Hoeken, and Sanders 2015), implying that both news narratives with and news narratives without such quotes can engage readers. Nevertheless, it can be expected that attributive quotes result in lower engagement with news narratives compared to reconstructive quotes and the absence of quotes (Craig 2006). The first hypotheses are therefore formulated as follows:

H1a: News narratives with reconstructive quotes result in stronger reader engagement than news narratives with attributive quotes.

H1b: News narratives with reconstructive quotes result in stronger reader engagement than news narratives without quotes.

Research showing that direct quotes can increase a news article’s credibility made use of experimental stimuli including only attributive (rather than reconstructive) quotes (Sundar 1998). Since attributive quotes are verifiable whereas reconstructive quotes are unverifiable (Craig 2006; Van Krieken and Sanders 2016), the following hypotheses were formulated:

H2a: News narratives with attributive quotes lead to higher perceived credibility than news narratives with reconstructive quotes.

H2b: News narratives with attributive quotes lead to higher perceived credibility than news narratives without quotes.

Both types of quotes may be needed to convince readers that the described events are realistic and, thus, to enhance the news narrative’s perceived realism (Cho, Shen, and Wilson 2014). This leads to the final hypotheses:

H3a: News narratives with attributive and reconstructive quotes are perceived as more realistic than news narratives with only attributive quotes.

H3b: News narratives with attributive and reconstructive quotes are perceived as more realistic than news narratives with only reconstructive quotes.

H3c: News narratives with attributive and reconstructive quotes are perceived as more realistic than news narratives without quotes.

To test the hypotheses, an experiment was conducted of which the details are described below.

Method

Stimulus Materials

An original Dutch news narrative was selected and manipulated on both the use and type of quotations. This news narrative of over 2,000 words was originally published in the Dutch broadsheet newspaper De Volkskrant on December 20, 2014. The narrative reconstructs the murder of a woman by her husband and focuses on the period prior to the murder, in which the husband loses his job, sinks into a depression and starts developing thoughts of suicide and murder upon starting taking antidepressants. The story is written in the present tense and mainly describes the events from the internal point of view of the husband, expressing his thoughts, feelings and perceptions. To illustrate the narrative style of the story, excerpt 3 presents the opening sentences.

Excerpt 3

It is in the middle of the night and Johan is lying awake. Next to him, in a basket on the ground, are a hammer and a knife. Hidden under a book. They have been there all night and Johan cannot think of anything else.

(De Volkskrant, 20 December 2014)

The original story includes reconstructive speech and thought reports in the present tense, uttered during the narrative events (e.g., “What are you going to do?” he asks) as well as attributive speech reports in the present or past tense, uttered at a court trial months after the events took place (e.g., “Only then did the absurdity get through to me,” he later explains).

The story was shortened and then manipulated on the use and type of quotation to create four different versions (see Table 1 for excerpts of each version). The first version featured only reconstructive quotes. This version included ten direct quotations representing what the news actors were saying and thinking during the narrative events. The second version featured only attributive quotes. This version included ten direct speech reports representing what the news actors were saying after the events, during a court trial. The third version featured five reconstructive quotes and five attributive quotes, i.e.,: half of the quotes from the story version with reconstructive quotes only and half of the quotes from the version with attributive quotes only. No quotes were included in the fourth version of the narrative.

Table 1. Example excerpts of the four story versions.

