Words and Action: Modeling Linguistic Leadership in #BlackLivesMatter Communities

TL;DR

This work investigates how linguistic leadership emerges within #BlackLivesMatter online communities by integrating time-binned community detection on Twitter with temporal word embeddings to detect semantic change. It then induces a leadership network that identifies which communities introduce new word meanings and which adopt them, revealing that BLM activists and Black celebrities are primary drivers and that conservatives often follow or co-opt emerging terms from the start. The methodology combines seed-based clustering, cross-temporal matching, and significance-tested leadership edges, applied to a large-scale dataset spanning 2014–2020. The findings offer quantitative evidence for how meme- and word-level discourse shifts propagate through online publics and influence mainstream political conversations, while also highlighting data gaps and the ongoing evolution of platform dynamics.

Abstract

In this project, we describe a method of modeling semantic leadership across a set of communities associated with the #BlackLivesMatter movement, which has been informed by qualitative research on the structure of social media and Black Twitter in particular. We describe our bespoke approaches to time-binning, community clustering, and connecting communities over time, as well as our adaptation of state-of-the-art approaches to semantic change detection and semantic leadership induction. We find substantial evidence of the leadership role of BLM activists and progressives, as well as Black celebrities. We also find evidence of the sustained engagement of the conservative community with this discourse, suggesting an alternative explanation for how we arrived at the present moment, in which "anti-woke" and "anti-CRT" bills are being enacted nationwide.

Figures (7)

• Figure 1: Semantic leadership in #BlackLivesMatter: Nodes represent the most central communities of the BLM network. The thickness of the edges are proportional to the number of word changes shared by each pair.
• Figure 2: Volume of tweets per month: We visualize the temporal distribution of tweets in our dataset in the style of Freelon et al (freelon_beyond_2016). The seven temporal partitions are indicated here both by color and a reference label (e.g., Period 1). Some pivotal events to the BLM movement in each partition are annotated.
• Figure 3: Tweet counts over time after filtering, by community. Grouping subcommunities allows for more consistent representation of each community over the entire span.
• Figure 4: Community size: Count of unique users in each community. Bar segments represent subcommunity clusters with height proportional to cluster size.
• Figure 5: Linguistic leadership: The linguistic leadership of key communities including (a) activists, (b) Black celebrities, (c) conservatives, and (d) progressives with their contributions highlighted. The height of each bar is proportional to the overall number of words a community leads (left) or follows (right). The height of the connecting band is proportional to the number of words the community on the left leads the one on right.