Research methodology

The PATDISC project empirically studies and analyses discursive data extracted from Twitter, Instagram and TikTok. The core innovation of this approach lies in evaluating platforms where verbal discourse either dominates, is closely combined with multimodal elements or where multimodal stimuli take primary importance. The project focuses exclusively on public discourse, the free availability of which in cyberspace allows for rigorous academic analysis.

To achieve its goals, PATDISC implements an integrative methodology that combines the frameworks of corpus linguistics, discourse analysis and pragmatics. This process is divided into three main operational phases:

1. Data collection and corpus compilation

Data collection targets digital discourse produced directly by social media users, capturing the various ways they share content (text, images, emojis, links, videos, etc.) to ensure a truly multimodal analysis.

  • Parameters: Data is harvested based on specific parameters such as keywords, themes or comment threads.
  • Software tools: The team utilises specialised software.
  • Structure: The final corpus will encompass distinct sub-corpora or datasets organised by the underlying theme that motivates the pathologisation (e.g., age, nationality or gender identity, ideology or beliefs).

Once compiled, the corpus undergoes a rigorous two-tier analytical process:

  • Macro-analysis: The collected data is segmented into «discursive moves» (textual fragments where a concrete action is perceived). The team then identifies the specific actions performed in each move using an ad hoc taxonomy of speech acts built upon classic linguistic classifications, followed by a close examination of how these moves are sequenced.
  • Micro-analysis (text mining): Utilising text-mining software, the team analyses co-occurrences and combinations to identify recurring pathologising strategies. This phase also involves extracting key terms and identifying neologisms to pinpoint platform-specific linguistic features.

Qualitative analysis is deployed to clarify how pathologising processes are driven by specific ideologies that characterise agents, foster distinct social relationships and construct digital realities by either including or excluding subjects.

The research maintains a balanced approach: it is data-based (supported by the data through complemental theoretical models) while also remaining data-driven (guided by the data to potentially generate new theoretical models). The analysis relies on the following interconnected linguistic frameworks:

  • Verbal (im)politeness models: Drawing on classic and modern theories of (im)politeness, relational work and the management of interpersonal relations.
  • Language aggression and conflict: Incorporating specialised research on mockery, online public shaming, insults, disagreements, complaints, cursing and interactive conflict.
  • Multimodal perspective: Attending carefully to multimedia elements such as memes and emojis to understand how visual stimuli alter or enhance the verbal text.

Ethical note on data management

In alignment with the ethical standards of the Association of Internet Researchers, all collected data is strictly anonymised to eliminate any personal identifying information before analysis or storage in open-access institutional repositories.