Reconstructive quotes 893 words	Attributive quotes 924 words	Reconstructive quotes + attributive quotes 907 words	No quotes 772 words
He then weakens and puts the hammer back into the basket. “This is too absurd,” Johan thinks.	He then weakens and puts the hammer back into the basket. “I thought: ‘This is too absurd,’” Johan later explains.	He then weakens and puts the hammer back into the basket. “I thought: ‘This is too absurd,’” Johan later explains.	He then weakens and puts the hammer back into the basket.
The hammer goes out of the basket. Into the basket. Out, in. Again and again he hangs above Nicole with the hammer. And again and again he thinks: “No, I won’t do it. I have to call 911 to have myself locked up.”	The hammer goes out of the basket. Into the basket. Out, in. Again and again he hangs above Nicole with the hammer. Johan: “Again and again I thought: ‘No, I won’t do it. I have to call 911 to have myself locked up.’”	The hammer goes out of the basket. Into the basket. Out, in. Again and again he hangs above Nicole with the hammer. And again and again he thinks: “No, I won’t do it. I have to call 911 to have myself locked up.”	The hammer goes out of the basket. Into the basket. Out, in. Again and again he hangs above Nicole with the hammer.
[…]	[…]	[…]	[…]
Then Nicole enters the room. She grabs her purse and turns on the light. “Where are the pills?” she asks.	Then Nicole enters the room. She grabs her purse and turns on the light. “She asked: ‘Where are the pills?’”, according to Johan at the court hearing.	Then Nicole enters the room. She grabs her purse and turns on the light. “She asked: ‘Where are the pills?’”, according to Johan at the court hearing.	Then Nicole enters the room. She grabs her purse and turns on the light.
When she tries to walk away, he suddenly stands in front of her. They meet at the foot end of the bed.	When she tries to walk away, he suddenly stands in front of her. They meet at the foot end of the bed.	When she tries to walk away, he suddenly stands in front of her. They meet at the foot end of the bed.	When she tries to walk away, he suddenly stands in front of her. They meet at the foot end of the bed.
“I am going to call an ambulance,” she says. “Your stomach needs to be pumped empty.”	Johan: “She said: ‘I am going to call an ambulance. Your stomach needs to be pumped empty.”	Johan: “She said: ‘I am going to call an ambulance. Your stomach needs to be pumped empty.”
[…]	[…]	[…]	[…]
When the police arrive, Johan walks outside. “I am the culprit,” he immediately says.	When the police arrive, Johan walks outside. “I immediately said: ‘I am the culprit,’” he explains during the court hearing.	When the police arrive, Johan walks outside. “I am the culprit,” he immediately says.	When the police arrive, Johan walks outside.

Measures

A questionnaire was developed to measure readers’ engagement with the story, the perceived credibility of the story and the perceived realism of the story. The measurement of readers’ engagement relied on the conceptualization of narrative engagement as a multidimensional experience (Busselle and Bilandzic 2009; De Graaf et al. 2009). The questionnaire included scales to measure five dimensions of narrative engagement: (1) the feeling of being present in the story world (narrative presence), (2) the identification with the main narrative character (identification), (3) the focus of one’s attention on the story world rather than one’s immediate surroundings (attentional focus), (4) the experience of emotions (emotional engagement), and (5) the overall understanding of the story (narrative understanding). For each item, participants indicated their agreement on a 7-point Likert scale ranging from 1 (completely disagree) to 7 (completely agree).

Narrative presence was measured with five items, taken from De Graaf et al. (2012): (1) During reading, I had a vivid image of the events in the story; (2) During reading, I had the feeling as if I was present at the events in the story; (3) While I was reading the story, I was in the world of the story in my imagination; (4) During reading, I pictured the described events; and (5) When I was reading the story, it seemed as if I was there in my thoughts. The reliability of this scale was good: α = .82.

Identification was measured with five items about the degree to which readers had put themselves in the position of Johan, the news narrative’s main character. The items were taken from the identification scale of De Graaf et al. (2012): (1) While reading the story, I had the feeling I went through what Johan went through; (2) In my imagination it was as if I was Johan; (3) During reading I put myself in the position of Johan; (4) When I had been reading for a while, it seemed as if I had become Johan in my thoughts; and (5) While I was reading, I pictured what it would be like for Johan to experience what was described. The reliability of this scale was good: α = .85.

Attentional focus was measured with five items, also adapted from De Graaf et al. (2012): (1) During reading, I was fully concentrated on the story; (2) During reading, I did not really notice things that happened around me; (3) When I read the story, my thoughts were only with the story; (4) During reading, I did not think for a while about the things that had been on my mind lately; and (5) While I was reading the story, I forgot my daily affairs. The reliability of this scale was good: α = .92.

Emotional engagement was measured with four items (De Graaf et al. 2012): (1) The story affected me; (2) I found the story moving; (3) The story stirred emotions in me; and (4) Because of the news story, feelings arose in me. The reliability of this scale was good (α = .90).

Narrative understanding was measured with three items taken from Busselle and Bilandzic (2009): (1) I had a hard time making sense of what was going on in the story; (2) My understanding of the characters is unclear; and (3) I had a hard time recognizing the thread of the story. The reliability of this scale was good (α = .71).

Credibility was measured in two ways: credibility of the story and credibility of Dutch journalism. The story’s credibility was measured with six items adopted from Sundar (1998): I found the story: (1) accurate; (2) believable; (3) biased (reversed); (4) fair; (5) objective; and (6) sensationalistic (reversed). The reliability of this scale was low (α = .57) and could not be improved by removing one of the items. Therefore, the six items were included as separate items in the statistical analysis.

