The Civic Media Observatory: Making sense of people’s speech, memes, and tweets

Illustration by Giovana Fleck/Global Voices with Midjourney Bot

When a friend sends a meme through WhatsApp, and we laugh, we tend to overlook that we need a wide range of political, cultural, and historical knowledge to grasp its underlying message.

At the Civic Media Observatory, we decode the meaning of media items from different contexts in order to make sense of the world we live in – and the stories we tell ourselves.

Memes may seem trivial, but they can be vehicles for hate, propaganda, or false information. Newsrooms, citizens, and governments are often confused about what to do about wrong and false information. For example, misinformation is abundant but sometimes harmless and may simply reflect a lack of knowledge or context. Disinformation on the other hand is intentionally false information created to mislead.

A fact-checking industry has developed in response to these problems, and social media companies have endlessly tweaked content moderation rules and opened channels for people to report content that they define as misleading or harmful. Yet, unlike fact-checking, our goal is not only to say what is true or false but to explain the wider environment into which stories are woven.

The Civic Media Observatory explores key aspects of how information moves among humans, exploring the effects of emotions, cultures, and storytelling.

The stories we tell ourselves

Observatory researchers comb information networks to find relevant and important narratives, which are underlying assumptions and beliefs that help shape how people talk about things. We identify narratives by analyzing media items. A media item could be a witty tweet, a harmful meme on Instagram, an anti-war graffiti on a wall, a YouTube video, or a wire service news story. 

Depending on the project, researchers may delve into transnational themes, such as digital authoritarianism. For example, earlier this year, investigations revealed that journalists in El Salvador have been targeted by Pegasus spyware in their phones. What is less known is that El Salvador’s government has been blaming businessman and philanthropist George Soros for the surveillance. Anonymous Twitter users have also been drumming up suspicion of Soros. Our researchers found a pattern of similar attempts to blame Soros for installations of Pegasus spyware in other countries, including in Hungary. 

For Giovana Fleck, the Civic Media Observatory Lead, narratives are the key to understanding our media ecosystems. “How are narratives formed? Who pushes them, who's behind them, who's benefiting from them, how do they spread, impact people, and how can they be harmful?” she asks. “All those questions guide the CMO.”

In the case of Russia, our researchers closely tracked the escalation of narratives for months before Putin’s invasion of Ukraine. Examples of these narratives include: “The Russia-Ukraine crisis is a fiction created by the West” or “Russia always wins.” These narratives evolved after February 24, and Russian media ecosystems are now rife with altered storylines such as: “The Russian army is saving Ukraine and Ukrainians from Nazism” or “The war crimes that Russia is accused of committing are actually staged by Ukraine and the West.” 

Key to our research is understanding why people are creating and sharing a particular tweet, meme, or story and what impact it can have on society, especially when it reaches thousands of people.

Unlike automated tools that parse thousands of tweets, the Observatory focuses on qualitative analysis of our information ecosystems. We rely on local knowledge to explore nuances of speech, symbolism, and context.

Contextualizing information

To understand what people mean when they create or share information, context and subtext are key. That is why observatory researchers explain the overall political, historical, and sociological context surrounding any media that they are studying. They make the implicit explicit, by unpacking implied meaning and bias in texts and images.

For example, one of the Observatory’s India researchers searched Hindu nationalist pages on Instagram and found a harmful montage claiming that the Black Lives Movement in the United States is a hoax. At first glance, we might wonder why this video actually also propagates Islamophobic narratives such as “Islam preaches violence, Muslims are brutal, without culture” and “Hindus are under threat,” as the post never mentions Muslim people. But in the context of India, the video acquires another layer of meaning: it equates racist narratives targeting Black people in the United States with Islamophobic narratives targeting Muslims in India. The comments under the post confirm that this subtextual intent was not lost on the audience.

A systematic approach to journalism

“Approaching narratives requires a certain methodology,” says Giovana Fleck, and requires careful, critical reading. Fleck believes that this research approach has much to offer journalism. “It is reporting, it is investigating, and it is answering questions that we have about the world, but in a systematic form,” she says.

The work of discovering media items, detecting their underlying messages, and linking them to narratives helps researchers to evaluate the civic impact of a media item.

“The most important aspect of the CMO is to ask what are the possible impacts of an item,” explains Fleck. Some items have already affected social discourse, while others may lead to the hardening of a narrative, or spill over into offline action. Researchers are asked to consider hypothetical effects, based on empirical and local knowledge.

For example, Turkey’s biggest pop star, Tarkan, arguably has a strong influence on public opinion. Our researchers write that “he was generally seen as a non-political figure, but lately he is becoming a symbol for the opposition,” and has been vocal against developing infrastructure in Turkey’s historic and cultural sites. As one of Turkey’s most influential voices, his efforts to protect cultural heritage may have a positive civic impact. Likewise, a Tweet that includes subtextual racism may incite violence, and would have a negative civic impact.

“We are not predicting the future,” Fleck says, about the goal of the CMO, “but we want to understand how a piece of misinformation can impact people.”

For example, in 2020, our Myanmar researchers warned about the possibility of increased polarization in the country due to intense pro-military and anti-democratic narratives circulating online [PDF]. As our research was drawing to a close in early 2021, Myanmar lived through a coup d’état and violent repression on behalf of the Junta.

Freely accessible information

Readers and fellow journalists can access our public datasets on Airtable, a database platform. Datasets are regularly updated with new media items and narratives. Here are links to our 2022 country-based research and to our Unfreedom Monitor research on digital authoritarianism. This newsletter and our stories also aim to make our findings digestible for a general audience.

Open to collaborations and speaking

The Civic Media Observatory has initiated several collaborations with newsrooms. For example the Miami Herald has set up its own observatory to analyze narratives related to upcoming elections in Florida.

Our researchers regularly speak at international conferences and universities, such as the M100 Sanssouci Colloquium in Potsdam, CIPESA’s annual Forum on Internet Freedom in Africa in Lusaka, the Council of Europe’s Lisbon Forum, World Press Freedom Day in Uruguay, Mozfest, and RightsCon.

Observatory research has been supported by DW Akademie, BBC Media Action, the MacArthur Foundation, NED, and Meta, among others.

Learn more about the Civic Media Observatory

Start the conversation

Authors, please log in »


  • All comments are reviewed by a moderator. Do not submit your comment more than once or it may be identified as spam.
  • Please treat others with respect. Comments containing hate speech, obscenity, and personal attacks will not be approved.