Media Coverage of Muslim Devotion: A Four-Country Analysis of Media Coverage of Muslim Devotion: A Four-Country Analysis of Newspaper Articles, 1996–2016 Newspaper Articles, 1996–2016

: Scholars have identiﬁed Muslims’ religiosity and faith practices, often believed to be more intense than those of other religious groups, as a point of friction in liberal democracies. We use computer-assisted methods of lexical sentiment analysis and collocation analysis to assess more than 800,000 articles between 1996 and 2016 in a range of British, American, Canadian, and Australian newspapers. We couple this approach with human coding of 100 randomly selected articles to investigate the tone of devotion-related themes when linked to Islam and Muslims. We show that articles touching on devotion are not as negative as articles about other aspects of Islam—and indeed that they are not negative at all, on average, when focused on a key subset of devotion-related articles. We thus offer a new perspective on the perception of Islamic religiosity in Western societies. Our ﬁndings also suggest that if newspapers strive to provide a more balanced portrayal of Muslims and Islam within their pages, they may seek opportunities to include more frequent mentions of Muslim devotion.


Introduction
Muslims are a highly stigmatized group in many Western societies.In recent surveys of 15 European countries, the Pew Research Center found that between 8% and 26% of respondents felt that the teachings of Islam promoted violence, and between 23% and 41% agreed that "Muslims want to impose their religious law on everyone else" (Pew Research Center 2018, pp.70-71).The Western media tend to portray Muslims in an overwhelmingly negative manner (Ahmed and Matthes 2017;  Knott et al. 2013), frequently associating Islam with terrorism, extremism, and a cultural otherness that conflicts with mainstream values (Cowan and Hadden 2004; Kalkan et al. 2009).Given the influence of the media in shaping public perceptions, this negative representation of Islam likely plays a role in how Muslims are viewed and treated by the public (Boomgaarden and Vliegenthart 2009; Entman 1989;  Saleem et al. 2016).
Scholars have identified Muslims' religiosity and faith practices, often believed to be more intense than those of other religious groups, as a point of friction in predominantly Christian-but generally secular-liberal democracies.Though there is little research focused centrally on Western media coverage of Muslim devotion, broader scholarship strongly suggests that devotion practices will be portrayed negatively given cultural strains perceived to exist between the West and Islam (Foner 2015,  p. 887; Brubaker 2016), because of tensions between predominantly Christian societies and Muslims (Casanova 2006; Pew Research Center 2018), or in light of what some see as pervasive secularism that is linked to skepticism of religiosity in general (Cesari 2007; Roy 2016).Yet, in this article, we show that Western media articles containing references to Muslims and devotion are not as negative as articles about other aspects of Islam-and indeed that they are not negative at all, on average, when focused on a key subset of devotion-related stories.
We employ both a quantitative and qualitative approach to understand articles related to Muslims and devotion in the print media.Quantitatively, we use computer-assisted methods to analyze a corpus of over 800,000 articles that reference either Muslims or Islam in major British, American, Canadian, and Australian newspapers from 1996 to 2016.To focus on stories most likely to reveal perspectives on devotion as a cultural issue within liberal democracies, we then identify a subset of those articles containing words related to devotion and that are not set exclusively in a foreign location and that do not deploy words related to extreme conflict (such as militia, insurgency or fanatic).We use lexical sentiment analysis to gauge whether the tone of these articles is positive, neutral, or negative.This process reveals that such articles are neutral to marginally positive, on average, rather than negative.To increase our confidence in these results, we subsequently read 100 randomly selected articles from this pool to check the validity of our quantitative findings.This qualitative analysis shows that while some articles are indeed negative, the majority are neutral to positive in tone.
Juxtaposing the neutrality or marginal positivity of these articles to the rest of the articles about Muslims demonstrates that while there is a great deal of negative coverage of Muslims in the media, stories mentioning devotion do not necessarily contribute to that negativity.We thus offer a new perspective on the perception of Islamic religiosity in Western societies.Our findings also suggest that if newspapers strive to provide a more balanced portrayal of Muslims and Islam within their pages, they may seek opportunities to include more frequent mentions of Muslim devotion, especially if they are set within the newspaper's own society.
To make our argument, we first discuss how prevailing scholarship suggests that Western print media coverage of devotion will be negative, and then explain in detail our data and research methods.Turning to our evidence, we elaborate on our quantitative findings and describe our qualitative findings, specifying the themes and trends that we discovered in our close reading of the 100 articles.Finally, in our conclusion, we provide a broader perspective by addressing the potential implications our findings have for debates over the role of religion in Western society, the promise of building interfaith understanding, and the print media's role in perpetuating Muslim stereotypes.

