Abstract
In this webtext, I develop an in situ approach for the rhetorical study of large-scale social media data. Grounding this in situ methodology in rhetoric and feminist critiques of data and visualization, this webtext models techniques and strategies for collecting, analyzing, and visualizing Twitter data. I utilize a large Twitter data collection, visualization applications, and grounded theory to locate unfolding rhetorical encounters that occur around shared interests rather than unified and stable hashtags or images. This approach extends rhetorical and interdisciplinary scholarship calling for attention to materiality and infrastructure in digital methodologies, and engages rhetoric's focus on contemporary everyday phenomena and lived experience. I offer a feminist rhetorical framework for data analysis and visualization, and demonstrate this framework with a Twitter data collection of approximately two million tweets and associated metadata gathered through Twitter's Application Programming Interface (API) from February 2016 to 2018 using the keyword "gentrification." Taken together, the visualizations generated through this methodology create a nuanced, polyvocial antenarrative of anti-gentrification rhetorical circulation on Twitter. I conclude by discussing the limitations and future directions of this study and offer recommendations to extend this study's feminist rhetorical methodology through scholarly, institutional, and community collaborations.
Introduction
In recent years, rhetoric and writing studies scholars have begun to engage with large datasets and the technological tools involved in collecting, processing, and visualizing data related to a vast array of social phenomena. Laurie Gries (2017) combined data types, including social media data, to track, largely through data visualization tools, the ongoing circulation and transformation of the Obama Hope image. William I. Wolff (2015) described his methodology for collecting and examining a large dataset of tweets around a Bruce Springsteen concert to characterize Twitter-based fan writing practices. Jason Palmeri and Ben McCorkle (2017) took "A Distant View of English Journal, 1912–2012," using coding schemes and data visualization methods to historicize the disciplinary development of writing studies' relationship to technology. In each publication, the authors thoughtfully reflect on the constraints and limitations of their studies, particularly highlighting the ways in which their choices in data collection, processing, and visualization have shaped their findings.
Studies like these, which engage large datasets and also acknowledge the influence of "the interplay of material infrastructures, users and texts" (Leurs, 2017, p. 137), begin to trouble the positivist veneer glossing over distant reading, data analysis, and visualization. As rhetoric and writing studies scholars continue to tread into the waters of data and visualization, the landscape of technological tools and their affordances continue to transform, becoming simultaneously easier to use and more "blackboxed" (see Eyman, 2016). As such, it becomes increasingly difficult—and important—for digital rhetoric and writing studies scholars to contribute methodological perspectives that illuminate the often unseen rhetoricity of data visualizations and provide new possibilities for engaging technologies not just theoretically, but through critical and inventive data and visualization projects of our own. To contribute to existing efforts in this direction, this webtext offers a rhetorical and feminist methodological approach to social media data analysis and visualization that deepens data viz's critical possibilities and conclusions, even within hard-coded software, by expanding the possibilities of data visualizations rather than relying on data analysis to advance previously "undiscovered truths" (boyd & Crawford, 2012).
