A decade of discourse: Exploring sentiments and trends around immigration on social media from 2014 to 2024

Social media discussions contribute to the evolving public perception of refugees and immigrants. However, prior research often relied on a single platform and short-term analyses, offering a fragmented view of a highly dynamic phenomenon.

Examine trends in public narratives surrounding refugees and immigrants, including the evolution of sentiment and user engagement on Twitter, Facebook, and Bluesky.

We analyzed 6.3 million U.S.-based English-language posts from Twitter (2014-2023), Facebook (2014-2024), and Bluesky (2023-2024), using platform APIs. Posts containing one or more of 129 immigration-related keywords were grouped into 76 categories. Sentiment was classified using a supervised Support Vector Machine model, and engagement was aggregated at the keyword level. Twitter geodata enabled state-level sentiment mapping.

Peaks in volume and negativity aligned with major events, including the 2014 Syrian refugee crisis, the 2017 travel ban, and the 2018 family separation policy. From 2014 to 2019, negative sentiment increased on both Twitter and Facebook, then became more neutral in subsequent years. Bluesky began with predominantly neutral discourse in 2023 but grew more negative after its public launch. Refugee-related discourse was consistently less negative than immigrant-related discourse across all platforms, while enforcement-related and exclusionary rhetoric keywords emerged as the most negatively evaluated. Twitter geodata revealed widespread negativity across states, although refugee discourse remained more moderate or neutral than immigrant discourse overall.

Migration discourse is shaped by political events, emotional framing, and platform-specific dynamics, underscoring the need for cross-platform analyses to understand evolving digital narratives.