Expectations from Existing Scholarship
There is widely understood to be an underlying-and at times overt-tension in Western societies between the culture and practices of Muslims and those of what some scholars call "the Judeo-Christian mainstream" (Kalkan et al. 2009, p. 1).Brubaker describes Western Europe's "growing civilizational preoccupation with Islam" (Brubaker 2016, p. 2), while Foner explains that "Since the 1990s, a central issue in Western European public debates has been the concern that Muslims are failing to integrate, and that the cultures of Islam and the West are irreconcilable.Muslims are often disparaged for their values and practices by many in the general public as well as by politicians" (Foner 2015, p. 887).These practices, which range from the wearing of headscarves, to the ritual slaughter of animals, to rare examples of honor killings and forced marriages, are perceived by many to threaten the liberal values of countries like France and Britain that host large numbers of Muslim immigrants (Cesari 2007;  Foner 2015, p. 887).Looking specifically at Britain, for example, the 2018 Pew Research Center report found that 42% of respondents answered affirmatively to the statement that "Islam is fundamentally incompatible with our country's culture and values" (Pew Research Center 2018, p. 21).
Negative views and skepticism of Muslims are not exclusive to Western Europeans, however.In American minds, the "strange" beliefs and practices of Muslims also distance them from the majority (Kalkan et al. 2009, pp.2-3).For Kalkan, Layman, and Uslaner, Muslims constitute a "distinctive" outgroup that fits into both "bands of others" that Americans use to cognitively organize minority groups: racial and religious minorities, such as African Americans and Jews; and cultural minorities, which include illegal immigrants, gays and lesbians, and atheists (Kalkan et al. 2009, p. 2).This double otherness may generate a greater public suspicion of and antipathy towards Muslims than toward many other groups in American society.
It is hardly surprising, then, that this perception of "otherness" extends into the media realm.The Western media's portrayal of Muslims is widely understood by scholars to be quite negative (Ahmed and Matthes 2017; Knott et al. 2013, pp.87-88).However, our study is specifically interested in how newspapers treat devotion and devotion practices, of which there has been little to no systematic analysis.Notwithstanding, there are two distinct (even if somewhat cross-cutting) reasons for believing that the Western media's coverage of Muslim devotion will be predominantly negative.
First, despite a rapidly growing Muslim population, the vast majority of Western Europeans still self-identify as Christian, as do a similar majority of Americans. 1 The Pew Research Center also found that Christian identity in Western Europe correlates with higher levels of nationalism and "negative sentiment towards immigrants and religious minorities," including Muslims (Pew Research Center 2018, pp. 7, 9; Casanova 2006).We thus expect Western newspaper coverage of Muslim devotion to reflect this negative sentiment that leads to skepticism of Muslims, many of whom are not only religious minorities but also immigrants.
The second reason for anticipating that media references to devotion in articles about Muslims will be negative is that Western societies-although nominally Christian-have become increasingly secular over a number of decades (Brubaker 2016; Cesari 2007; Lewis and Kashyap 2013; Pew Research  Center 2015; Roy 2016).Analyzing laws and constitutional changes that aim to exclude religion from the public sphere, such as rulings against the wearing of the hijab and the secularization of public education, it is evident that faith is increasingly considered and treated as a private rather than public matter (Casanova 2006; Cesari 2007; Roy 2016).Numerous scholars argue that this secularization has brought along with it an increased skepticism and disparagement of religiosity, directed predominantly toward non-Christian religions (Brubaker 2016; Casanova 2006; Cesari 2007; Roy 2016).Cesari goes so far as to assert, "The idea that religion cannot contribute positively to the public good, a hallmark of secularized thought, is consistent through most of Europe . . .European perceptions of certain manifestations of Islam as troublesome or even unacceptable are thus a consequence of a more general invalidation of religion" (Cesari 2007, p. 37).Brubaker echoes this perspective, arguing that Western European secularism "makes the religiosity of populations of immigrant origin stand out as problematic" (Brubaker 2016, p. 2).