In this webtext, I share the feminist rhetorical methodology of social media data and visualization I developed as I've worked over the past two years to collect and analyze a large dataset of tweets containing the keyword "gentrification." Like Gries's (2017) "Mapping Obama Hope" project, I'm interested in how writing and visual rhetoric circulate to activate publics and ignite collective actions; however, rather than tracking an object, like an image, I'm particularly interested in locating contemporary, unfolding encounters in their diverse materiality as they impact political and cultural phenomena like gentrification and contribute to resisting dominant narratives and power structures. In the larger project connected to this webtext, Twitter has served as a kind of fieldsite, providing for the study of gentrification encounters and rhetoric in circulation. Extant scholarship in urban studies and in rhetoric and writing studies, for the most part, examines gentrification rhetoric in hindsight. While there are benefits to historical studies, for those interested in how residents and activists are understanding and communicating about present-day gentrification, social media sites like Twitter are places of contemporary social encounters from which rhetoric emerges. Indeed, Twitter has proven to be a communication space for marginalized populations to voice dissent and to collectively take action to challenge dominant power structures (Jules et al., 2018). Rather than focusing on any particular, discrete artifact, social media platforms as places of encounter are appropriate fieldsites for the "study of vernacular, material, and 'live' rhetorics" (McHendry et al., 2014, p. 294); these types of in situ social media data studies fill an important gap, bringing digital methods to critical, participatory rhetoric and bringing rhetorical perspectives to contemporary phenomena as they unfold. Additionally, by collecting, reading, visualizing, and preserving contemporary social media data on gentrification, the larger research project connected to this webtext locates anti-gentrification rhetoric circulating on Twitter and participates in the recovery and amplification of non-dominant narratives not frequently circulating in dominant media outlets. Such in situ Twitter studies enhance rhetorical circulation theories and methodologies and amplify knowledge and narratives about gentrification created and framed not by dominant media, government officials, or urban planners, but by ordinary citizens and activists experiencing various forms of gentrification first hand.
In Racial Shorthand: Coded Discrimination Contested in Social Media, rhetoric and writing studies scholars Cruz Medina and Octavio Pimentel (2018) "set out to unpack the dominant narratives that undermine the media produced by communities of color, further erasing the rhetorical, oral, and aural traditions of these communities" (para. 5). In a collection that spans themes and methodologies, Medina and Pimentel argued that social media has become a space in which activists and ordinary citizens can "mediate and create for the purposes of activism, critique, literacy documentation, and culturally relevant storytelling" (para. 5). Likewise, Leigh Gruwell (2018) observed that rhetoric and writing studies scholarship has "embraced social media as both an object of study as well as a methodological tool" (para. 2) and yet has not gone far enough to consider how the platform shapes our research. The study described in this webtext engages with Twitter as an entry point into a wider ecology of emergent social media platforms, physical environments, and bodies that participate in encountering and shaping anti-gentrification rhetoric on Twitter and in discrete but interconnected gentrification scenes. Exploration of these encounters informs our understanding of activism in the digital age and the way in which networked visuality and communication technologies complicate our understanding of material processes like gentrification, collective actions related to activism, and, when taken together, the "ecological relationality as participating in rhetorical practices and their theorization" (Rickert, 2013, p. 3).
Analyzing this large social media dataset has been inseparable from the processes of data visualization. Due to the enormous size of this collection, I wouldn't have been able to work with this dataset without the technological tools of visualization, or "the reduction and spatial representation of datasets in such a way as to make them more intelligible than in their pre‐visualization, tabular format" (Hepworth, 2017, p. 8). This webtext won't only focus on the visualizations I've generated, however, but will also delve into how feminist rhetorical methodologies might work with and against their technological affordances. In the case of this study, quantification and pattern finding served as beginning points for formulating subsequent research directions that use both analytical and inventive approaches to studying and visualizing Twitter data. As such, data analysis and visualization are enhanced and complicated by adopting feminist rhetorical methodologies that value qualitative data, multiplicity, and narrative. In referring to feminist theories, I'm connecting to diverse and varied critical perspectives that "encompasses a range of ideas about how identity is constructed, how power is assigned, and how knowledge is generated, as well as how a range of intersectional forces such as race, class, and ability, combine to influence the experience of being in the world" (D'Ignazio & Klein, 2016, p. 1). This work responds to Johanna Drucker's (2017) call "to attend to how visualizations do their work if we are to fully understand the work that they do" (p. 913). Such a methodology provides for greater understanding of the way in which such data is shaped and adds nuance to the way in which we work with large datasets to make knowledge and share insight gained through data analysis and visualization. In the following pages, I demonstrate how I've developed a rhetorical feminist methodology through strategies of rhetorical arrangement that recover stories of activism, value the small data in large collections, and create multiple visualizations that add polyvocal nuance to large-scale data analysis and visualization.