Are the Digital Humanities Pragmatic? Should They Be?

Digital humanists and information scientists, with their skills at analyzing networks, trawling institutional histories, and processing text, are in many ways already developing techniques to intervene in the climate crisis. Using the tools of text mining, for example, scholars and digital practitioners could create systems that highlight and track the politicians around the globe who have resisted climate change legislation, focusing sustained attention on each speech or act of denial. Yet in the main, the digital humanities and social sciences have largely failed to engage climate change, especially as a space for concrete action-oriented projects. Here is a crucial missed opportunity for scholarship in the service of collective knowledge and collective action.

The textual analysis tools typical of certain digital humanities (DH) projects have rarely been applied to the discourse about the environment (for an exception, see Grubert and Algee-Hewitt). In the European Union, United States, Canada, and Australia, nationally backed grant programs have invested heavily in topics such as the twentieth-century novel and nineteenth-century newspaper, producing tools and initiatives that delve into questions of rhetoric, form, authorship, and content. Groundbreaking though these studies are, they operate—as much of the university and the humanities operate—as if national interests depend on engaging with a critical and accurate account of the development of the nation rather than what is truly needed: a planetary account of our current environmental crisis.

It might be objected that the humanities themselves exist less to serve any pragmatic “interest” than for the purposes of investigation into human experience and the history of ways of knowing more generally. Indeed, the shift that Matthew Gold and Lauren Klein have identified in “A DH That Matters” already demonstrates how humanistic values can be applied to events both current and past—for instance, using the tools of the digital humanities to aid the victims of hurricanes, to narrate the history of race, to document the bias of the technology industry, and to map Immigration and Customs Enforcement (ICE) detention centers. Where an earlier generation of digital humanists stressed the importance of “playfulness” when looking at data, data scientists engaging with questions of climate change more often experience an urgent drive to action, one that privileges pragmatic intervention in public discourse over playfulness per se.

The roots of such a pragmatic tilt have already been sown in certain quarters of the digital humanities that might, for our purposes, be identified as the “pragmatic digital humanities.” It is already evident in projects such as Torn Apart/Separados, which was designed to draw public attention to the pervasiveness of ICE detainment centers and specific social emergencies such as the children imprisoned at the U.S. border (Ahmed et al.). Digital humanists have also pioneered the building of infrastructural tools and standards of fair academic labor (Applegate; Risam; Guldi, “Scholarly Infrastructure”). They have begun important conversations about how “minimal computing” can minimize energy expenditure and make our projects accessible to citizens and scholars in the developing world (Schreibman et al.; Edmond; Gilio-Whitaker). In addition, digital humanists have documented standards for participatory, peer-to-peer design mechanisms that allow maps to become directly responsive to feedback from local communities—as embodied, for example, in the standards of the Bay Area’s Anti-Eviction Mapping Project (Golumbia and Koh; Graziani and Shi; Maharawal and McElroy; McElroy; Senier).

“Action-oriented” digital humanities projects such as these offer an important set of examples for how the light of DH scholarship can be brought to bear on the life of data, rhetoric, and argumentation in the textual documentation of the climate emergency, the most pressing issue of our time. These projects demonstrate a shift from the ideational work of imagined (or actual) readers toward the purposes held up by communities in all their complexities. As a result, these projects benefit from the skills of humanists and digital humanists—for example, in the dynamics of “reading” and “distant reading”—but also apply those skills in new ways, with new audiences and ends in mind and with the goal of creating concrete change in the world.

By classifying these projects as part of a “tilt,” I intend to describe a reorientation that does not require a wholly new critical or methodological stance. A “tilt” is proposed, in contrast to the epistemological “turns” of recent decades—the “linguistic turn,” the “cultural turn,” the “spatial turn,” and the “global turn” among them. In a spatial sense, turning suggests a diversion to one’s path to explore a hitherto unknown area. Such a description makes sense for all of those turns by which scholars trained in an existing field diverted their attention, with the help of a new set of theory, to explore a hitherto unknown arena of practice—say, the hidden, global interconnections between social movements or the spatial implications of historical technology. A tilt, by contrast, suggests that one might stay on one’s path, continuing to work on the same field—be it cultural history, diplomatic history, global history, the temporality of the novel, feminist criticism, or histories of the nation-state—while inclining one’s attention elsewhere. A tilt suggests less new content or a new method of inquiry than a reorientation from wherever one stands, taking account of the world and one’s own talents anew.

A pragmatic “tilt” in the digital humanities would not ask feminist scholars or historians of immigration to abandon their work in favor of some more noble calling, or to turn their scholarship toward some compelling new body of theory. Rather, a “tilt” would take the form of listening and learning to examples of scholarship from far-flung fields—including journalism, economics, and engineering—where scholars are already trying to join their labor to support the climate crisis.

