Have you ever been cautioned against talking about politics during family gatherings? The commonly held idea that political discussion can lead to strife between participants is something that can surface frequently, especially around the holiday season. However, is this statement true? Does the discussion of political events and figures actually create an identifiable increase in tension? In this article, I will be dissecting the language used in public forums in regards to political events. How many people are engaging in this type of discussion online? What are the current demographics of those involved in these discussions? Will numbers change at all as a major presidential election draws near, and thus, the perceived risk increases? I hope to reveal answers to these questions throughout this analysis. 

The validity of using social media in this type of research is something to be discussed primarily. Though the subject matter at hand is very different, I found the article “Using social media to measure language use” by Anastasia Zhuravleva et al to be supportive of my convictions in utilizing social media as a resource in order to document cultural movements and norms in regards to political profanity in public spaces. The authors discuss how social media, specifically Facebook, accurately reflects traditional questionnaire data and allows researchers to study social interactions and contact rather than just the stagnant use of language via said questionnaires. 

Similarly, the article “Personality, Gender, and Age in the Language of Social Media: The Open-Vocabulary Approach” by Andrew H Schwartz et al supports the findings of the article above. In this article, the authors analyzed the Facebook messages of 75,000 volunteers in regards to their language use. What they found was that social media sites such as Facebook can be useful tools in order to gather data about language use. The social connection between users is highlighted again as a reason for why it yields more accurate results than other, traditional methods of data collection, thus supporting my use of social media for data collection again. Now that the issue of the validity of social media as a research space, there needs to be discussion on profanity itself and how it spreads online and who is utilizing it.

Since social media is a key component of my research, how does the use of language spread online? Has there been studies done on how that impacts profanity usage? Does the usage of profanity in others’ language, and in particular political profanity, encourage an increase of usage in those who are exposed to it? In the article “Is offensive commenting contagious online? Examining public vs interpersonal swearing in response to Donald Trump’s YouTube campaign videos”, K Hazel Kwon and Anatoliy Gruzd explore that very question. They break comments down into ‘parent’ and ‘child’ comments and conclude that offensive comments online are contagious and shape the community’s interactions socially. This is similar to what I am doing, but on a different social media platform.

What about gender and swearing? Swearing has traditionally been denoted as a more masculine use of language. “Gender and Swearing: A Community of Practice” by Karyn Stapleton looks at how the framework of communities of practice impact the use of swearing in a group of undergraduate drinking friends. In her research, Stapleton finds that though both men and women in this like-minded group are comfortable using profanity in their language with each other, there are persistent gender differences. Women use their profanity in nuanced ways in order to construct a subtle gendered identity via their linguistic practices. This will be something that I paid attention to when collecting and compiling my data. Since women use profanity in very specific ways to create identities, will that impact my results at all? 

This research is invaluable for several reasons. Firstly, there is a lack of literature focused on this topic in academia. There are many instances of literature studying portions of this discussion, but none viewing profane language used by the general populace in regards to political matters on social media as a whole. I found one peer-reviewed article that did discuss something very similar to this topic, but the authors were investigating language and its effects in South Korean commenters, not those residing in the United States. The article, “Swearing Effects on Citizen-to-Citizen Commenting Online: A Large-Scale Exploration of Political Versus Nonpolitical Online News Sites” by K. Hazel Kwon and Daegon Cho did ultimately find that the use of profanity online in regards to political discussions incites attention from the audience and that there is an acceptable degree of profanity use online. However, this study is focused on South Korean commenters, not Americans. This lack of literature presently is impactful on the study of sociolinguistics; this is an area of communication which impacts a large amount of people in our current society. 

Secondly, political matters seem to be at the forefront of our daily lives in the United States, especially with a new presidential election on the horizon. One is bombarded with discussions on debates, political ads, and stories broadcasted via a multitude of venues during their day-to-day routines. In light of this, it stands to reason that discussions on political happenings happen frequently enough to be impactful on the speech patterns of adults in the United States.  Tracking the usage trends of political profanity can potentially demonstrate the perceived risk felt by the populace at large. 