Although not all liberal democracies may be as secular as those in Western Europe, there is evidence that many of them have also experienced declining levels of religious affiliation over time.Even in the United States, which "remains home to more Christians than any other country in the world," the Pew Research Center found that "the Christian share of the . . .population is declining, while the number of U.S. adults who do not identify with any organized religion is growing" (Pew Research Center 2015, p. 3).In 2007, for example, 16% of the American population self-identified as religiously unaffiliated; by 2014, this number had climbed to 22% (Pew Research Center 2015, p. 3).In Canada, Jukier and Woehrling similarly conclude that "the percentage of people who claim no religious affiliation has increased dramatically over the past several decades" (Jukier and Woehrling 2010, p. 156).Finally, Australia has also seen both a "significant decline" in Christian self-identification and a growth in lack of religious affiliation and atheism (Ormerod 2010,  p. 16).
Secularists tend to focus attention on Islam in particular, as Muslims are often perceived to be more religious than non-Muslims.While scholars disagree over whether Muslims actually are more 1 The Pew Research Center (2018, p. 6) found that of the 15 European countries surveyed a median of 71% of the respective countries' populations self-identify as Christian.Its 2014 survey found that 71% of Americans describe themselves as Christian (Pew Research Center 2015, p. 4).religious than their American or European non-Muslim counterparts, they generally acknowledge that Islam is widely associated with a high degree of religiosity in Western society (Fischer et al. 2007;  Lewis and Kashyap 2013; Fish 2011).Fish (2011, p. 19) explains that "Islam is sometimes seen, by its very nature, as calling its followers to a greater level of devotion than other faiths demand of their adherents."Knott et al. (2013, p. 79) suggest that in Britain, Muslims display more "open religiosity" than the white middle class.
The secularization of Western societies, coupled with the widespread belief that Muslims are more religious than other groups, makes it reasonable to expect that the media will emphasize the extremeness of Muslim devotion and the tension it engenders in liberal democratic countries.In fact, Knott et al. (2013, p. 87) assert that British media discourse "positions Islam as a missionary religion and Muslims as those who seek to change British culture and practices."In a similar vein, Edmunds (2012,  p. 73) argues that "Islam alone is judged, in the media, to be fundamentalist, and other religions, which also contain fundamentalist strands, are absent from discussions of religious radicalization." In addition to identifying tensions surrounding Islam and Muslims in particular, studies that focus on the media suggest that religion in general receives significantly less newspaper attention than other aspects of American life-such as politics, business, foreign policy, sports, entertainment, and cooking, for example, which often have their own sections-and that when religious groups are discussed, it is usually in the context of scandal, violence, or perceived deviance from cultural norms (Wright 1997, p. 104).Cowan and Hadden (2004, p. 69) concur with this: "Nothing breaches the threshold of media significance like an intersection of religion and violence.In this moment it becomes significant; it becomes newsworthy."Routine faith practices, and other positive aspects of religion, on the other hand, are rarely deemed "newsworthy" and given media coverage (Cowan and Hadden 2004).
In sum, existing scholarship emphasizes the negative perception of Muslims in the United States and Western Europe, as well as cultural tensions between Muslims and Western society.It also stresses the rise of Western secularism and general skepticism about religion, as well as the association of Muslims with high religiosity, and the likely predominance of scandal, violence, or deviance in media stories about religion.Applying these findings to our analysis, we expect that newspaper articles that refer to Muslims and devotion will be predominantly negative, and that we will find few neutral or positive mentions of routine Islamic practices and rituals in newspaper articles.

Data and Methods
To explore media coverage of Muslim devotion, we use a combination of quantitative and qualitative techniques.We assemble an extensive corpus of articles from prominent newspapers published in Great Britain, the United States, Canada, and Australia over the 21-year period from 1 January 1996 through 31 December 2016.Our primary goal is to obtain a broad sample of newspaper coverage of Muslims in these four liberal democracies. 2Our sources range from left to right in political orientation, encompass different domestic geographic regions, and include broadsheets as well as tabloids.We also ensure inclusion of high-circulation national newspapers that frequently influence regional and local newspapers as well as elite opinion in each country.
Using searches of the Lexis-Nexis, ProQuest, and Factiva newspaper databases, we select all articles containing the root words "Muslim/Moslem" or "Islam" in the title or in the article text, for each newspaper across the entire 21-year period.After eliminating duplicate articles and incidental mentions (such as "Yusuf Islam," or "Islamabad"), our corpus contains 802,924 articles, of which 334,554 are from British newspapers, 254,804 appear in American newspapers, 120,788 are in Canadian 2 Our methods are best suited to analyzing English-language texts.We therefore selected large Anglophone countries for which media databases provided access to multiple newspapers.newspapers, and 92,778 are published in Australian newspapers.Table A1 in Appendix A displays the distribution of articles across all newspapers in our dataset.
