‘Dehumanization can lead to
hatred’

Detecting dangerous
patterns in media:

PHOTO: REYER BOXEM

PROPOSITION

Dit onderzoek wordt gefinancierd door het Ubbo Emmius Fonds, universiteitsfonds van de Rijksuniversiteit Groningen. Benieuwd welke projecten het fonds financiert,
kijk op uef.nl. Ook bijdragen aan de wetenschap? Word Vriend of Supporter 
van het fonds. 

Titel van het project:

“Visual argumentative inferences and soft hate speech online: A corpus-based multimodal argumentative perspective to raise critical awareness”

Lead to hatred
Marta’s PhD project focuses on how the so-called European ‘refugee crisis’ of 2015-2017 is portrayed in the media. “’Crisis’ is a widely known term, but in academia we approach it very critically”, Marta remarks. The term is ideologically charged, she implies, so in fields that aim to be objective, it should be avoided: in academia, but also in journalism. “In my research I take a multimodal approach: I look at the language used in headings and subheadings, but also at the accompanying photos.”

Words are often explicit, but images, as Marta explains, can carry meaning much more implicitly. “For instance, refugees are often portrayed as a broad, anonymized mass of people”, she says. “Often, such representations erase individual identities and histories, dehumanizing the people involved. This may end up reducing the empathic engagement among audiences.”

As an example, Marta names one famous poster used in the Brexit campaign, depicting a large group of people walking in a field. The accompanying text reads ‘Breaking point’. “It was implied that masses of people were entering Britain”, explains Marta. “In reality this picture was taken at the Croatia-Slovenia border. The recontextualized photo was almost used as a war narrative. The makers wanted viewers to associate refugees with an emergency, a threat. If internalized, this might lead to hatred and aggression against the targeted populations.”

Trend of dehumanization
Marta studies how text and images are used to convey such implicit messages. “The first step is to develop a methodology to study large databases”, she says. Her project collected around 4,000 articles – and their accompanying photographs – about refugees between 2015 and 2017 in Italy, Malta and Greece. “We built a very large and fine-grained annotation scheme”, explains Marta. Annotation is a way to categorize data, allowing for systematic analysis. “We are now testing it with different annotators. We want to make it reproducible.”

Marta gives an example. “Imagine a picture of me, sitting at this table, drinking tea. The annotation makes you answer questions such as: who is sitting at the table? Is she included in the text of the article? If so, is she individualized, or categorized, for instance as ‘a young woman’? In addition to drinking tea, is she speaking? Is she an agent – is she actively doing something – or is someone speaking about her?”

The annotation allows researchers to do statistical analyses and detect patterns. “When you look at the articles in the pilot study, for instance”, says Marta, “you see that they hardly ever depicted refugees in any active speaking role. They were only walking, or approaching, or engaged in other material actions. They were often passivated, while the people speaking, or ‘making meaning’, were always politicians or officials. This fits into a trend of dehumanization of refugees, making them voiceless.”

‘Soft hate speech’
In her PhD, Marta has two main aims: to discover patterns like she described, but also to develop a methodology to pursue this kind of research. “For now, we are doing this annotation manually”, she says, “but my guess is that at a later stage, the process could be automatized. We hope to build a solid tool that allows researchers to study large datasets, and detect these representational patterns, also in other contexts or media.”

Ultimately, Marta and her colleagues aim to identify how these representations may develop into so-called ‘hate speech’: language use that is prosecutable because it incites to violence against vulnerable societal groups. “Many of the images and articles that we come across do not meet the legal threshold. This is called ‘soft hate speech’. But they do lead to intolerance, stigmatization, and exclusion. That is what makes this ‘grey area’ so important to study. If these patterns get normalized in society, they might lay the groundwork for ‘hard hate speech’.”

Structure of argumentation
Marta’s research is “super-interdisciplinary”, as she puts it. It touches not only on linguistics, research methodology and social science, but also on philosophy. “I am very lucky to have a well-balanced and successful team of supervisors”, she says. “This allows me to really dive deep into the structure of argumentation that can normalize hatred. We try to trace it back: what are the implicit assumptions in this context? What is the argumentation that makes certain premises stick?  I am sure our research could not be as solid if it wasn’t done at the interface of all of these disciplines.”

The team hopes to increase what they call ‘multimodal awareness’: people’s awareness of the fact that modalities other than text, such as images, can in fact convey messages. “People are less trained to be critical towards images”, says Marta, “and this is concerning, because images are becoming increasingly dominant. And we still lack a toolbox to deconstruct the meanings that are conveyed. They often pass under the legal radar, but also under the radar of individuals.”

In addition to a toolbox for researchers, the team aims to develop guidelines and education materials for professionals in media and in education. As for her own future, Marta envisions a career that combines research with its practical applications. “Not necessarily in academia”, she remarks. “Perhaps in a non-governmental organization.  Strengthening people’s capacity to detect harmful patterns is crucial – and I would love to contribute to that.”

Language shapes ideologies, and in turn, ideologies shape language, as Marta Marcora puts it quite eloquently. “This complex interaction manifests itself in our everyday lives, but also in our politics and power structures”, she says. “I’ve always found this fascinating. This connection between language and society, and how they influence and shape each other.”

Following this specific interest, Marta studied linguistics, communication, and social sciences. “I never wanted to be a specialist in a single field”, she explains. “Rather, I like to approach things from different angles. From different disciplines. And I like complex issues. I’ve always been like this.” She laughs heartily.

The interdisciplinarity is what drew Marta to the M20 Programme. This programme, initiated by the Ubbo Emmius Foundation, funds PhD scholarships for interdisciplinary research tackling major societal challenges. “I saw the project advertised on EURAXESS, the European platform for academic careers. It was like the project had my name written on it. It is very much linked to my Master’s thesis. Sometimes you just have to get lucky.” Coincidentally, Marta’s older sister was already living in The Netherlands, and she had a partner from Groningen. “I’ve always liked the city and its people. Super welcoming.”

TEXT: NIENKE BEINTEMA

How are text and images used to convey implicit messages, and to influence perceptions of migration? This is a question that intrigues Italian PhD student Marta Marcora. She studies how refugees are represented in online media. “Many images erase individual identities and histories, and thus contribute to dehumanization.”