Data Sources & Methodology 

Firstly, I must define the language I am studying. In order to accurately gather data to tabulate, my working definition of political profanity and slurs must be clear. According to the Oxford English Dictionary, profanity is defined as “… the fact, quality, or condition of being profane; profane conduct or speech; (also) an instance of this, a profane or obscene act or word…” and the definition of a slur is “… a deliberate slight; an expression or suggestion of disparagement or reproof.”. For my research, I am specifically looking for language that:

1. Is in reference to another person. It may be directed at an individual or a group, but it must always be in reference to humanity. I did not take examples that were in reference to inanimate objects, animals, or used in speech that wasn’t directed at a specific subject.
2. Is derogative and/or profane in nature to an excess. I did not take examples that were ‘lightly’ profane; I wanted to focus on language that wouldn’t normally be observed in common situations. For these reasons, I did not collect data from comments that contained language like ‘idiot’, ‘dumb’, ‘troll’, or ‘stupid’. 
3. May or may not contain expletives. Even though many of my insult tokens turned out to contain expletives, I did not want to rule out language that did not contain them yet still met the qualifications for a slur or profanity.

I was particularly interested in the language used in public forums online. The anonymity that social media can afford a person – and thus, a ‘freer’ venue in which to communicate thoughts and beliefs – was something I wanted to focus on. I decided that looking at the comment section of various articles from different news sources would give me a diverse pool from which to pull data. To keep with the trend of diversity, I decided to choose three local news sources (Anchorage Daily News, KTUU Channel 2 News, and KTVA 11 News) and three national news sources (CNN, Fox News, and MSNBC). All of these news sources vary slightly on political leaning; CNN skews slightly left whereas MSNBC is very liberal and Fox News is very conservative. This broadens the potential data pool compared to just looking at one political leaning or even just a moderate news source. 

I then pulled data from two different forms of social media, Twitter and Facebook. Twitter and Facebook both have different demographics; as such one would expect the distribution of comments to correlate with those demographics. Facebook currently is utilized by 69% of U.S. adults; 63% of men use the online platform and 75% of women. (Pew Research 2019) Twitter currently is utilized by 22% of U.S. adults; 24% of men use the platform and 21% of women. (Pew Research 2019) They are both used extensively throughout the United States by a variety of age groups as well, which allowed for a more varied potential data pool. I created a separate quasi-anonymous social media account on both Facebook and Twitter in order to collect my data and interact with willing participants. 

Once I narrowed my venues for data collection, I had to collect data points. I chose to cover a random national political event weekly, usually one that encouraged a significant amount of discussion online. I then would go through the comments attached to the various articles, looking for instances in which language was used that fit my definition of political profanity. Comments slowed down around 1-3 weeks after the article’s inception, so I gathered the bulk of the comments during that time frame. I then recorded each insult token on a spreadsheet. Firstly, I indicated where I found the insult token; I recorded the social media site in which it was found and then what specific news source produced the article. Then, I recorded the comment itself in its entirety, the specific profanity referenced as an insult token, and who the comment was directed towards. I recorded the commenter’s name and then their proffered gender and political leaning if they were publicly available. For organization, I color-coded the spreadsheet to designate the different articles in which the comment and profanity was found.

In order to maintain some semblance of consistency from my data pool, I chose articles that covered the same event by the different news agencies. The first event that I chose to pull data sets from was the coverage of a poll denoting Elizabeth Warren’s lead on Joe Biden in late September. The second event that I chose to pull data sets from was the coverage of Nancy Pelosi’s announcement of the impeachment investigation for President Donald Trump. The third event that I chose to pull data sets from was the coverage of Mike Pompeo’s failure to meet deadlines for documents in the impeachment investigation. As a control, I used the same methods listed above but chose a non-political article instead. I chose to look at the latest developments on the 737-Max investigation; however, not all of the news agencies covered the story, so instead of one specific article I had to pull data from several different ones. 

In looking at the data, there are several different speech communities and communities of practice involved with the set. At a macro level, we are dealing with American politics. We are also looking solely at users of social media that have public comments available. We are also looking at communities that affiliate themselves with these specific news sources.