To analyze the tone of coverage, we use automated sentiment analysis techniques developed by scholars in computational linguistics, machine learning, and business analytics (Grimmer and  Stewart 2013; Hopkins and King 2010; Young and Soroka 2012).Sentiment analysis aims to identify whether the tone of a text is positive or negative.In this article, we apply a lexicon-based approach that provides information not just on whether a text is positive or negative, but also about how positive or negative it is.Automated sentiment analysis can thus capture underlying positivity and negativity systematically associated with groups across the media as a whole.
We base our assessment of tone on eight widely used and externally validated general-purpose lexica developed by scholars by a range of different methods for a variety of tasks, each of which contains lists of positive and negative words. 3 To gauge the positivity or negativity of an individual article in our Muslim corpus, we use computer-assisted methods to compare each word in every article to the list of words in each lexicon.If the word in the article exists in the lexicon, the (positive or negative) valence score assigned by the lexicon is added to a running count. 4At the end of the article we divide the sum of the valence score by the total number of words in the article 5 and average the scores across all 8 lexica.This produces an initial score of positivity or negativity for each article.
Yet, even if a given article about Muslims has a negative tone, there is a chance that it is no more negative than the average newspaper article; research suggests that the media devote more coverage to negative stories than to positive ones, and that its coverage of negative stories is more intense than its coverage of positive ones (Garz 2014; Lengauer et al. 2011; Soroka 2006).What we need, therefore, is a method for calibrating the valence measures for our Muslim articles against a body of texts that is (on average) neutral within the context of newspaper coverage.Accordingly, we select a random set of articles from across all of our US and British newspapers to produce a "neutral corpus" of 107,720 articles. 6After conducting automated sentiment analysis as described above, we assign the average valence of this corpus a value of 0 and standardize the valence measures so that the standard deviation of the neutral corpus is set to 1.These calibration parameters are then applied to each article in our Muslim corpus.This allows us to say precisely how positive or negative each Muslim corpus article is relative to our neutral corpus of newspaper articles.
In addition to estimating the valence of articles within our corpora, we identify a subset of articles related to devotion through collocation analysis.Collocates are words that are more commonly found in proximity to one another than would occur by chance across the entire corpus (Baker et al. 2013;  Blinder and Allen 2016). 7We use collocation analysis to isolate words in our corpus of 802,924 articles that are proximate to the root words Muslim and Islam.Among these are devotion-related words such as devout, devotion, observant, pious, piety, faith, pray, pilgrimage, and fasting as well as their 3 For more information about these lexica, see the Supplementary Materials.We also account for intensification and negation (intensification refers to words that modify the strength of the polarity of a word, such as the "very" in "very good").

5
A long set of sentences with 5 words of positive valence may be perceived as neutral by a reader, while a single sentence with 5 such words will be read as having a strong sentiment.

6
Because our newspaper databases do not permit random sampling of articles, we generate a list of 17 words designated as neutral by our lexica (such as "boot," "fourth," or "throat") and use these as search terms to generate the neutral corpus.The resulting articles are thus not focused on any particular topic, and range from highly negative to highly positive in tone.

7
Collocation analysis requires that researchers define a window of words around terms of interest within which to search for collocates.For our analysis, we follow standard practices by using a 10-word window (five words each to the left and right of our Muslim target words, or "L5R5 collocates") and by examining what is called the "L1 collocate," which captures just the single word to the left of the target word (Blinder and Allen 2016, p. 12), which is often an adjective that modifies the main term of interest.variants and plurals. 8We then extract all articles that contain one or more mentions of any of these root words as a first step toward obtaining our devotion sub-corpus.
