1. Home
  2. Europe

Taking the temperature: Germany and refugees

Tracking Germany's mood swings with big data

Thousands of refugees, mostly Syrians, arrived at Munich central train station in Germany on 6 September, 2015. Gordon Welters/UNHCR

The tone of German media coverage of the refugee situation is wavering, and public interest past its peak, according to an analysis of two big data sources.

Our first-hand reporting from Germany also suggests the welcome may be wearing thin. 

In the graphic below, we plot the tone of over 80,000 articles published on German websites, as provided by the GDELT database, and compare that with Google's ranking of the volume of searches in Germany for refugee-related terms.

Key moments:

In early August, officials announced they expected 800,000 refugees to reach Germany. In the third week of August, Germany announced that Syrian refugees could stay, rather than be removed to their original point of entry to the EU. This corresponds to a further negative turn in the tone of German media coverage.

The photos of Syrian toddler Aylan [Alan] Kurdi published in the first week of September triggered a jump in searches on refugees in Germany. The mood of German news coverage peaks as Angela Merkel addressed the UN General Assembly on 25 September. 

The tone of the coverage, as measured by GDELT, never reaches a neutral value of 0. Further dips in the tone of coverage in October coincide with debates on the shaky EU quota plan and objections from the state of Bavaria about plans for refugees. 

After a dramatic summer, these two big data sources suggest the tone of news coverage about refugees has settled back to the levels of early August, but that public interest remains relatively high.

About GDELT's tone data : 

Computerized "tone mining" essentially assigns a score to each English word based on the emotional response it is most likely to generate; for example, recording that the word "delightful" traditionally has a positive connotation, while "horrific" usually connotes something bad, and saves this into a large dictionary. Tone-mining software then assigns an emotional score from positive to negative to a passage of text by looking up each word in the dictionary and computing the average score of all its words. While highly simplistic, this yields an approximation of the overall tone of a document. And when you feed in enough documents, patterns start to emerge."

Note: GDELT translates German automatically.

Share this article

Hundreds of thousands of readers trust The New Humanitarian each month for quality journalism that contributes to more effective, accountable, and inclusive ways to improve the lives of people affected by crises.

Our award-winning stories inform policymakers and humanitarians, demand accountability and transparency from those meant to help people in need, and provide a platform for conversation and discussion with and among affected and marginalised people.

We’re able to continue doing this thanks to the support of our donors and readers like you who believe in the power of independent journalism. These contributions help keep our journalism free and accessible to all.

Show your support as we build the future of news media by becoming a member of The New Humanitarian. 

Become a member of The New Humanitarian

Support our journalism and become more involved in our community. Help us deliver informative, accessible, independent journalism that you can trust and provides accountability to the millions of people affected by crises worldwide.