Is the Great Awokening Really Winding Down? Part I: Some Multifaceted Evidence from Twitter Content

There has been some discussion lately by Eric Kaufmann, Tyler Cowen, Balaji Srinivasan, Paul Graham and Musa Al Gharbi as to whether The Great Awokening is winding down.

I’m going to write a series of blog entries about this topic to contribute to the discussion. In order to do that, I will be looking at data on language usage by several relevant institutions such as news media, the US Congress, social media and the Academy.

I begin this series with the following analysis of Twitter content.

Mentions in Twitter of terms that signify prejudice (i.e. racism, sexism, homophobia, etc) are passed previous peaks. Despite the recent drop in prevalence, mentions of prejudice in Twitter 2022 are still much more abundant than as recently as 2010 (see percentage increases from 2010 to 2022 in the figure below). The percentage increases from the 2010 baseline to the peak of each trend are striking.

Twitter chatter about different prejudice types has come in waves which peaked at different points in time. Mentions of gender prejudice peaked first in 2018, mentions of sexual orientation prejudice, antisemitism and islamophobia peaked next in 2019, mentions of ethnic/racial and nation of origin prejudice followed in 2020. Mentions of gender identity prejudice peaked in 2021.

In contrast to the dynamics displayed by prejudice signifying words, mentions of terms often associated with social justice discourse and with mostly positive connotations display a different and more multifaceted pattern, with numerous terms still growing and others mostly stabilizing around their record highs.

I next plot another set of terms associated with social justice discourse, but similarly to the prejudice signifying terms in the first plot above, these terms mostly have largely negative connotations and are often suggestive of bad/deviant behavior or dysfunctional social processes. The patterns here are similar to those of the prejudice signifying words, with many terms, but not all, appearing to be past their previous peaks.

Interestingly, terms that focus on victimization have also grow dramatically in prevalence in Twitter within the span of just 10 years. Notice that despite these terms having negative connotations, they reflect a very different type of negativity than the terms above by focusing on the receiving end of the victimization rather than the negative behavior of the victimizer.

The trends in the figure below provide empirical support for the extraordinary work of Bradley Campbell and Jason Manning who first predicted this trend as early as 2014 in their work on victimhood culture. The patterns displayed by the set of terms below are also heterogeneous, with some terms appearing to be past their peaks and others still growing or maintaining their previous highs.

There are a bunch of different patterns in the charts above. In order to visualize the aggregated underlying dynamics for the four subsets of terms I have analyzed (prejudice signifying words, negative terms associated with social justice discourse and suggestive of bad behavior/processes, terms focusing on the receiving end of victimization and positive terms associated with social justice), I will average the min-max scaled frequencies time series of all the terms within each subset. This is to obtain an overall sense of the prevalence of each construct of terms while avoiding that any term in the construct with an outsized frequency magnitude dominates the aggregated average. This approach produces similar dynamics to factor analysis but it is more transparent and does not need arbitrary hyper-parameter tuning like factor analysis does for eigenvalue cut-offs, rotation method to use, number of factors to select, thresholds for inclusion of a word in a factor, etc. The resulting averaged min-max scaled frequencies metrics allows discerning overall minimum, medium and maximum usage of the terms in each construct.

The chart above leads me to formulate the following preliminary and tentative hypothesis still to be confirmed/disconfirmed by future analysis. The patterns above suggest a decoupling between negative and positive terminology associated with The Great Awokening.

Terms with overwhelmingly negative associations that focus on bad behavior by victimizers such as racism, misogyny, homophobia, micro-aggressions, discrimination, police brutality, injustice or hate speech (see blue and orange trends above) are no longer growing in prevalence in Twitter content. They instead appear to have decreased slightly in prominence from their previous peaks around 2018-2020.

In contrast, terms often associated with positive social justice discourse such as diversity, equity, inclusion, sensitivity or safety (see green trend above) or that focus on the receiving end of victimization: abused, marginalized, wronged, harmed, silenced, or disadvantaged (see red trend above) are either still growing or have stabilized around their previous peaks.

There is another factor worth considering. The beginning of the COVID global pandemic in 2020 surely captured a substantial fraction of global attention. Trends interrupted by the pandemic grip on public consciousness might yet well resume as the pandemic recedes.

To conclude, social justice discourse terminology with positive connotations (inclusion, equality, safety, etc) is not decreasing in prevalence on Twitter. Similarly, terms that highlight the receiving end of victimization (traumatized, marginalized, wronged, exploited, etc.) are also not decreasing in Twitter chatter. In contrast, negative social justice discourse terminology mostly suggestive of bad behavior/processes (racism, misogyny, homophobia, toxic masculinity, police brutality, discrimination, injustice, etc) have decreased moderately from their previous peaks around 2019/2020.

Importantly, two of the four sets of terms analyzed above grew slightly from 2021 to 2022. This suggests that the jury is still out in terms of whether the Great Awokening is winding down. The possibility that the phenomenon might be mutating by emphasizing more social justice terminology with positive connotations and those at the receiving end of alleged victimization while toning down its more negative/corrosive/aggressive terminology deserves further consideration.

The analysis I have presented above is itself preliminary and in need of being complemented by additional analysis of other institutions’ content, which I hope I will be able to report about soon.