Given that our goal is to understand how the media frames stories that are primarily about the cultural aspects of devotion within Western societies, we make two further refinements to this sub-corpus.Existing research has shown that newspaper articles set uniquely in foreign locations (i.e., those outside of the newspaper's country of publication) or that contain words highly associated with conflict (such as terrorism, extremism, or radicalism) are likely to be negative in tone (Chermak and Gruenewald 2006; Entman 1991; Mertens and de Smaele 2016; Nossek 2004).To control for these effects, we use collocation analysis to identify a list of root words commonly associated with conflict, 9 and we use our computer-assisted method to identify articles that mention a foreign location, but contain no mention of a domestic location in the nation of the newspaper where the story is published.We then extract articles that contain a root word related to devotion and that do not contain a root word associated with conflict and that are not set uniquely in a foreign location to finalize our devotion sub-corpus.Therefore, our analysis of articles mentioning Islam/Muslims and devotion words (which we call "Muslim devotion" articles for short) excludes articles set solely in a foreign location as well as articles that contain words associated with conflict.This amounts to 37,454 articles, which constitutes 4.7% of all articles within our larger Muslim corpus.As shown in Table A1 in Appendix A, of these articles, 14,013 come from British newspapers, 13,672 from American newspapers, 5378 from Canadian newspapers, and 4391 from Australian newspapers.
This procedure thus allows us to compare the tone of articles in the Muslim devotion sub-corpus to the tone of articles in the larger Muslims corpus.We use the summarize function in the Stata statistical package to identify the mean predicted valence and standard deviation of articles within each set.We use the kdensity function in Stata to display the distribution of valences among articles within both the Muslim devotion sub-corpus and the remaining articles within the Muslim corpus compared to the neutral corpus.As discussed in the following section, these steps confirm that the Muslim devotion sub-corpus of articles contains substantially more positive articles than the rest of our Muslim corpus.
Given that our quantitative findings cut against the grain of existing scholarship that suggests such articles would be negative, it is possible that Muslim devotion articles may include latent negativity that our quantitative analysis did not adequately capture.We therefore undertake a qualitative analysis of a random sample of 100 articles drawn from the Muslim devotion sub-corpus. 10Each author independently analyzed all selected articles to assess whether human readers would view them on the whole as positive, negative, or neutral in tone.We also identified common topics present across the 100 articles to see if any particular type of coverage was especially associated with positivity or negativity in a way that our quantitative analysis might miss.This process permits a deeper understanding of the tone of articles that mention Muslims and devotion, and gives us greater confidence in our findings.

Quantitative Analysis of Media Data
Applying our quantitative methods to our entire corpus of 802,924 articles that mention Muslims or Islam demonstrates that these articles have a very negative tone, on average, with a mean valence of -.86 and a standard deviation of 1.07.The mean valence of our corpus is thus nearly a full standard deviation lower than the mean article valence within our neutral corpus of randomly selected articles 8 We do not include words related to clothing (such as veil) and diet (such as halal) in our searches, as these terms are frequently used in the context of political and policy debates in Western liberal democracies, rather than being simply about religious devotion.9 These include the root words: extreme, fundamental, ultraconservative, radical, militant, fanatic, Taliban, 9/11, jihad, al Qaeda, Bin Laden, ISIS, ISIL, Islamic State, Islamism, caliphate, theocracy, fascist, war, insurgency, revolution, separatism, militia, and armed. 10These articles ranged in valence from −2.04 to 2.24, and had a mean valence of 0.27.on all topics, which was set at 0 with a standard deviation of 1 for ease of interpretation and comparison with other corpora.To frame this finding in a more intuitive way, the average article in our Muslim corpus is more negative than 82% of all articles in our neutral corpus.It is thus probable that an article that mentions Muslims contains more negativity than over four out of five randomly selected newspaper articles.
Isolating the sub-corpus of "Muslim devotion" articles, however, presents a different story.Taken together, the 37,454 articles referencing Muslims and devotion (that aren't set exclusively in a foreign location and that don't contain conflict words) have a mean valence of +0.09 and a standard deviation of 1.09.This result reflects a great deal of neutrality, and indeed, even modest positivity.To provide a comparison to the Muslim corpus data provided above, the average article in the Muslim devotion sub-corpus is more positive than 52% of all articles in the neutral corpus.More importantly, the difference between our sub-corpus of devotion articles and all other articles in our Muslim corpus is stark.Extracting our devotion sub-corpus articles from the Muslim corpus leaves 765,470 articles.These remaining articles have an average valence of −0.91; an article with that valence would be more negative than 84% of all articles in our neutral corpus.
To illustrate these distinctions visually, Figure 1 compares the distribution of article valences in our devotion sub-corpus to all other Muslim articles and to the neutral corpus.11It shows that the vast majority of articles mentioning Muslims or Islam that are not within our devotion sub-corpus are negative, many substantially so.By contrast, articles mentioning devotion that do not contain conflict words and that are not set exclusively in a foreign location have, on average, quite similar valences to those of the neutral corpus.The quantitative evidence is thus clear.Contrary to expectations, articles that mention Muslims or Islam and that also mention devotion (but that do not contain conflict words and that are not set exclusively in a foreign location) do not have a negative tone. 12 We turn now to a qualitative analysis of 100 randomly-selected articles from this sub-corpus to enable a deeper understanding of this finding.

