Loyalty, Purity, and Two Smoking Dumpsters – Moral Rhetoric in Occupy Wall Street Tweets Predicts Violence and Civil Disobedience in Collective Action Events

by Daniel Komáromy

Thesis supervisor: Dr. Frederic Hopp

Who hasn’t had images of police brutality against civilian demonstrators in recent years burned into their memory? The Yellow Vest movement, the George Floyd protests, the civil unrest in South Africa and in Belarus demonstrates that state repression spans across geographic locations and political systems. Whereas state violence may be an organic part of authoritarian regimes, in democracies they are often claimed to be an answer to the destruction caused by the non-normative violent acts committed by protestors. By contrast, normative nonviolent events, such as peaceful demonstrations, aren’t destructive, but they also don’t disrupt life of average citizens enough to capture public attention and bring about societal change. Therefore, a perennial dilemma in the life of social movements is how to set the agenda without committing any compromising acts. In other words: is there a chance to find a healthy balance between being constructive and disruptive?

Yes, there is. Furthermore, there are countless examples of this constructive disruption in nondemocratic countries as well. At the beginning of this century, in Serbia, Nepal, Lebanon, Madagascar, and Ukraine, the entrenched power was overthrown by means of non-normative nonviolent collective action. Or in its maiden name, civil disobedience. Civil disobedience incorporates a set of tactics that often violate social norms, but don’t involve any violence; such as occupation of public spaces. And since it blends constructive (nonviolent) and disruptive (non-normative) elements, it is more likely to attract mass participation, overthrow authoritarian regimes, and generally achieve its long-term goals than normative or violent action. Too good to be true?

A recent study investigated exactly this question. More specifically, it examined what sort of Tweets led to what sort of collective action events during the Occupy Wall Street (OWS) movement. And as fighting for social causes is rooted in moral convictions, to identify the building blocks of these messages, it used the Moral Foundation Theory (MFT) that describes five pairs of basic elements that represent the moral domain: care-harm, fairness-cheating, purity-degradation, loyalty-betrayal, and authority-subversion. The study combined the NAVCO 3.0, a database containing manually labelled news stories about collective action events, with dataset incorporating over 33,000 Tweets posted during the OWS events (1st August – 31st December, 2011) with the hashtags #ows or #occupy. Additionally, it used the Moral Foundations Twitter Corpus, which consists of Tweets labelled with the aforementioned 10 moral categories, to fine-tune a pretrained BERT architecture built for multi-label classification. The neural network first generated contextual embeddings for the OWS Tweets based on the uncased BERT base model. To avoid overfit, it was followed by a dropout layer with a rate of 0.2. Finally, a linear layer classified the Tweets into 11 categories. The fine-tuning with the MFTC corpus was carried out on this final layer . To map the relationship between the moral content and collective action type, time-series and non-parametric (partial) correlation analysis models were estimated.

The study has various implications regarding the moral rhetoric–collective action type relationship. First, whereas purity frames were negatively (and degradation frames positively) related to violent acts, they positively predicted nonviolent events. Second, topics about authority and subversion were related mostly to non-normative action; they were positively associated with civil disobedience, and negatively with normative action. Third, messages on loyalty and betrayal were positively related to violent and normative action and negatively to civil disobedience. Fourth, communication about harm increased the chances of a violent, and decreased the chances of a normative act in the following days. Finally, the main topic of the movement, fairness seemed to predict only civil disobedience.

  • As violence often ensues violence and many of these Tweets probably talk about activists being hurt, the relevance of the harm foundation is clear. The importance of loyalty is more surprising, but it actually resonates with literature on hostility prevailing among nativist and right-wing populist communities. Previous findings also demonstrate that positive discrimination towards one’s own group overrides fairness and harm concerns and justifies violence against others. In this sense, loyalty makes a virtue of evil.
  • As an ethical purist, Gandhi repeatedly emphasized purity and selflessness: he believed that removing anger, vanity, and dogmatic thinking constituted the efficacy of nonviolent action. In fact, communicating purity may be the key to persuade those citizens about the constructive nature of civil disobedience, who believe that these acts infringe the rights of average citizens.
  • Interestingly, Tweeting about subversion (sort of a synonym for disobedience) was inversely related to civil disobedience, which can be explained by slacktivism. Tweeting about subverting the system can give the false sense that you have already done something. Instead, you have to organize occupation and coordinate the (social) media presence of the movement to have successful actions. And this is what the authority messages were about. Although the General Assemblies of the movement lacked any form of hierarchy, they were astonishingly successful in tactical planning. Actually so successful that according to the critics of the movement, its “process fetishism (…) overshadowed its ideological thrust”. And by this they meant that, to assure a broad-based support, the OWS maintained a weak ideological core and rather focused on the successful actions. However, the discussion of the movement’s political strategy was beyond the scope of the study.

The take-home message for activists is that communicating harm and loyalty may incite violence – and in those games, the authorities usually come out on top. Instead of emphasizing the distinctiveness of the in-group or the evil done by the out-group, it is better to accentuate the purity of the movement’s means and the lack of hostility in its actions. These nonviolent acts also demand quite some coordination and organizing. Nevertheless, beyond the right communication, socio-political transformation requires also strategic politics based on institutions and communities centered on collective demands. If we are willing to learn from the history of past social movements, we must keep these in mind.

The coevolution of relative moral label frequency and collective action type over time.