The overall credibility of Dutch journalism was measured to examine the possibility that a single news story could affect readers’ overall evaluation of journalism’s credibility. This variable was measured with the same six items adopted from Sundar (1998): I find Dutch journalism: (1) accurate; (2) believable; (3) biased (reversed); (4) fair; (5) objective; and (6) sensationalistic (reversed). The reliability of this scale was good (α = .79)

Perceived realism was measured with five items based on items developed by Cho, Shen, and Wilson (2014): (1) The story is based on facts; (2) The story is realistic; (3) The story showed events that have really happened; (4) The story is true; and (5) The story is representative. The reliability of the perceived realism scale was good: α = .84.

Finally, in line with the study by Weaver et al. (1974), participants’ newspaper reading habits were measured with the question: How often do you read a newspaper? Answers were provided on a 7-point scale ranging from “Never” to “Daily”. This measure was included to control for individual differences in exposure to newspaper narratives.

Design

A between-subjects design was used. Each participant read one of the four story versions: the story with reconstructive quotes only (n = 31), the story with attributive quotes only (n = 31), the story with both reconstructive and attributive quotes (n = 31) or the story without quotes (n = 30). All participants responded to all items and questions after reading the news story.

Participants

A total of 128 participants took part in the study, which is comparable to the number of subjects taking part in previous studies on the effects of quotations in news stories (Weaver et al. 1974; Matthews 2012). Five participants were excluded because Dutch was not their native language. The final sample consisted of 123 Dutch participants (82.9% female), of which 60% took part for course credit. The remaining 40% did not receive any type of compensation. Participants’ age ranged from 17 to 66 (M = 27.5, SD = 13.5). Level of education varied from secondary education (11.4%) to middle-vocational training (5.7%), higher professional education (16.3%) and scientific education (66.7%). Participants were equally distributed over the four experimental conditions in terms of gender (χ²(3) = .696, p = .874), age (F (3, 119) < 1), level of education (χ²(18) = 20.68, p = .296), and newspaper reading habits (F (3, 119) < 1).

Procedure

The study was conducted online via Qualtrics. A participant pool including students as well as nonstudents was used to recruit participants. After agreeing to take part in the study, participants were randomly assigned to one of the four experimental conditions. They first read an introductory page with instructions stating that they were about to read a newspaper article that had been published in 2014 in newspaper De Volkskrant. They were asked to read the article in the way they would normally read a newspaper article and to answer a set of questions afterwards. The following page showed the story, which was divided into five parts, visually separated by three page-centered asterisks, for ease of reading. After having read the story, participants moved on to the next page which showed the first set of items. Each scale was presented on a separate page. The final page included the demographical questions. Participation took about 10–15 min and was anonymous. All participants also completed a second, unrelated study.

Results

The hypotheses were tested using multivariate and univariate analyses of covariance (MANCOVA and ANCOVA). Participants’ self-reported newspaper reading habit was included as a covariate to control for individual differences in exposure to news narratives.

The first analysis included story version as independent variable, the five dimensions of narrative engagement as dependent variables, and participants’ newspaper reading habit as covariate. The analysis revealed a trend towards a significant effect of the reading habit covariate on participants’ engagement with the narrative (Wilks’s λ = .91, F (5, 114) = 2.17, p = .062, partial η²= .09). The covariate was positively related to identification (F (1, 118) = 5.31, p = .023, η²= .04), emotional engagement (F (1,118) = 4.83, p = .030, η²= .04), and attentional focus (F (1, 118) = 5.05, p = .026, η²= .04). This means that the more frequently participants read a newspaper, the stronger their identification, emotional engagement, and attentional focus were. The covariate was not related to the dimension of narrative presence (F (1, 118) < 1) and neither to the dimension of narrative understanding (F (1, 118) < 1).

Table 2 shows the means and standard deviations for all narrative engagement dimensions after controlling for participants’ newspaper reading habit.

Table 2. Means and standard deviations (between brackets) for the various dimensions of narrative engagement by condition (1 = low engagement, 7 = high engagement).

