Do They Speak the Same Language? Exploring Differences in Linguistic Styles between Supporters of #Buildthewall and #Nobannowall in Social Media

Background and Purpose: In an increasingly online society, social networking sites such as Twitter have largely changed how different parts of society communicate and engage in addressing policy issues. While bringing public attention to bear on issues of the moment is a necessary step for promoting policy change, little is known about how supporters of a specific public policy express themselves in social media. Therefore, this study aims to explore the differences in linguistic styles of supporters and opponents of building a US/Mexico border wall on Twitter.

Method: We identified 19,342 Twitter users who posted tweets between January and March in 2017 that contained the “#buildthewall” or “#nobannowall” hashtags. We collected the historical tweets of each user (approximately 40 million). From these, a random sample of 200 tweets that used #buildthewall and 200 that used #nobannowall.  Two independent coders labeled the tweets as indicating support (80% agreement) or opposition 84 agreement) to the boarder wall. For the study, we used only unanimously-labeled tweets. The resulting sample includes supporters (57 users, 222,621 tweets) and opponents (73 users, 218,685 tweets) of the border wall.

We used the LIWC (Linguistic Inquiry and Word Count) to measure psychological processes (e.g., group identity, anger and anxiety) and moral foundations (e.g., fairness, purity and authority) latent in tweets. We also defined keyword patterns for references to immigrants; three neutral (“immigrant*”, “latino*”, “muslim*”), and one negative (“illegals”). We applied these measures to aggregate tweets to each user in the sample. Independent t-tests were used to compare differences in these categories between supporters (M_s) and opponents (M_o).

Results: On average, supporters and opponents of building a US/Mexico border wall posted a similar number of tweets [M_s = 1,589; M_o =1,514]. Supporters of the wall emphasized more on authority [M_o =20, M_s = 42; t (76.92)= -4.48, p =.00] and purity [M_o = 17, M_s = 29; t(76.31)= -2.72, p =.01], while opponents of the wall emphasized more on fairness [M_o = 70, M_s = 34; t(76.42)= 2.68, p =.01] and care [M_o = 135, M_s = 98; t(110.24)= 2.75, p =.01]. Supporters of the wall were more likely to use “illegals” to refer to immigrants [M_o = 0.17, M_s = 8; t(53.32)= -5.03, p =.00], while opponents of the wall were more likely to refer to immigrants as “immigrant*” [M_o = 26, M_s = 7; t(64.52)= 2.91, p =.00] or “latino*” [M_o = 4, M_s = 24; t(55.36)= -5.71, p =.00]. We did not find any significant differences among supports and opponents of the wall in the emotional content of their tweets.

Conclusions and Implications: Our preliminary result is that supporters and opponents of building a US/Mexico border wall emphasize different moral foundations in their postings. Supporters were also more likely to use what many regard as a stigmatizing label “illegals” to refer to immigrants. A reliable and repeatable method to identify such differences may help practitioners and policy advocates both target and tailor social-media interventions to different groups.