From hate speech to its counter narratives

In what circumstances did the kNOwHATE project emerge? 

It came up during the confinement, in a brainstorming that resulted from an open call by FCT about the potential consequences of the pandemic on hate speech and hate speech incitement crimes. International reports already showed this trend of increasing hate speech directed at certain communities during the pandemic. Since I work in the area of prejudice and discrimination, I thought I would apply and use an interdisciplinary approach, crossing social sciences and technology. I contacted GAI – the Research Support Office at Iscte to find out if there were any researchers working in these areas at Iscte. They quickly referred me to several colleagues, we got in touch via zoom, and in about two or three weeks we wrote and submitted a proposal. But we didn’t get the funding. 

But you didn’t give up? 

No! Eventually a new opportunity arose, with the opening of a call by the European Commission. We decided to recycle the project, involving a wider consortium of academic partners and the target communities involved: several research units of Iscte, partners from INESC-ID and LARSyS – from Instituto Superior Técnico -, and several community partners, Casa do Brasil, ILGA, SOS Racism, and the Commission for Equality and Against Racial Discrimination. This application was more elaborate and complex, and it was eventually funded. 

This project seeks to produce knowledge about the expressions of hate speech in Portugal, about the counter narratives that exist, and to use this knowledge to inform, raise awareness and empower victims and policy makers

What do you want to achieve, in terms of the project’s goals? 

There is a great interest from the European Commission to understand hate speech and how best to combat and regulate it. There is a discussion in academia and in the public space about the most effective ways to regulate these phenomena: Should we block/denounce content, or does that violate the right to freedom of expression? 

In the project we want, on the one hand, to understand the different manifestations of online hate speech, the more explicit and open ones and the more implicit and subtle ones. Hate speech is something very complex to define. I am particularly interested in the nuances that this discourse adopts in relation to different target communities. The manifestations/expressions are not the same in discourse against the Brazilian immigrant community, against the Roma/Gypsy community, or against the African descendant community. The basic premise is that if we can identify these nuances, we can design more effective and specific strategies to combat hate speech.  

Another goal of the project is to understand counter speech, that is, the counter narratives that arise to hate speech with the aim of eliminating it. Some UN reports suggest that the production of counter narratives is an effective strategy to combat hate speech, which allows the empowerment of communities that are victims of hate speech, as well as people who are not targets and witness this speech, the bystanders, and who can also be important actors of counter discourse. Basically, the project seeks to produce knowledge about the expressions of hate speech in Portugal, about the counter narratives that exist and use this knowledge to inform, raise awareness and empower people and policy makers. 

What research methods were used? 

In the first phase, we had to conduct interviews and focus groups with people in the target communities, to find out how they experienced hate speech and if they identified any form of counter narrative. Doing this work during the pandemic was very challenging and the partners were fantastic in conducting these activities. We sent a preliminary qualitative analysis of this work to the European Commission, which confirmed that different communities, in our case immigrant communities, LGBTI+ communities and ethnic-racialized communities, highlighted different aspects and different forms of expression in hate speech. This bottom-up approach, brought by the communities themselves, was then crossed with the literature review and the conceptual framework we defined, very much anchored in approaches from Social Psychology and Linguistics. In the current phase, the Data Science colleagues are extracting a set of social network data to be further analysed. We have focused at this stage on Twitter – because it is easily accessible and allows interaction between people – and on Youtube videos, a more linguistically rich material because it does not have the character limitation inherent to Twitter. 

Decoding the messages captured online, how is this done? 

We are creating annotated guidelines. This work is being done by me, as a social psychologist, and by Paula Carvalho, a linguist at INESC-ID. We are going to develop a manual that will gather the guidelines that will then guide the research fellows in coding the first messages extracted by the data science team. It is these guidelines that define what is or is not hate speech, that characterize its multiple expressions, and that are then used to develop automatic detection mechanisms. From the experience of previous projects, we know that the complexity of this phenomenon is such that it is very difficult to automatically detect hate speech by simple keywords – a more complex analysis of expressions, sentences is required. Often people don’t use very extreme words because they know they will be detected automatically. Therefore, the linguistic approach brought by Paula Carvalho is fundamental to identify linguistic, rhetorical and discursive mechanisms underlying messages in which hate speech appears in an indirect, hidden way, resorting for example to the use of humour, irony, etc.  These more veiled and subtle expressions are equally damaging to the victims and difficult to detect. 

Different communities highlight different forms of expression in hate speech 

 

What do you hope to achieve at the end of the project? 

Essentially, we want to have a greater knowledge of this phenomenon, its multiple expressions, and ways to combat it. We want to have automatic detection tools that capture the different more “open” and more “subtle/hidden” expressions of hate speech in Portugal, but we know that the complexity of these expressions brings added challenges for their automatic detection. We want to produce knowledge that can empower the communities that are the target of these speeches, but also raise awareness and empower the bystanders of this phenomenon. 

We know that the more people are exposed to hate speech, the more they desensitize and normalize it, which can ultimately condition our ability to react and fight these speeches. At the end of the project, we hope to have this knowledge reflected in materials that feed awareness campaigns and inform policy makers. 

Rita Guerra

Researcher at CIS-Iscte 
Visitar perfil