sentiment analysis

What did 9/11 mean for U.S. media coverage of Muslims and Islam?

In this article, we focus on broad patterns and trends in the coverage of Muslims to better understand the effect of 9/11. We show that key changes in the content of coverage were both nearly instantaneous and persistent. In particular, 9/11 dramatically tightened the connection in US newspaper reporting between Muslims, on the one hand, and terrorism and extremism, on the other hand. Moreover, the attacks also brought about an enduring increase in the volume of newspaper articles mentioning Muslims and Islam. These trends are evident not just in leading national titles such as the Times and the Post, but also across a range of more locally focused newspapers, including tabloids. On the other hand, although coverage also became more negative in the immediate aftermath of the attacks, this negative shock was comparatively short-lived, and the average tone of articles about Muslims and Islam had returned to pre-9/11 levels by the end of 2001. Sadly, those pre-9/11 levels were already systematically far more negative than for other minority groups in the United States.

Covering Muslims

We present the first systematic, large-scale analysis of American newspaper coverage of Muslims. By comparing it over time with reporting on other groups and issues as well as coverage of the subject in other countries, we demonstrate conclusively how negative American newspapers have been in their treatment of Muslims across the two-decade period between 1996 and 2016, both in an absolute sense and compared to a range of other groups. The same pattern holds in other countries, such as Australia, Canada, and the UK. While 9/11 did not make coverage more negative in the long run, it did dramatically increase the prevalence of references to terrorism and extremism.

Translation4tru (Translation for the Rest of Us)

Embedding-based word-level translation

Automated machine translation has come a long way, and neural machine translation (as used e.g. by Google Translate) is now very good. However, translating large quantities of texts for research purposes remains resource intensive. I make available bilingual word-level translation dictionaries that exceed in both size and quality any readily available alternatives. Translation4tru includes these dictionaries, along with python notebooks that can be used to apply them to translate text, as well as notebooks and code that can be used to develop translation dictionaries for additional language pairs.

Illustration by M.C. Escher - Tower of Babel (fair use)


High-quality dictionary-based sentiment analysis

Sentiment analysis (the measurement of the positivity or negativity of texts) is one of the most widely used tools in computational text analysis. MultiLexScaled is a sentiment analysis tool that can be applied off-the-shelf, has been proven to work well across a range of domains, and obtains performance comparable to dedicated machine learning applications.

2020 Report on Media Portrayals

In this Report, we analyze 2020 coverage of six significant American racial, ethnic, or religious groups: African Americans, Asian Americans, Native Americans, Latinos, Jews, and Muslims. Latinos and African Americans are the largest of these groups, constituting approximately 19% and 13% of the US population, respectively. Asian Americans are the next most numerous, at roughly 6%. Native Americans, Jews, and Muslims each make up approximately 1-2% of the American population. We address key questions about media coverage of these groups: Are some mentioned more frequently in US newspapers than others? How positive or negative is coverage of these groups, and why? How does 2020 compare to earlier years in terms of the amount and tone of coverage? What themes are present in the reporting of all six groups, and which were distinctive to each in 2020? In brief, our analyses show that coverage of African Americans stands out as being the most frequent and that of Muslims as the most negative, both by a wide margin.

The good, the bad, and the ugly

We examine how the US print media portray Latinos and Hispanics, drawing on computer-assisted coding of 185,244 articles in 17 American newspapers between 1996 and 2016. We explore the prevalence of themes of criminality, immigration, illegal immigration, and economic threats. Among these themes, we find that criminality and illegal immigration are associated with the greatest degree of negativity. Yet, the overall tone of articles is neutral rather than negative. Using collocation analysis, we examine the topics associated with positivity within Latinx articles. Stories relating to Latinx achievement and culture have the strongest positive associations with the tone of newspaper coverage. Our research thus identifies the themes associated with both negativity and positivity, and shows that coverage of Latinx has been relatively neutral rather than predominantly negative.

Atheism in US and UK Newspapers

We analyze coverage of atheism and atheists in American and British newspapers using computational text analysis techniques, including sentiment analysis and topic modeling. In particular, we show that greater negativity is associated with atheism as a concept than with atheists as individuals. Our findings add a new dimension to scholarship on differences between individual-targeted and group-targeted tolerance in public attitudes, establishing for the first time that media coverage mirrors such differences.

Media portrayals of Muslims

Are Muslims portrayed more negatively than other religious groups? If so, what factors are associated with this negativity? We apply computer-assisted, lexicon-based coding to over 850,000 articles that mention Muslims, Hindus, Jews, or Catholics in 17 national and regional US newspapers over the 20-year period of 1996– 2015 and compare them to a representative baseline of articles. We show that the average tone of articles about Muslims is considerably more negative than both this baseline and compared to articles about the other groups. The negative tone is most strongly associated with stories about extremism and events in foreign settings. However, even controlling for a wide range of factors does not eliminate the negativity in stories mentioning Muslims. We discuss the implications of these findings for media objectivity and for public attitudes and policy preferences with respect to Muslims and other social groups.

Afro-pessimist or Africa Rising?

Is media coverage of Africa systematically negative or increasingly positive? Several scholars have argued that too little empirical evidence exists to address the debate between “Afro-pessimist” and “Africa Rising” perspectives. We contribute to this discussion by analyzing 139,012 articles drawn from The New York Times, The Washington Post, The Wall Street Journal, and USA Today over the 25-year period between 1 January 1994 and 31 December 2018. We find modest support for the Afro-pessimist viewpoint: Articles mentioning Africa are negative on average, are even more negative during peak periods of coverage, and have not become more positive over time. In addition, we examine the thematic coverage most strongly associated with negativity and positivity. Stories that reference conflict, government, and specific African countries account for a significant portion of the negativity in our corpus. Conversely, stories related to culture and education constitute a subset of positive articles. Overall, our analysis not only sheds light on an ongoing debate about the tone of coverage of Africa, it also provides a better understanding of prevalent negative and positive thematic coverage in four major US newspapers.

Media coverage of Muslim devotion

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.