	Reconstructive quotes	Attributive quotes	Reconstructive + attributive quotes	No quotes
Narrative presence	5.02 (1.11)	5.15 (0.99)	5.39 (1.07)	5.60 (1.20)
Identification	3.16 (1.27)	3.03 (1.39)	3.45 (1.13)	3.08 (1.38)
Emotional engagement	4.88 (1.39)	5.19 (1.36)	5.49 (0.79)	5.13 (1.32)
Attentional focus	4.41 (1.46)	4.95 (1.28)	4.87 (1.32)	4.95 (1.58)
Narrative understanding	5.45 (1.08)	5.68 (1.13)	5.74 (1.04)	5.79 (1.08)

Note: Means are adjusted for individual newspaper reading scores.

The multivariate analysis showed no effect of story version on narrative engagement (Wilks’s λ = .88, F (15, 315) = 1.03, p = .439). This finding does not provide support for Hypotheses 1a and 1b.

Table 3 shows the means and standard deviations for participants’ perceived credibility of Dutch journalism as well as their perceived credibility and realism of the news narrative after controlling for their newspaper reading habit.

Table 3. Means and standard deviations (between brackets) for the perceived credibility of journalism, the perceived credibility of the news story, and the perceived realism by condition (1 = low, 7 = high).

	Reconstructive quotes	Attributive quotes	Reconstructive + attributive quotes	No quotes
Credibility journalism	4.18 (0.64)	4.14 (0.82)	4.15 (0.78)	4.27 (0.52)
Credibility story:
- Partiality	4.95 (1.25)	4.24 (1.50)	4.38 (1.61)	4.41 (1.19)
- Sensationalism	3.88 (1.47)	3.87 (1.71)	3.18 (1.33)	3.75 (1.44)
- Objectivity	3.85 (1.41)	3.77 (1.77)	2.92 (1.53)	3.65 (1.59)
- Accuracy	4.55 (1.18)	4.74 (1.03)	4.49 (1.03)	4.27 (1.23)
- Credibility	5.11 (1.17)	5.34 (0.95)	5.36 (0.99)	4.73 (1.34)
- Fairness	4.71 (1.24)	5.16 (1.34)	5.19 (1.20)	4.83 (1.32)
Realism	4.50 (0.90)	4.46 (1.07)	4.48 (0.86)	4.33 (1.09)

Note: Means are adjusted for individual newspaper reading scores.

The second multivariate analysis of covariance included story version as independent variable, the credibility of Dutch journalism as well as the six items about the credibility of the news story as independent variables, and the newspaper reading covariate. The analysis revealed no effect of the reading habit covariate on perceived credibility (Wilks’s λ = .93, F (7, 112) = 1.19, p = .314). In contrast with Hypotheses 2a and 2b, the analysis showed no effect of story version on credibility (Wilks’s λ = .78, F (21, 322) = 1.36, p = .134).

A separate univariate analysis of covariance with perceived realism as dependent variable showed no effect of the reading habit covariate on perceived realism (F (1, 118) = 1.14, p = .287). The analysis furthermore revealed that story version neither had an effect on perceived realism (F (3, 118) < 1). This result does not lend support to Hypotheses 3a, 3b, and 3c.

Conclusion and Discussion

The aim of this study was to examine the impact of attributive and reconstructive quotes in news stories in an attempt to explain and possibly resolve conflicting evidence found in previous studies (Sundar 1998; Matthews 2012) as well as to advance our understanding of the role of quotations in readers’ experience and evaluation of journalistic texts. The results showed that levels of engagement, credibility and perceived realism did not differ between the narratives with quotes and the narrative without quotes. These results are in line with Matthews’s (2012) study, which showed that news articles with quotes and news articles without quotes were judged as equally credible. However, Matthews’s (2012) findings as well as the results of the present study contradict Sundar’s (1998) finding that quotations enhance a news text’s credibility.

A possible explanation for this discrepancy could be that Sundar (1998) studied online news articles whereas Matthews’s (2012) study and the present study examined quotations in offline newspaper stories. It might be possible that the effects of quotes on a news text’s credibility manifest only in online environments because in such environments the default reading mode might be one of distrust and low credibility, whereas the default reading mode for offline newspaper texts might be one of trust and credibility (see Mitchelstein and Boczkowski 2010). The inclusion of quotations might not be necessary to confirm or reinforce that credibility. Future research could test this explanation by examining the impact of quotes in a direct comparison between online and offline news articles. Studies in this direction should also take into account reader demographics, such as age. A limitation of this study is the sample, which included participants with a relatively young age on average. It might be possible that younger news consumers evaluate online versus offline news differently than older news consumers, which could result in differences in the impact of quotes.

