Framing the Covid-19 Pandemic Crisis in Financial Discourse. A Sentence Embeddings Approach

This study aims at identifying and interpreting recurrent frames in corporate communication, and in particular in CEO Letters to Shareholders, sent in 2020, the first year of the COVID-19 pandemic crisis. The identification and analysis of discursive frames can shed light on the discourse strategies deployed by companies as a function of their performance, and in particular on how in their Letters CEOs accounted for unexpected positive or negative financial performance due to force majeure contextual events, i.e., those related to the pandemic, through the use of linguistic resources designed to convey specific interpretations and evaluations of situations and issues. To accomplish this, the study proposes a novel methodology of approaching text by implementing computer assisted text analysis and state-of-the-art natural language processing techniques. The research has been carried out from a quantitative point of view experimenting with state-of-the-art neural networks models in generating frames from textual corpora. Key-words: natural language processing, topic models, sentence embedding, framing in financial discourse, CEO letters, strategic communication.