Qualitative Analysis of 100 Muslim Devotion Articles
Given the nature of the existing scholarly literature, we read the 100 randomly selected articles from the devotion sub-corpus with a skeptical eye, even in spite of our quantitative analysis. 13e recognized that our method of selecting articles would some that contained a devotion keyword (such as pray, faith, or devout) that were substantively unrelated to Muslims or Islam, or that were only tenuously about devotion itself.Yet for the majority of those that explicitly linked Muslims with devotion practices to at least some degree, we were prepared to find articles that reflected Western society's antipathy towards Muslims and broad skepticism of religion.Specifically, we anticipated articles emphasizing the "otherness" of Islamic practices, disparaging public expressions of Islamic devotion, and highlighting violent and scandalous events involving Muslims, perhaps couched in language that did not contain enough overtly negative words to trigger low values through automated sentiment analysis.
We did find some stories that fit these expectations.A few articles contain elements of Islamophobia, deliberately highlighting the "otherness" of Muslims and implying the superiority of Western culture.One such article from Britain's The Daily Mail, which mainly laments the decline of Christianity in the UK, asserts that "The absence of functioning democracies in the Middle East beyond the state of Israel is the direct result of a different set of religious teachings," and condemns Britain's effort to treat all faiths as equal. 14Although it contains references to faith, this article takes aim primarily at Islam as a religion rather than specific devotion practices.On the whole, blatant anti-Islamic sentiment in our devotion sub-corpus was rare; more frequently, articles with negative undertones were subtler in their critique of Islam, especially with respect to devotion.
In line with the arguments of Cowan and Hadden (2004) and Wright (1997), however, some of the articles in our sample did center on violence or scandal involving Muslims or Islamic events.Two stories discuss the stampede in Mecca during the 2015 Hajj pilgrimage;15 others cover the violent deaths of Muslim individuals. 16However, despite the negative subject matter and negative valences of these articles, we found that they more often than not seek to generate sympathy for Muslims and aim to dispel negative stereotypes associated with Islam.One article, for example, speaks of the high-profile killing of Amadou Diallo in 1999-but rather than focus on the violence itself or Diallo's identity as a Muslim immigrant in New York City, the story describes the Islamic acts of devotion performed by his family members at Diallo's memorial service and the international mourning that followed his death.17Another article, chronicling a Muslim father's murder of his daughter, weaves in a staunch defense of Islam, emphasizing that "murdering daughters is no more an Islamic value than murdering estranged wives is a Western one...murder is also not a phenomenon peculiar to Muslim families...To judge a faith and a culture on this one squalid incident is absurd." 18The scholarly literature does not often discuss the defenses of Islam and the opposition to negative stereotypes that we encountered in our articles.
Nor did we expect the media to portray Muslim individuals in a neutral or positive way when referring specifically to their religious practices or devotional acts.To be sure, there were some instances in which individuals were "tagged" as Muslim in an apparent effort to orientalize, or "otherize" them.
In these cases, individuals were situated in a negative context and the articles highlight how their behavior and actions violate Islamic values.Take, for example, The Sun's coverage of professional soccer player Kolo Touré's extra-marital affair with a prostitute: "They romped in a guest bedroom as Kolo pretended the house belonged to his parents who were away.Kolo who is a devout Muslim along with 32-year-old brother Yaya, actually lived at the house with his wife Awo and two children." 19espite the irrelevance of Touré's Muslim devotion to the story, the journalist's casual mention of it serves to reinforce negative associations with Islam.
