#DearCongress was prevalent within Twitter discourse during the 2013 U.S. government shutdown. The hashtag allowed a community to form by assembling around a common topic. What a group communicates through social networks cannot be separated from its context. This micro-analysis of #DearCongress utilizes discourse tracing through word clouds, guided by Bruns & Burgess’ concept of an ad hoc public. By looking at the commentary during a political and time stamped event, we can better understand the role discourse plays in group formation, evolution, and impact.
In 1774, the First Continental Congress wrote a letter to King George to “address grievances.” United States citizens still contact their government officials, creating culture through ritualistic practices (Carey, 2009). The Pew Internet Research Center noted in 2013 that “39 percent of adults contacted a government official or spoke out in a public forum through off-line methods” and 34 percent conducted similar activities online (Smith, 2013). In the same year, a portion of Americans airing grievances to government officials very well may have tried to do so without hope of response. For 16 days in October 2013, the United States Government closed its doors and sent its employees home. It shut down. For 16 days, it is likely that no one replied to a letter, e-mail message, or phone call.
The platforms and transmission modes, which the American people utilize to discuss politics with or without their legislators, are continually in flux. The transmission mode has altered the transparency of these communications. Digitally-mediated political communication which is posted online is often in a public forum. In public forums, the path to group formation, organizing, or advocacy efforts eases. This exists in tandem with a level of performance, relative to context (Goffman, 1959). Individuals would not choose to contact their legislator or speak in a public forum if they did not want to receive some level of attention or raise awareness, resulting from their public efforts.
When the government shut its doors on 1 October 2013, Twitter was one social media platform that individuals turned to in order to share phatic communication — expressing and sharing commentary about Congress and the shutdown. Implementing a hashtag into a tweet formed what Axel Bruns and Jean Burgess (2011) refer to as an “ad hoc public” or a temporary community with shared interests. Hashtags’ nature and subject matter vary in use and expression. Despite the numerous discussions, hashtags, and groups, this paper presents just one from this time period: #DearCongress. This hashtag was chosen because of its connotation. At first glance, the hashtag #DearCongress brings to mind a letter — a public, open letter to Congress from those they govern. It should be noted that the 2013 government shutdown was not the first time this particular hashtag was utilized; tweets dating as far back as 2010 were found to use the tag.
This is a micro-level analysis of the Twitter hashtag #DearCongress, looking at the space and shared values within the texts produced during the government shutdown. To do so, this analysis utilized a process of discourse tracing (LeGreco and Tracy, 2009), through quantitative, visual tools to work through a “big data” set — 56,015 cases obtained over a 16–day period. Because it is pertinent to look at the content in relation to context to prevent altering the original meaning, the analysis is presented with a storyline, secured from the Associated Press and specific tweets. This hybrid presentation interrogates the ad hoc public’s evolution by remixing it with context.
Twitter is often described as a microblog or social networking platform. As a social network, Twitter allows its users to interact with one another. One can share content with their own group of followers or they can post comments, observations, and emotions.
Twitter is a construct, both technical and social. It serves as a “container for human action” . It utilizes the “@” reply as a symbol so that content can be directed toward a specific user, further connecting users in the virtual container. It also allows for users to connect with a larger, public audience as well through a sharing function, called a “retweet.”
Twitter differs from other social networks, such as Facebook, because all posts or tweets must be within 140 characters. The meaning or intent behind any singular tweet can be difficult to ascertain with such a limited space.
Additionally, Twitter serves as a quasi-public forum; posts are public and can be read by anyone. One can utilize Twitter’s search engine for a user or topic, cuing into a group or conversation. Conversations can be found or easily accessed by marking a word or phrase with a pound sign (#) or what Twitter users organically referred to as “hashtags.” One does not have to include a hashtag to appear within a search however; in some cases, including the same phrase, absent the “#” preface, will still place a tweet in Twitter’s search results. However, utilizing a hashtag within one’s tweet ensures the likelihood that a tweet and an individual is placed within a group or ad hoc public.
