Within Western societies, the issue of migration is often accompanied by high levels of public anxiety and translates to a significant increase of the use of hate speech towards immigrants and minorities. Social media seem to be a fertile ground for hate speech. Focusing on the social dimension of hate speech, the project M-PHASIS seeks to study the patterns of hate speech related to migrants in user-generated content. The project will address the following aspects to provide a better understanding of the prevalence and emergence of hate speech in user-generated content in France and Germany:

  1. Advance the understanding and assessment of hate speech by considering multiple features of the phenomenon (lexical, syntactical and contextual facets of hate speech) and taking into account explicit and implicit forms.
  2. Develop a research protocol to detect hate speech in text and classify it in terms of its referents (i.e., themes associated with hate speech) and the representations conveyed, as well its circulatory characteristics.
  3. Improve the methods to detect hate speech in terms of validity, reliability, and the equivalence across cultures.
  4. Conduct a cross-cultural comparison of the prevalence of hate speech in France and Germany and the factors that give rise to hate speech in both countries (e.g., platforms on which comments appear, homogeneity of surrounding user-generated context, journalistic intervention).
  5. Proceed to the archiving and annotation of real-life examples of hate speech from social media sources, to be released to the research community for secondary analyses at the end of the project.

Our research hypotheses are that hate speech against migrants in social media:

  • Is context-dependent: it must be apprehended and analyzed in relation to its surrounding topical contents, its supporting media outlets as well as the sociocultural conditions of its appearance.
  • Materializes in different ways that can be systematized: hate speech can be understood through its linguistic features but can also be implicit, conveyed in more subtle manners.

In this project, we therefore want to examine what types of contexts relate to which types of hate speech. The project embraces an interdisciplinary approach to its object and seeks to benefit from the inputs provided by computerized processing of hate speech in social media. The insights gained will allow producing a software app detecting and/or blocking hateful comments automatically.

Resources developed during this project will be made accessible in Open Access Platforms to the scientific community. A web demonstrator will be implemented to validate the scientific developments achieved in the project.