However, the majority of articles that "tag" individuals as devout Muslims discuss their faith in a neutral or positive manner.Multiple articles describe the challenges that professional Muslim athletes face in trying to reconcile the demands of Islam with their athletic careers, acknowledging the determination required of the athlete in order to remain fully committed to both their team and religion.For example, Britain's The Sun interviewed and quoted soccer player Nathan Ellington: "I have played three games during Ramadan.I played 90 min 80 min and 60 min.It has been difficult.You have to fast between sunrise and sunset.Obviously during matches playing in the intense heat your mouth goes very dry as you can not drink." 20In Australia, The Age published an exuberant piece about Yassmin Abdel-Magied, the first female formula one race car driver. 21After describing her as a "conservatively-dressed Muslim woman," the article explains that she "revel[s] in breaking stereotypes" and has "clearly thrived and learned to traverse cultural boundaries"-in high school she became "the first girl at [her Christian college] to wear a headscarf."An article from Canada's The Globe and Mail, which recounts the tragic death of a father and daughter in a house fire, offers that the father "was a very nice guy, always jolly and laughing," as well as a "devout Ismaili Muslim" and "very religious." 22n addition, there are a number of articles discussing both Christians and Muslims that generally range from neutral to positive.The exceptions include a few texts that articulate Christians' frustration with what they perceive to be a double standard regarding the public display of religious symbols.In a British The Sun article, for example, a teenage girl banned from wearing her Christian chastity ring at school points out that "Muslim girls at our school wear headscarves and Sikhs wear Kara bangles, but the school refused to allow me to manifest my belief." 23This subset of articles is difficult to pinpoint as either positive, neutral, or negative; on the contrary, it lends itself to various interpretations.On the one hand, these stories could be seen to convey a type of envy on the part of Christians, who wish they shared Muslims' ability to display open religiosity.On the other hand, the articles could be understood to be somewhat Islamophobic, disparaging Muslims for being overly demanding that society accommodate their religiosity and for benefitting from policies that are not extended to Christians.Either way, articles such as this one call into question the widespread assumption that Muslims are more religious than other groups, and highlight the ongoing tension between secularism and religion, even in the secularized countries of Western Europe.
The majority of articles that reference both Christians and Muslims, however, use Islam as a neutral counterpoint, offering passing mentions of Muslims or Islam to prove a larger point rather than make negative commentary.An example is The Daily Telegraph's coverage of a landmark court case regarding religious discrimination in the workplace.The article quotes a lawyer's argument that "unlike the Muslim headscarf, wearing a cross is not a generally recognized act of Christian worship and was not required by scripture." 24This type of comparison was more typical in our set of 100 articles.
We also found that many articles portray Islamic religious practices largely benevolently, commonly depicting religious acts like prayer or the observance of religious holidays in a manner that displays tolerance of, if not admiration for, Islamic devotion.In a piece from the Richmond Times Dispatch, a Christian woman describes a positive encounter that she had with one of her Muslim colleagues: More recently, during a meeting of the Richmond Metropolitan Habitat for Humanity Executive Committee, my friend and colleague Imad Damaj arose and quietly asked if there was place in my office he could pray.The time in the day had come to pray and, as a devout Muslim, he heeded the call.No fanfare, no show; just a whisper in my ear.His witness inspires and humbles me-always. 25ntrary to the scholarly literature's assertions that Muslim religiosity is feared and shunned in secular Western society and that Christians in particular harbor anti-Islamic sentiment, articles like this one illustrate a clear respect for Islamic devotion.These articles also prove an exception to the observation that routine faith practices are not "newsworthy" enough to warrant media coverage.Some of the most positive articles discuss Islamic events (conferences, dances, etc.) and interfaith initiatives.American papers regularly publish listings of local events taking place that week that may include information about Islamic events, such as the International Islamic Heritage Festival or Eid-al-Fitr.At first glance, a brief mention of Islam in a long list such as this may seem insignificant.But, take this excerpt from the St. Petersburg Times: Following a Jewish event and taking place in a Christian Church, the "Understanding True Islam" gathering isn't hugely attention-grabbing.However, the casual and neutral inclusion in this listing normalizes Islamic observance as just another part of American daily life, no rarer or more extreme than Christian or Jewish events.
In our set of articles, we found that coverage of Islamic services and events themselves also tend to be positive in tone.The Sydney Morning Herald for example, reported on Sema, a Sufi ceremony in which "whirling dervishes" spin in a manner that is said to bring them closer to God.The coverage of the event quoted some of the participants who emphasize the beauty and spiritual significance of this act of devotion.One whirling dervish explained, "Beautiful voice and melodies take us away from ourselves and we enter a secret and mysterious world . . . the physical aspects are difficult to master but the mental effort is greater.Whirling is an expression of love from me for God, he says." 27This emphasis on the participants' perspective on the ceremony offers deeper insight into a little known religious practice without condemning its devotional aspect.