Ad hoc public
In their research of Twitter hashtags, Bruns and Burgess (2011) observed and collected tweets in moments of crisis, arguing that communicating within this “container” created an ad hoc public — a temporary group or community to share information. Their writing was applied to crisis communication during a natural disaster. The U.S. government shutdown in 2013 may differ from the immediacy of crisis and survival modes, yet it shares the time constructed nature and urgency with Bruns and Burgess’ notion of an ad hoc public.
It is through shared use, negotiation, and assemblage (Markham, 2013) around a hashtag itself that places one within a culture or community. Individuals may stay within this culture, or simply add to it and leave. The idea of an ad hoc public is necessarily applied to the Twitter conversation collected because of its underlying assumption — the groups “form, engage, and (potentially) dissolve again” . Within the ad hoc public, events can “cycle back inside and rearrange both the organization and the organism” .
Utilizing T.A.G.S., a Twitter Archiving Google Spreadsheet (Hawksey, 2013), we requested API (application programming interface) access from Twitter. This allowed the spreadsheet’s code to “scrub” Twitter’s database and textually archive the tweets. Upon the commencement of the government shutdown, we noticed one Twitter hashtag growing in frequency: #DearCongress. This hashtag was then added to T.A.G.S. as a collection limit.
The T.A.G.S. code that was used to scrape, store, and archive tweets has a limit of 16,000 tweets. Because of this, we had to create a new document each time the spreadsheet reached its numerical capacity (Within the first 36 hours, the original database reached its file limit). Tweets were collected for the time period ranging from 1 October through 16 October 2013 (the duration of the U.S. government shutdown). 3 October is missing from the dataset due to a script error. Images contained within a tweet (such as a Twitter picture) were collected only through hyperlink; they were not included in the scope of this specific textual analysis. It should be noted that no technology permits a complete data capture (Bruns and Burgess, 2011).
In total, four databases housed the collected tweets, accounting for 56,015 cases. This dataset supported the following questions guiding this analysis:
• In what ways has the #DearCongress conversation changed during the shutdown?
• If the discourse changed, why?
Following the collection period, the tweets were compiled and separated by date. The data visualization tool Wordle.net was used to aid in tracing discourse within the data by creating a word cloud. Word cloud tools produce a visual representation of words. After adding a dataset to Wordle, the “size of the words reflects how often they have occurred in the text” . A word cloud for each day’s body of tweets was made.
The word clouds alone do not provide an entire cue into the psychological processes associated with language, however. The tweets for each day were run through Linguistic Inquiry and Word Count (LIWC) software (2007) to numerically analyze the text for emotion. This software has been broadly accepted in the research community, particularly in psychology, communication and sociology fields. LIWC codes for “content words ... generally nouns, regular verbs, and many adjectives and adverbs [that] convey the content of a communication” . For example, Tausczik and Pennebaker (2010) categorically identified anger to include 184 operationalized words such as “hate,” “kill” and “annoyed”. The operationalized words were implemented in LIWC’s construction, creating categories. Additionally, targeting emotion frequencies (anxiety, anger, and sadness) provides “a deeper understanding of how people are processing a situation or event” .
All of the collected tweets shared one commonality: #DearCongress. This term was removed from the text prior to creating word clouds for each day’s conversation because while the hashtag is part of the overall content, it is viewed in this case as a signifier of the ad hoc community or topical interest (Bruns and Burgess, 2011). Its inclusion within the analysis would predominate the word cloud, thus hindering any attempt at discourse tracing. If this public were to dissolve after the collection period, visualizing the underlying discourse could at least reflect the changing nature, opinions, and commentary during the shutdown.
We present the findings in three acts — three days during the shutdown that display the ad hoc public’s evolution through discourse during the government shutdown. They are presented first with an event, cross referenced against a timeline from the Associated Press (2013), a visualization of the dominant Twitter discourse for that day, and specific examples, iterative of both. These presentations do not encompass the entire #DearCongress story; they merely represent the ad hoc public’s progression.