Results

For my control, I looked at recent articles that covered new developments on the issue of the 737-Max investigations. Out of 606 comments, I only found one insult token. That insult token was found on Facebook under Fox’s coverage of the story. It was directed at President Trump and was said by someone who identifies as a male. 

For my first news event choice, I looked at articles that covered Elizabeth Warren’s slight lead over Joe Biden in the 2020 Presidential race that occurred in late September of 2019.  [1] [2] [3] [4] [5]  Out of a total of 2,481 comments I found 52 instances of political profanity being used within a comment. This amounts to 2.09% of all comments for the articles I pulled data from. For these data sets, Fox had the highest cumulative percentage of profanity usage on both Facebook and Twitter at 3.24%, followed by MSNBC at 1.67% and CNN at 0.84%. Twitter had a slightly higher usage of profanity than Facebook at 2.41% compared to 2.08%.

For my second news event choice, I looked at articles that covered Speaker of the House Nancy Pelosi’s announcement of the impeachment inquiry against President Trump that occurred in late September of 2019. [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] Out of a total of 2,669 comments I found 35 instances of political profanity being used within a comment. This amounts to 1.31% of all comments for the articles I pulled data from. For these data sets, Fox had the highest cumulative percentage of profanity usage on both Facebook and Twitter at 2.95%, followed by MSNBC at 2.35%, KTVA at 2.00%, CNN at 0.97%, Anchorage Daily News at 0.77%, and KTUU at 0.46%. Twitter had a higher usage of profanity than Facebook at 2.31% compared to 1.31%.

For my third and final news event choice, I looked at articles that covered Secretary of State Mike Pompeo’s failure to reach the House’s subpoena deadline that occurred in early October of 2019. [17] [18] [19] Out of a total of 1,943 comments I found 29 instances of political profanity being used within a comment. This amounts to 1.49% of all comments for the articles I pulled data from. For these data sets, Fox had the highest cumulative percentage of profanity usage on both Facebook and Twitter at 3.09%, followed by CNN at 0.81%. Facebook had a slightly higher usage of profanity than Twitter at 1.54% compared to 1.17%.

Gender

The only instance of a female majority for article set #1 was on Facebook under the article by CNN. The only instance of an unspecified gender majority for article set #1 was on Twitter under the article by CNN. In reference to the content of comments, there was a stark difference in gender as well. Out of the 52 total insult tokens, 11 insult tokens were in reference to males, 20 insult tokens were in reference to females, and 21 insult tokens were in reference to things that were unisex or not specified. As percentages, that is 21.2% in reference to males, 38.5% in reference to females, and 40.4% in reference to things that were unisex or not specified. 

The only instance of a female majority for article set #2 was on Facebook under the article by Anchorage Daily News. Two instances of an unspecified gender majority for article set #2 were both found on Twitter under the articles by CNN and MSNBC. In reference to the content of comments, out of the 35 total insult tokens, 6 insult tokens were in reference to males, 7 insult were in reference to females, and 22 insult were in reference to things that were unisex or not specified. As percentages, that is 17.14% in reference to males, 20.00% in reference to females, and 62.86% in reference to things that were unisex or not specified.

There were no instances of a female majority for data set #3. One instance of an unspecified gender majority for article set #3 was both found on Twitter under the article by CNN. In reference to the content of comments, out of the 29 total insult tokens, 6 insult tokens were in reference to males, 2 insult tokens were in reference to females, and 21 insult tokens were in reference to things that were unisex or not specified. As percentages, that is 20.69% in reference to males, 6.90% in reference to females, and 72.41% in reference to things that were unisex or not specified.

Overall, out of 116 insult tokens 26 points were generated by those who identify as female. That is a percentage of 22.41%. 75 insult tokens were generated by those who identify as male. That is a percentage of 64.65%. Finally, 15 insult tokens were generated by those who did not identify their gender or used gender neutral pronouns. That is a percentage of 12.93%.