Aside from the local listings and events, there are also a number of articles about multiculturalism, interfaith dialogue, and the promotion of interfaith relations.These often explicitly advocate for religious tolerance and interfaith cooperation.Within these articles, comparisons between Judeo-Christian and Muslim holidays and practices are sympathetic and equalizing.In one example from Australia's The Age, a former Jesuit priest and now head of faith and mission at St. Kevin's College in Melbourne, speaks to religion and its importance.In the essay, he compares the three monotheistic religions, saying: "This weekend Western Christians celebrate Easter.At the same time, Jewish people are celebrating Passover.In a couple of months Muslims enter the holy month of Ramadan.These traditions are often seen as rivals but they share several elements."28Here, the priest expressly highlights the similarities between Judaism, Christianity, and Islam, asserting that these religions are not "rivals" and are not drastically different from each other.
Finally, some articles explicitly promote religious cooperation and increased understanding for the practices of devout Muslims.For example, a handful of articles report on Pope John Paul II and his efforts to bring Islam and Christianity together.One such article by The Boston Globe notes that the Pope timed an initiative "to coincide with the end of Ramadan, the Muslim month of fasting, because he wanted Catholics and Muslims to fast together for peace." 29The author of an article in The Toronto Star recounts joining a family in their observance of Ramadan for a day.His concluding comments display an understanding and recognition of the struggle that many Muslims contend with in Western society: "For us the fast was not so much about empathy for the poor, as experiencing the challenge Muslims face observing these rituals in a non-Muslim world . . .I cannot help but admire the determination of Muslims during Ramadan."30These articles not only recognize the significance of Ramadan but also advocate for a stronger understanding and acceptance of this practice and of Muslims in general.On the whole, therefore, we find that a majority of our 100 articles are either neutral or positive rather than negative, confirming our quantitative findings for the Muslim devotion sub-corpus.

Conclusions
Newspaper coverage of Muslims in our four liberal democracies has been strongly negative.The average article from our corpus of over 800,000 articles mentioning Muslims or Islam has a more negative valence than over four out of five articles from our randomly selected neutral corpus of articles.We anticipated that stories referencing Muslims and devotion would also be negative, especially given that scholars who write about Muslims in Western societies have emphasized that perceived religiosity is typically stigmatized.Yet, our combined quantitative and qualitative analyses show that newspaper stories that mention Muslims and devotion tend to be neutral or even modestly positive in tone.
These findings must be tempered by recognizing several limitations of our study.Our selection of devotion articles does not include those set exclusively in foreign locations or that contain words commonly associated with violence, conflict, or extremism.It also doesn't focus directly on certain cultural aspects of Muslim devotion that have been at the center of prominent public and political controversies, such as veiling, halal food, and adherence to sharia law.Our analysis therefore may not apply to all aspects of Muslim devotion that researchers may legitimately wish to investigate.Moreover, it is worth noting that most articles in our sub-corpus are not lengthy reflections on Muslim devotion.Often, this topic is a small portion of a much longer article.
Even with those caveats in mind, tens of thousands of articles have appeared in British, American, Canadian, and Australian newspapers over the past two decades that mention Muslims or Islam and that touch on aspects of devotion, and most of these articles do not have an overtly negative tone.These findings demonstrate that media coverage of devotion is more positive than the prevailing scholarly literature about Muslims leads us to believe.Our qualitative analysis also uncovered a significant number of articles that discussed religious devotion with nuance and even sympathy.This suggests that there is substantial room for further research on how Muslim devotion is perceived within Western societies that are commonly understood to be either predominantly (Judeo-)Christian, strongly secular, or both. 1 A small number of articles downloaded from Australia and Britain were not labeled according to the specific newspaper title by Lexis-Nexis. 2 The articles listed in the Muslim and Muslim devotion corpora for The Independent include articles from: i-independent print ltd, Independent on Sunday, Independent Magazine, Independent Traveller, and Independent.co.uk. 4

Figure 1 .
Figure 1.Valence distributions for Muslim devotion sub-corpus articles, all other Muslim corpus articles, and the neutral corpus.
Religion highlights TIMES STAFF.Secrets of Rejewvenation-What it Means to be God Partner, 9-11 March.Congregation B'nai Emmunah retreat held at Chinsegut Hill Conference Center in Brooksville.Members only, Call 938-9000; Foundations of Faith Understanding True Islam and Our Muslim Neighbors, 7 pm Monday.Chapel-by-the-Sea Community Church, 54 Bay Esplanade, Clearwater Beach. 26

Table A2 .
Valences for All Newspaper Articles Discussed in the Qualitative Analysis.