1 October: The start of a new fiscal year for the United States Government. With no operating budget passed, the government shuts down. (AP)
The most dominant word within the Twitter conversation during the first day of the shutdown was not a word at all, but rather two letters: “RT.” “RT” stands for “retweet” — meaning on Twitter, that an original piece of content was “shared” with another group of followers, most likely without alteration of the original text. During the collection period, “RT” appeared as a technical addition to the original body of text; it is usually only added to the content by the user through an act of clicking, “retweet.”
Figure 1: 1 October word cloud.
For example, this tweet appeared several times:
RT @AndrewJenks: #DearCongress the dude and the kids from those AT&T commercials could get more done than you guys. http://t.co/umqjH5LXAz
So did these:
RT: @NBCNews: #DearCongress: ‘You shouldn’t be getting paid’ http://t.co/U3zelRAiz8
RT: @NBCNews: #DearCongress: Americans weigh in with shutdown messages to Congress http://t.co/kLwmmn2DT5e
Tweets from NBC News, the Today Show, or some of their anchors appeared numerous times within the data. This may be a result of tweets from these entities, published the day prior to the government shutdown, asking their viewers to utilize the hashtag. Whether a result of NBC’s campaign or not, the 1 October discourse presented varied commentary:
lrntch: #dearcongress It is time to put the American people ahead of partisan politics.
Saints09chick: #DearCongress This land is our land, not yours. #KeepParksOpen
The discourse within the first day’s tweets varying in topic, scope, and pure differences in diction, numerically held less significance or frequency to the day’s top terms. Therefore, the varied discourse is not visually displayed above with the same prevalence as “RT.”
The collected cases from 1 October totaled 189,296 words. While the visual cue from the word cloud may be vague, the LIWC analysis for the key emotive words (anger, sadness, and anxiety) provided a base of insight. The quotients cited below represent the total percentage of the target words in the text.
Figure 2: 1 October target word frequency.
8 October: The government shutdown begins its second week after no progress is reported from meetings between Congressional Leadership and President Obama. (AP) The President held a press conference that afternoon to report on the shutdown and budget meetings.
The terms “RT” and “shutdown” remained at the top of conversation frequency. 8 October was the first day within the dataset in which RT began to significantly fall in its use, evident here by its size reduction. The decrease in RT signifies a decrease in content sharing and an increase in original composition. The second week of conversation in the ad hoc public began to visualize an increase in the use of emotive words such as “embarrassment.”
Figure 3: 8 October word cloud.
Those still actively speaking within the public shifted diction:
Clubmeds: If this goes past the 17th I could possibly lose my veterans disability pension which means my apartment and everything I own #DearCongress
jensmarriedlife: #DearCongress why are the democrats so afraid to use a medical system they designed?
PerleChampion: Get to work congress @TODAYshow #dearcongress http://t.co/2amvSYisOn
Figure 3 illustrates the shift the ad hoc public displayed from earlier discourse, shown in Figure 1. Shared content, marked with an, “RT,” decreased in prevalence, giving rise to an increase in original content. Tweets began to: reflect personal vignettes, be directed at someone, and usernames increased in frequency. Additionally, LIWC results confirmed a diction shift: anxiety signifiers rose to 1.02 frequency (an increase of 0.76), anger – 1.10 (increase of 0.34), and sadness increasing slightly to – 0.39 (increase of 0.03).
Figure 4: 8 October target word frequency.
16 October: The last day of the shutdown. After failing to gain enough support from Republican members in the House of Representatives on 15 October, Senators Harry Reid and Mitch McConnell announce a plan to reopen the government through 15 January and extend the U.S. debt limit through 7 February 2014. (AP)
Retweet practices decreased once again as emotive descriptors increased in prevalence in the word cloud, compared to Figures 1 and 5. The words prevalent in the ad hoc public’s conversation: going, disgrace, deadline, proved, serve, failed. When these words are analyzed alone in LIWC, they mark highly on program’s quotient for negative emotions.
Figure 5: 16 October word cloud.