Political Leaning

Another point of interest for me was the topic of political leaning. How often were different political parties talked about? When looking in black-and-white terms of Republicans versus Democrats, it feels like both parties accuse the other of consistently attacking their opponent; there is an assumed war of words occurring and each side seems to think that they are blameless in it. So, I tabulated all of the instances in which political parties or members of certain parties were mentioned. 

In the first data set, I found that 46 of the 52 insult tokens I collected dealt with political leaning. Of these comments, they referred to Democrats or Liberals 78.26% of the time. In contrast, Republicans or Conservatives were only referred to 21.74% of the time. 

In the second data set, I found that 27 of the 35 insult tokens I collected dealt with political leaning. Of these comments, they referred to Democrats or Liberals 85.18% of the time. As with article set #1, Republicans or Conservatives were only referred to 14.81% of the time. 

In the third data set, I found that 22 of the 29 insult tokens I collected dealt with political leaning. Of these comments, they referred to Democrats or Liberals only 40.91%. In contrast to the two other data sets, Republicans or Conservative were referred to 59.10% of the time. 

Self-Bowdlerization

An interesting phenomenon that I came across was the tendency for some commenters to practice self-bowdlerization in regards to their use of political profanity. Of the 116 insult tokens I collected, 10 displayed this self-bowdlerization. The fact that 8.62% of comments tended to self-censor their comments online is significant. Out of these instances of self-bowdlerization, 5 of these censorships came from those who identified as male, 4 came from those who identified as female, and 1 came from a person who did not note their gender. 9 of these insult tokens were found on Facebook with 1 on Twitter. Cumulatively, the 5 instances of self-bowdlerization from males came from a pool of 75 insult tokens, bringing the male percentage of bowdlerization to 6.67%. The 4 instances of self-bowdlerization from females came from a pool of 26 comments, bringing the female percentage of bowdlerization to 15.38%. The 1 instance of self-bowdlerization from the person who did not denote gender came from a pool of 15 comments, bringing the gender-neutral percentage of bowdlerization to 6.67%.  

Discussion 

In light of the extreme lack of political profanity found in my control, based on the data I collected it seems that political issues do tend to incite people to utilize profanity. Fox News is the news source that tends to have a more frequent percentage of political profanity usage compared to the other news sources studied. There also was a marked difference in the engagement of the audience between political and non-political stories. 

Early on in data collection, I realized that there was a disparity in the amount of comments put forth by different genders. When I pulled insult tokens from comments, I made sure to notate the designated gender set forth by each individual. A majority of comments were written by those who self-identified as male, both on Facebook and on Twitter. In all three data sets, men commented more frequently than women or those who did not specify their gender or used gender-neutral pronouns. In addition to the authors of the comments, the content of the comments were largely directed at females if the comment was not directed at a group or something that is not a person. Then, the article referencing Elizabeth Warren had the highest percentage of overall political profanity usage. This is concerning to me. There is a clear difference between how men discuss and are discussed versus how women discuss and are discussed. There is a line between sex here, and usually it is those that identify as female who are put in a more negative light. This suggests that there are cultural biases towards male speakers in public areas.

The comments on political leaning were interesting to tabulate. Largely, Democrats were talked about negatively, save the data set on article #3. Could this be because of the current President and makeup of the House and Senate? America is under a Republican majority currently, and I wonder if this frees citizens to talk more negatively about the opposing party and party members because of it. Overall, however, both parties are guilty of utilizing political profanity to attack members of the party opposite of themselves. There was interaction from both sides, varying in intensity. 

When there is censorship, there indicates a certain level of unease with whatever is being censored. The results of the self-bowdlerization found in my data set surprised me; the number was much higher than I anticipated, and the numerical majority of those using it were male. Traditionally, swearing was something done by those who identified as male; the use of profanity was something that violated the cultural norms for women, and thus, it was highly taboo for women to utilize it as mentioned by Stapleton in “Gender and Swearing: A Community Practice”. (Stapleton 2003) Researchers are now determining that the community of practice in which people are engaging in has more of an impact on the usage of profanity rather than gender. However, cultural and societal views on the matter are still more in line with the traditional gender-separated approach previously held; this makes the data collected on self-bowdlerization more significant in nature. 

