Social discourse and reopening after COVID-19

Abstract

Although the COVID-19 pandemic has not been quenched yet, many countries lifted nationwide lockdowns to restart their economies, with citizens discussing the facets of reopening over social media. Investigating these online messages can open a window into people’s minds, unveiling their overall perceptions, their fears and hopes about the reopening. This window is opened and explored here for Italy, the first European country to adopt and release lockdown, by extracting key ideas and emotions over time from 400k Italian tweets about #fase2 — the reopening. Cognitive networks highlighted dynamical patterns of positive emotional contagion and inequality denounce invisible to sentiment analysis, in addition to a behavioural tendency for users to retweet either joyous or fearful content. While trust, sadness and anger fluctuated around quarantine-related concepts over time, Italians perceived politics and the government with a polarised emotional perception, strongly dominated by trust but occasionally featuring also anger and fear.

Online search trends and word-related emotional response during COVID-19 lockdown in Italy: a cross-sectional online study

“…Feelings of fear and reduced sense of agency might be the source of similar results found on previous studies using semantic and emotional network analysis on social discourse in Italian tweets at the end of the first lockdown ( Stella, Restocchi & De Deyne, 2020 ; Stella, 2020 ). Stella (2020) showed that Italian participants tended to re-share a greater number of messages expressing fearful ideas, probably triggered by the strong affinity of the tweets’ content and the feeling of individuals following the sudden raises in the COVID-19 contagion curve after the reopening. Fear is also the emotional concept most frequently produced by Italian participants in relation to the COVID-19 concept in a word association task ( i.e., participants listed concepts coming in mind in response to a given concept) ( Mazzuca et al, 2021 ).…”

“…The results concerning the valence and dominance dimensions are consistent with the expected individuals' stronger feelings of fear and reduced sense of agency (and a consequent subjective perception of being in an out-of-control situation) in the current context, that is, an imminent threat to the humanity health (Stevenson, Mikel & James , 2007;Warriner, Kuperman & Brysbaert, 2013). Feelings of fear and reduced sense of agency might be the source of similar results found on previous studies using semantic and emotional network analysis on social discourse in Italian tweets at the end of the first lockdown (Stella, Restocchi & De Deyne, 2020;Stella, 2020). Stella (2020) showed that Italian participants tended to re-share a greater number of messages expressing fearful ideas, probably triggered by the strong affinity of the tweets' content and the feeling of individuals following the sudden raises in the COVID-19 contagion curve after the reopening.…”

Cognitive Network Science for Understanding Online Social Cognitions: A Brief Review

“…Boccaletti et al, 2014) can account for multiple types of links within one network representation. Besides TFMNs (Stella, , 2020b, as described above, other cognitive approaches with multilayer networks were recently introduced in the literature. Conceptual distance in multilayer networks (combining phonological, categorical, and semantic associations) was found to be predictive of both normative word acquisition in children (Stella et al, 2017) and picture naming failures in people with a spectrum of aphasic disorders .…”

“…Stella (2020b) built on previous results of semantic relatedness/closeness being captured by semantic network distance(Kenett et al 2017) in order to model semantic prominence in terms of closeness centrality, i.e. the mean network distance of one node to all its connected neighbours.…”

“…a cognitive tendency for the human mind to simplify the acquisition of low-quality new knowledge by adjusting it to a pre-existing set of beliefs and info (The main advantage of cognitive networks is their intrinsic ability to provide jargon and quantitative metrics that can be interpreted from a cognitive perspective. For instance, "semantic relatedness" can be translated, quantified and explained in terms of network distance(Kenett et al 2017) and this, in turn, gives rise to distance-based measures that can complement frequency and n-grams in detecting trending online topics(Stella 2020b). Bridging Big Data analytics with interpretable modelling through cognitive networks can greatly benefit several research lines studying cognition like semantic framing, stance detection and perception, emotional dynamics and cognitive biases, as outlined above.Cognitive networks can be powerful assets also for exploring the cognitions portrayed by AI-driven accounts like social bots(Ferrara et al 2016).…”