Emotions were embedded within shared experiences,
donnageeoh: I explained the DC crisis to my 14 yr old daughter last night, in layman’s terms. Her reaction: “Well that’s just stupid.” Hmm #dearcongress
jccolyer: @ABC #DearCongress Ur method failed,going to deadline has proved ur motives,U serve urselves not people as u say,u r a disgrace
and pure emotion.
DeannaMorell: #dearcongress You’re a disgrace!! Do the country a favor and resign! You’re hurting this country! #shutdown
Emotive codes from LIWC shifted once again, returning to statistical values similar to the first day’s results. Anxiety measured 0.25 and anger dropped to 0.75. Sadness rose slightly, once again to 0.43.
Figure 6: 16 October target word frequency.
NBC, the news organization seemingly promoting #DearCongress, only appeared in others’ tweets. They do not appear within the collected dataset for the final day of the government shutdown as an originator of discourse; they were only mentioned.
Figure 1 presents one important consideration as we continue to use the Internet or online networks when sharing political opinions: structural limitations or constructs can impact the overall discussion. The prevalence of “RT” throughout the #DearCongress public is not one specifically created by the users. Rather, it signifies action and connectivity within the ad hoc public. Its prevalence throughout this particular discourse can signify: the rate at which people shared content or opinions about the shutdown; the alteration of argument due to Twitter’s structural design which governs its discourse (Freelon, 2010); the degree to which content may have dominated this conversation. A high rate of content may have originated from a pre-determined group and shared in greater length than original content was contributed to the conversation.
“RT” nearly serves as a disruption when analyzing the overall discourse through the dataset. It reminds observers that “at issue is the claim that the machines, structures, and systems of modern material culture can be accurately judged not only for their contributions to ... positive and negative side effects, but also for the ways in which they can embody specific forms of power and authority” . The transmission modes are just as political as the event being discussed in this case — data “artifacts” and collection method included.
Alternatively, the technical construct or insertion of “RT” into the discourse, particularly in the first week of the data, also signifies high engagement because the “users are not merely tweeting into the hashtag stream, but are also following what others are posting” . If the public’s members are following the stream, actively engaged they would contribute original content as the stream continued. So, within the ad hoc public, “as quality decreases, criticisms increase, which then increase quality ...” .
The shifting emotion quotients exemplified this notion of storming before norming. Anxiety and anger peaked by the middle of the shutdown, marked on 8 October. The observed movement supports Weick’s (1974) claim — criticisms increased and subsequently the quality of the conversation increased thereafter. The conversation quality is signified by the settling of anxiety and anger quotients by 16 October.
Emotion cues (anxiety, anger, and sadness) shifted. Considering that the language used to discuss an event “can reveal something about the extent to which a story may have been established or is still being formed” , the movement of emotive words requires attention. The numerical changes assigned to text by LIWC displays that this group continued to form and identify itself through language.
Members within the ad hoc public formed connections by directing commentary at each other. Structurally, Twitter reads a tweet with a username marked by an “@” symbol and listed exactly at the beginning of a tweet, as a mention or reply. The discourse utilized in the #DearCongress public develops further meaning when one observes where the conversation was directed. The “greater the total of participants who engage in this way — the more the hashtag community can be said to act as a community” (Bruns and Burgess, 2011). Speaker of the House, John Boehner’s Twitter account, @SpeakerBoehner, received the most number of comments from the public with 604 comments directed at him. These could have been in reply to a tweet or specifically directed comments. The next highest accounts receiving comments: @TODAYshow with 364 mentions and @NBCNews with 161 mentions. The Today Show and NBC News accounts appeared high on the reply list because they campaigned for #DearCongress’s use before the government shutdown, praeter hoc (in anticipation of the event).
@CarsonDaly: “Today the #OrangeRoom launched #DearCongress soliciting your comments on Govt shutdown. Its #1 trending topic on twitter rt now. Keep it up”
This very implementation of the word “launched,” implies that Carson Daly and the Today Show’s “Orange Room” not only launched, but intended to start the use of #DearCongress. In reality, the government shutdown is not the first time the hashtag appeared. If you search for “#DearCongress” in Twitter’s search engine, tweets dating back to 2010 will appear within your results (a result only viewable within the last year, due to changing techno-political structures on Twitter). #DearCongress’ earlier use and resurgence during the shutdown supports Bruns and Burgess’ claim that ad hoc publics are in flux. The NBC praeter hoc use of #DearCongress may have merely impacted what the ad hoc public looked like during this time period.
