Peter A. Chew, a linguist and president of the data analytics consulting firm Galisteo Consulting Group, Inc., spoke to an April 9 Middle East Forum webinar (video) about the friction between popular and elite views of Islamism in France since last September’s beheading of French schoolteacher Samuel Paty.
According to sources cited by Chew, the Samuel Paty case signaled a shift in French attitudes about the threat of Islamism in France. One poll revealed that, following Paty’s murder, 87% of the French public believes that the constitutional principle of laicité, the secularism separating religion from civic affairs and the state in France, “is in danger.”
Chew set out to apply the science of data analytics to gauge popular and elite sentiment, particularly about the “highly sensitive” topic of Islamism. His methodology is related to a technique called “blind source separation” to “factor out [his] own biases” and “compare two sources statistically” to see if there are significant differences.
French teacher Samuel Paty was targeted and beheaded after he came under fire for displaying a cartoon critical of Islam during a civics lesson on freedom of expression. |
Chew used the “Twittersphere” to find data for popular opinion, collecting approximately 18,000 tweets (of all languages, mostly French) mentioning Samuel Paty published March 7-9, 2021, after the teenager who had accused the teacher of “Islamophobia” admitted to lying, which triggered a lot of social media activity. Chew sorted the massive number of tweets into “buckets” sharing similar key words that included a consistent component, enabling him to distinguish “the forest from the trees.”
Chew cautioned that Twitter is “not fully representative of population” as a whole and can give a “skewed picture of what the population thinks.” Additionally, Twitter can be “manipulated by bots” which Russia has used to promote “divisive opinions” and “sow division.” Still, Chew believes Twitter remains a valuable window into popular cultural attitudes.
To find data for the elite opinion, Chew collected a limited number of samples to assess media coverage about the Paty murder, including various media sources across Europe.
Chew used a measure in computational linguistics called “Pointwise Mutual Information” (PMI) to compare the social media and journalism data. Chew then used “human analysis” to draw conclusions from the data analytics.
Chew found that words used in reference to the assassination in each data set differed remarkably. The words “Islamism” and “Islamophobia” appeared far more often in the popular social media category, while neutral or vague words such as “obscurantism” were more commonly found to describe the assassination in the elite data set.
One “commonly recurring theme” in the popular opinion Twittersphere is that the accusation of “Islamophobia” frequently “shuts down criticism of Islam, and it serves as a call for murder, paints a target on the back of the victim.” Many tweets also included the sentiment that there is a double standard in the treatment of Muslims compared to non-Muslims, and anger about how welcoming France has been to Islam and Muslims. “Civilizationist ideas” expressing fears of losing the prevalent European culture were also common.
Media reports “downplay[ed]” the cause of Paty’s murder by avoiding words like “Islamism” and “Islamophobia.”
In contrast, media outlets appeared to downplay the cause of Paty’s murder by using vague terms like “victim of a controversy” and avoiding use of the words “Islamism” and “Islamophobia.” Chew said that media avoidance of such words may be for a “variety of reasons,” but his analysis did not seek to explain intent.
Chew said his use of data analytics provides insight into the “forest” of “populist complaints” about Islamism and has “strategic communication implications” for the media. News publishers will have more “traction” with the general public if they challenge elite “taboos” regarding avoided words.
Marilyn Stern is communications coordinator at the Middle East Forum.
