FCT project EXPL/LLT-LIN/1104/2021
Period: 01-01-2022 to 31-12-2023 (extended date)
The aversive feelings that people experience in scenarios of crisis, such as fear, uncertainty, and the feeling of being out of control, stimulate adherence to conspiracies and promote their fast propagation [PROO17]. The COVID-19 pandemic, one of the most challenging global societal crises we are facing nowadays, has boosted the proliferation of conspiracy theories on social media to unprecedented levels. According to recent studies, denialism, i.e. the psychological predisposition to reject expert and authoritative information, the conspiracy thinking, i.e. tendency to view challenging societal events as the product of conspiracies, and partisanship are the the key ingredients behind COVID-19 conspiracy beliefs [USCI20]. While some conspiracy theories may have no huge impact on people’s lives, others, such as the ones related to medical and public health topics can be particularly dangerous for individuals and a menace to democratic societies. Therefore, research on misinformation, including conspiracy theories, is becoming increasingly relevant.
Motivation
Although there is a diversity of approaches on conspiracy theories from different fields, including psychology, political science, sociology, history, and media studies [DOUG19], research on the language mechanisms and strategies used in conspiracy narratives abounding in social media is still scarce. The advent of large social media platforms offers an unique opportunity to access an immensity of real-world data that will allow answering crucial research questions related to language use and communication dynamics. One critical question concerns the identification of the main conspiracy sociolects (i.e. the sociolinguistic characteristics shared by authors of conspiracy theories) in social media. Another important issue concerns the identification of the most important relationships between complementary modes of communication – particularly, text and image – co-occurring in the same conspiracy content.
Multimodal approaches, relying on the joint analysis of language patterns from multiple modalities, are critical to dissecting conspiracy theories, key to the development of robust and reliable conspiracy detection systems. This exploratory project aims at developing an innovative multimodal (text-image) conspiracy grammar for Portuguese, based on a multidisciplinary approach, crossing knowledge and methods from Corpus Linguistics, Forensic Linguistics, Discourse Analysis, Social Semiotics, and Natural Language Processing (NLP).
Research Objectives
The research will be based on a large and representative multimodal corpus composed of text-image conspiracy content posted by Portuguese speaking users in a variety of social media platforms and blogs. This corpus will be finely-grained annotated with syntactic-semantic, pragmatic-discursive, and social semiotic attributes. These will allow performing complex and refined queries on the corpus, in order to identify relevant multimodal patterns, which will be thus systematically described in our conspiracy grammar. The challenges in the grammar creation involve, among other aspects, the integration and articulation of objective and subjective knowledge from different sources, and the dynamic description of a huge set of relationships involving all the language dimensions considered in this study.
This innovative resource will contribute both to the systematic characterization of multimodal conspiracy narratives in social media, and will help to understand how linguistics, pragmatics and social semiotics interact to create new and engaging conspiracy narratives. Furthermore, it will support the development of methods for conspiracy detection, and machine-assisted conspiracy sociolects identification.
We believe that this exploratory project will contribute to advancing the state-of-the-art in multimodal misinformation studies, answering, among others, the following research questions:
- Which are the most predominant morphosyntactic, lexico-syntactic, semantic and discursive features in conspiracy theory narratives that abound in social media, and how do they relate to each other?
- Which meanings can be inferred from images that include conspiracy narratives, from a social semiotics point of view?
- How do text and image articulate in multimodal conspiracy narratives to create either a unified or a dissociated meaning?
- Which are the most suitable approaches to describe and formalize the multimodal properties from text and image, and respective interaction, in view of their automatic processing?
- How can Portuguese conspiracy sociolects be characterized from a multimodal point of view?
We have formed a multidisciplinary research team including researchers from corpus linguistics, forensic linguistics, discourse studies, social semiotics and natural language processing, critical to successfully achieve the project’s goals.