Bruns and Burgess (2011) argue that breaking events can yield multiple hashtags until a “settling” process occurs. Given the shifting discourse in this analysis, the same may have occurred within this ad hoc public. However, instead of the Twitter community organically fleshing out a specific hashtag to form a public, the ad hoc public organically shifted through conversation. The fact that NBC’s @TODAYShow trailed off from their praeter hoc activity begs the question: were they acting in #DearCongress the ad hoc public or #DearCongress the praeter hoc public? We may never know but the discourse and context implies ad hoc.
Even as a microblog, utilizing Twitter as a public forum, impacts political communication. The commentary within the ad hoc public on a blog — micro or not — can have a lasting impact. Political commentary in networked settings allows individuals to “improve the practiced experience of democracy, justice and development, a critical culture and community” . Despite the news organization drop off (if not complete departure from the ad hoc public) other members stayed active within the public. In a paper on political participation on blogs, Woodly (2008) states that, “besides the indirect effects that blogs may have on political activity through their influence on mainstream news media, they have direct effects on the interested public” . Absent a news organization, the ad hoc public shifted, cued by the shifting discursive activity and the political context.
We utilize the Internet as a platform not only for discovery but connecting through communication. We form groups, communities, and ad hoc publics through the communicative process. Textual artifacts result from this process, lending itself to analysis. Discourse tracing through data visualization can provide queues to emergent themes within a public at a given time. As a container for human interaction, Twitter specifically is a “platform for political deliberation” .
This research focused on the visualization of discourse in relation to other events connected to and impacting the conversation at large. These events cannot be separated from the content because it relies on context. After all, would #DearCongress have become a trending topic on Twitter if not for the context? The #DearCongress tweets from 2010 are seemingly few in number and arguably did not make a performative statement in the same manner as the big data set in 2013. The fact that a news organization with a captive audience utilized the same hashtag, promoting its use, is only further evidence for factors that contribute to the flux that can occur within an ad hoc public.
If we can visualize the discourse in relation to other events, other conversation, or even quantitatively examine word choice, it furthers the ad hoc public’s agency — a benefit to political communication in shared networks. As a predictor for political activity, the structural limitations and constructs must however be taken into consideration for further application. For example, retweets were not quantified for this writing, to avoid shifting the scope of this paper to a social network analysis. Approaching this dataset through discourse tracing provides a lens to further understand the “container for human interaction.” Understanding the diction, the public conversation, the open letter, the advocacy effort, may be “performance” (Goffman, 1959) but it can only aid onlookers when trying to understand a group or their agenda. Nevertheless, the means by which #DearCongress, the ad hoc public communicated, shifting the container they conversed in, may have said more to their government on furlough, than what they were saying. Because what was said, can mean something entirely different ... post hoc.
About the authors
Elisabeth A. Montemurro serves as a Communications Strategist at Tribune Publishing. She holds a Master’s of Communication in Digital Media and Storytelling from Loyola University Chicago.
E-mail: elisabeth [dot] montemurro [at] gmail [dot] com
David Kamerer, Ph.D., is an associate professor in the School of Communication at Loyola University Chicago.
E-mail: david [at] davidkamerer [dot] com
1. Carey, 2009, p. 18.
2. Bruns and Burgess, 2011, p. 8.
3. Weick, 1974, p. 358.
4. Poynter, 2010, p. 234.
5. Tausczik and Pennebaker, 2010, p. 29.
6. Tausczik and Pennebaker, 2010, p. 31.
7. Winner, 1986, p. 127.
8. Bruns and Burgess, 2011, p. 6.
9. Weick, 1974, p. 359.
10. Tausczik and Pennbaker, 2010, p. 33.
11. Benkler, 2006, p. 9.
12. Woodly, 2008, p. 119.
13. Tumasjan, et al., 2010, p. 183.
Associated Press, 2013. “Timeline: Congress’ path to federal shutdown” (17 October), at http://www.bigstory.ap.org/article/timeline-congress-path-federal-shutdown-7, accessed 17 February 2016.