Though the data initially suggests a more frequent rate of usage for those who are male, when looked at in comparison to the total number of comments made by the genders included in this study, the female use of self-bowdlerization is 8.71 percentage points higher. This falls into the commonly held belief that women are less comfortable with swearing or profanity in public spaces; based on these results, I can theorize that the cultures that these women are associated with still are in line with the belief that women should not use profane language. 

As with any study like this, there are limitations associated with it. Firstly, there are other social media sites where conversations like these take place that I have not tabulated. There are also discussions in person; just focusing on online interactions alienates a large portion of people who do not have access to social media or choose to not have an online presence. Another limitation is the anonymity that social media provides (to an extent). There is no way to verify someone’s identity; there could be people creating different accounts to engage online. One can also delete comments made on social media; thus, the data is not permanent and can potentially change if re-visited in the future. 

Another limitation to this study is the fact that not all the news agencies picked up on the stories I chose. Some stories were not covered by all the news sources, so there are gaps in the data in regards to those events. Not all of the news agencies utilized both Facebook and Twitter, as well; Fox is an example of this. Fox does not use Twitter as a form of protest, so there are no data points from them on that social media site at all. 

This study encompasses a relatively brief moment of history - just over a month - and the potential for future research is near limitless. One could expand the potential data pool by incorporating more social media sites and news sources into the study; this move could reveal more about the demographics that engage in these discussions. It could also be beneficial to narrow the location of the data pool. I focused on the United States in general, but there could be studies done on hyper-local political events in order to see if there are differences in the political profanity used in different regions of the United States compared to others.  Future studies could also focus on topics other than political events and maneuvering; do percentages of profanity usage directed to other human beings change when topics change? However, the most obvious future direction for study is simply the expansion of time. I will be continuing data collection until November 2020 and the election of a new president. I will then look at the corpora to determine if there is a quantifiable difference in the use of political profanity in public spaces as the election draws nearer to a close.

K, H. K., & Gruzd, A. (2017). Is offensive commenting contagious online? examining public vs interpersonal swearing in response to Donald Trump’s YouTube campaign videos. Internet Research, 27(4), 991-1010. doi:http://dx.doi.org.proxy.consortiumlibrary.org/10.1108/IntR-02-2017-0072

Kwon, K. H., & Cho, D. (2017). Swearing Effects on Citizen-to-Citizen Commenting Online: A Large-Scale Exploration of Political Versus Nonpolitical Online News Sites. Social Science Computer Review, 35(1), 84–102. https://doi.org/10.1177/0894439315602664

Perrin, A., & Anderson, M. (2019, April 10). Share of U.S. adults using social media, including Facebook, is mostly unchanged since 2018. Retrieved from https://www.pewresearch.org/fact-tank/2019/04/10/share-of-u-s-adults-using-social-media-including-facebook-is-mostly-unchanged-since-2018/.

Schwartz, H. A., Eichstaedt, J. C., Kern, M. L., Dziurzynski, L., Ramones, S. M., Agrawal, M., . . . Ungar, L. H. (2013). Personality, gender, and age in the language of social media: The open-vocabulary approach. PLoS One, 8(9). doi:http://dx.doi.org.proxy.consortiumlibrary.org/10.1371/journal.pone.007379

Stapleton, K. (2003). Gender and swearing: a community practice *. Women and Language, 26(2), 22+. Retrieved from https://link-gale-com.proxy.consortiumlibrary.org/apps/doc/A112359435/LitRC?u=anch19713&sid=LitRC&xid=d6a9e6fa

Zhuravleva, Anastasia, de Bot, Kees & Hilton, Nanna Huag (2016). Using social media to measure language use. Journal of Multilingual and Multicultural Development, 37:6, 601-614. DOI: 10.1080/01434632.2015.1111894
__________________________________
KAITLYN WOLFE received a Baccalaureate degree in English in 2019. She is especially interested in the relationships between social interactions and language use. This piece was selected by Professor David Bowie.