The present study extends previous studies on the effects of quotation by including measures of five different dimensions of readers’ engagement, thus strengthening the link between experimental research on quotation effects on the one hand and theories of narrative engagement on the other, which hold that engagement is a multidimensional reading experience (Slater and Rouner 2002; Busselle and Bilandzic 2009). Although numerous journalism textbooks include the advice to use quotations as a means to engage the audience (see, e.g., Gibson, Hester, and Stewart 2001), no difference was found between the four story versions on levels of engagement. This result seems to challenge research arguing that direct quotations add liveliness and drama to news texts (e.g., Zelizer 1989). However, a possible explanation for the absence of significant effects on readers’ engagement with the news narrative manipulated in this study could be that the story topic and its events were engaging by nature. This possibility would be in line with the findings of a study comparing an environmental news story with a crime news story (Kelly et al. 2003). Results of that study showed that readers rated the crime story as more interesting than the environmental story. Yet, the mean engagement scores found in the current study do not point towards a ceiling effect (cf. Table 2), meaning that there was room for quotations to influence readers’ experiences. It is nevertheless possible that direct quotes increase readers’ engagement only when used in news articles covering topics that are considered less engaging by nature. Future research could examine this possibility by comparing the presence versus absence of quotes in news articles covering a range of topics, such as crime, finances, politics, sports, and environment.

Furthermore, the results of the present study did not support the hypotheses that attributive quotes would enhance credibility whereas reconstructive quotes would enhance engagement, and that a combination of both would increase the news story’s perceived realism. A possible explanation could be that readers are insensitive to the difference between attributive and reconstructive quotes. An alternative explanation could lie in the manipulation of quotation type. To maximize equality in the content of the attributive and reconstructive quotes, the attributive quotes were presented as quotes embedding a reconstructive quote. For example, the following quote was used in the story version with reconstructive quotes:

“I am going to call an ambulance,” she says. “Your stomach needs to be pumped empty.”

The attributive counterpart of that quote was as follows:

Johan: “She said: ‘I am going to call an ambulance. Your stomach needs to be pumped empty.”

The attributive quote thus included a slightly modified but complete version of the reconstructive quote. This might have increased readers’ engagement, which would explain why levels of reported engagement did not differ between the news story with attributive quotes and the news story with reconstructive quotes. Future studies comparing reconstructive quotes with attributive quotes that do not embed reconstructive quotes could shed further light on the presumed effects of different types of quotes (Van Krieken and Sanders 2016) and readers’ sensitivity to such differences. Studies in this direction could also include pre-tests to assess the potential of the manipulations to elicit effects in an effort to select the most appropriate experimental stimuli.

A main limitation of the current study is that it employed a single message design. Single message designs are vulnerable for potential confounds, and it has been found that treatment effects may vary across messages (Jackson et al. 1989). In the context of the current study, this means that the lack of quotation effects on readers might be attributable to the specific news narrative that was used in the experiment and that the results could have been different if another narrative had been used. Future studies could overcome this limitation by employing a multiple message design in which a variety of news narratives is used to test the impact of quotations on readers (Jackson and Jacobs 1983; Reeves and Geiger 1994). Such studies are important to advance our understanding of quotations in news stories and to draw general conclusions about the circumstances under which quotations do and do not affect reader responses.

To conclude, although the current study does not offer conclusive evidence about the impact of quotes in news texts on readers’ engagement, perceived credibility and perceived realism, it adds valuable insights to the existing body of knowledge by indicating that readers’ experience and evaluation of news narratives are not necessarily influenced by quotations. Moreover, a comparison of this study’s results with the results found in previous studies indicates that the potential impact of quotations might be dependent upon topic of the coverage as well as medium. This leads to new opportunities for future research examining the role of quotations in readers’ evaluation and experience of news texts. Such research is important in light of the central role of quotations in journalism (Zelizer 1989, 1995), the increasing number of direct quotes in news texts (Vis, Sanders, and Spooren 2012), as well as the current public distrust in the news media (Lewis 2019).

Disclosure Statement

No potential conflict of interest was reported by the author.

Additional information

Funding

This research was supported by the Netherlands Organization for Scientific Research [NWO; project number 275-89-038].

Notes

1 This newspaper story is the basis for the stimulus materials used in this study (see Method section).