Yochai Benkler, 2006. The wealth of networks: How social production transforms markets and freedom. New Haven, Conn.: Yale University Press.
Axel Bruns and Jean Burgess, 2011. “The use of Twitter hashtags in the formation of ad hoc publics,” Proceedings of the Sixth European Consortium for Political Research General Conference (25–27 August, University of Iceland, Reykjavik), at http://eprints.qut.edu.au/46515/, accessed 17 February 2016.
James Carey, 2009. Communication as culture: Essays on media and society. Revised edition. New York: Routledge.
Deen G. Freelon, 2010. “Analyzing online political discussion using three models of democratic communication,” New Media & Society, volume 12, number 7, pp. 1,172–1,190.
doi: http://dx.doi.org/10.1177/1461444809357927, accessed 17 February 2016.
Erving Goffman, 1959. The presentation of self in everyday life. Garden City, N.Y.: Doubleday.
Martin Hawksey, 2013. “Twitter archiving Google spreadsheet TAGS v5” (15 February), at http://mashe.hawksey.info/2013/02/twitter-archive-tagsv5/, accessed 29 September 2013.
Marianne LeGreco and Sarah J. Tracy, 2009. “Discourse tracing as qualitative practice,” Qualitative Inquiry, volume 15, number 9, pp. 1,516–1,543.
doi: http://dx.doi.org/10.1177/1077800409343064, accessed 17 February 2016.
Annette Markham, 2013. “Remix cultures, remix methods: Reframing qualitative inquiry for social media contexts,” In: Norman Denzin and Michael Giardina (editors). Global dimensions of qualitative inquiry. Walnut Creek, Calif.: Left Coast Press, pp. 63–81.
Ray Poynter, 2010. The handbook of online and social media research: Tools and techniques for market researchers. New York: Wiley.
Aaron Smith, 2013. “Civic engagement in the digital age,” Pew Research Center (25 April), at http://www.pewinternet.org/2013/04/25/civic-engagement-in-the-digital-age/, accessed 25 April 2014.
Yla R. Tausczik and James W. Pennebaker, 2010. “The psychological meaning of words: LIWC and computerized text analysis methods,” Journal of Language and Social Psychology, volume 29, number 1, pp, 24–54.
doi: http://dx.doi.org/10.1177/0261927X09351676, accessed 17 February 2016.
Andranik Tumasjan, Timm O. Sprenger, Philipp G Sander, Sandner, and Isabell M. Welpe, 2010. “Predicting elections with Twitter: What 140 characters reveal about political sentiment,” Proceedings of the Fourth International AAAI Conference on Weblogs and Social Media, at http://www.aaai.org/ocs/index.php/ICWSM/ICWSM10/paper/view/1441, accessed 17 February 2016.
Karl E. Weick, 1974. “Middle range theories of social systems,” Behavior Science, volume 19, number 6, pp. 357–367.
doi: http://dx.doi.org/10.1002/bs.3830190602, accessed 17 February 2016.
Langdon Winner, 1986. “Do artifacts have politics?” Daedalus, volume 109, number 1, pp. 121–136.
Deva Woodly, 2008. “New competencies in democratic communication? Blogs, agenda setting and political participation,” Public Choice, volume 134, number 1,, pp. 109–123.
doi: http://dx.doi.org/10.1007/s11127-007-9204-7, accessed 17 February 2016.
Received 15 August 2014; accepted 20 February 2016.
This paper is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
#DearCongress: A public letter
by Elisabeth Montemurro and David Kamerer.
First Monday, Volume 21, Number 3 - 7 March 2016
A Great Cities Initiative of the University of Illinois at Chicago University Library.
© First Monday, 1995-2017. ISSN 1396-0466.