By its nature, the climate crisis has many aspects that already overlap with existing fields. There are racial elements of the climate crisis, already well established in the scholarship around toxic pollution and ethnic neighborhoods (Camacho; Lerner; Taylor; Bullard). There are also global and feminist elements of the climate crisis, legal and diplomatic ones, and even aspects of climate change related to memory studies and the disappearing memories of the forests, coasts, and glaciers that have vanished in my lifetime. Humanists in general have an expertise in close reading and the critique of analysis that are well developed for understanding how communities and institutions respond or fail to respond to crises. And digital humanists, above all, have specific skills in project design, visual interfaces, and data analysis that can make almost any scholarly project more accessible—as well as tools that can make data actionable. Indeed, as Matthew Gold and Lauren Klein argued in “A DH That Matters,” a trend of orientation toward action is already visible in many DH projects today.

A truly pragmatic tilt would add to a general orientation toward action a concern with the powerful and pressing needs of the climate deadline, adding questions about climate response to each of the many subjects of inquiry and action that concern humanists and their interlocutors. Pragmatic thinking reminds us that we do not have infinite time for a response, and that many questions about politics where justice is needed will not solve the fundamental problem of survival. A pragmatic tilt does not mean setting aside problems of feminism or race—but it may require tilting those questions toward the climate crisis and the means of its repair by inspecting the failings of communities and institutions to limit carbon, to develop specific and fact-based discourses about the crisis, and to countenance the rights of all humans who may face new danger as a result of fire, flood, famine, and displacement.

Calls for a pragmatic response to the climate tragedy have abounded across the humanities. At the heart of many of these calls are questions about the data of climate change and its transparency, as well about the failure of society to respond to these and other warnings of climate change. Scholars in the environmental humanities have also undertaken investigations into what skills the public needs to interpret the data of climate change and its communication. One scholar of information studies, for example, emphasizes the need for “numeracy and visual comprehension” to prepare students to analyze discourse in a world where visualizations of climate change increasingly inform matters of public debate (Houser, “Climate Visualizations,” 1). Her critique is newly relevant in a world where a U.S. president can redraw the path of a hurricane on a map with a Sharpie, or when the climate denial articles of the Epoch Times go viral on Facebook. Reckoning with a widening divide between numerate scientists and naive (or willingly uninformed) citizens, an enormous variety of artists and writers have attempted to make climate change more knowable and more pressing through practices such as walking glaciers, describing arctic ice in detail, and inquiring into our emotional relationship with nature, the deep past, and futurity (Forman; Peterson; Lee et al.).

Certain experiments in the digital humanities have also attempted to reckon with the climate emergency, typically by engaging descriptive and emotional modalities to connect the present crisis with experiences of the past. DH scholars have experimented with, for example, mapping the sites of ecological change in Iceland against environmental descriptions in the medieval Icelandic sagas (Lethbridge and Hartman) or mapping the sites of British Romantic writing (D. Cooper). Some of these attempts concentrate on bridging data about geological timescales with human consciousness, thereby bringing abstraction into apprehensible reality; others consist of meditations about how to do the intellectual and emotional work of the digital humanities on a wider scale (Nowviskie). Exercises of this sort tend to emphasize the possibility for the subjective transformation of the individual through attention to nature, the individual’s integration with and dependence on nature, and the emotional responses that such an awareness inspires. One scholar of literature and the environment writes that scholars in her network “affirm the expertise of the humanities for transforming human preferences, practices, and actions in a time when there is a need for radical change” (Adamson, 354).

Emotional work of this kind is important for addressing the psychological impediments to acknowledging the stress of climate change, but it is not always clear whether aesthetic encounters with nature necessarily offer a bridge to timely, direct, pragmatic action. As experts in discourse and rhetoric, humanists have frequently offered aesthetic or subjective responses to nature as a solution to data’s seeming failure to persuade. But these are at best indirect and abstract responses—for instance, a concern with party-based denunciations of the evidence of climate change may motivate a critical investigation of the life of data in public discourse, or concern with climate denial may prompt experiments with mapping historical references to climate in the literature classroom. As one art historian concluded of his survey of climate-based photojournalism and art, “not all [approaches] have been effective in educating people about the dangers and causes of climate change or encouraging civic action and involvement” (Braasch, 33). It is quite possible that many current attempts at investigating the “digital environmental humanities” or other hybrid forms of thought will fail as the demands of a climate deadline force us to look at opportunities for organization and dangers in ever more concrete and specific ways. As one digital Victorianist was led to protest, “Why are we building and analyzing digital systems as ecosystems at the moment when entire natural ecosystems are being eradicated?” (Linley, 413).

A pragmatic tilt to the digital humanities might provide the answer to such a question—especially if digital humanists apply their talents at text mining and mapping to the voluminous materials created by corporations, governments, and communities wrestling with dilemmas that have their root in environmental peril. While much of the data about the climate crisis takes the form of quantitative measurements of carbon and methane, the governmental and social aspects of our failure to limit carbon emissions are documented in words. Likewise, words and narratives document the spreading realities of how communities are already suffering the consequences of wildfire, floods, storms, and the failures of institutions to take appropriate measures. Where the evidence is textual in nature, it can be analyzed with the skill sets of the digital humanities.

Text Mining as Real-Time Response

Where can the future authors of projects of this kind turn for inspiration? One answer is that there are already a large number of quantitative projects that build on text mining to make assessments of climate-related data. And while many of these projects have promising results, most of them could be improved through the skills of narrative analysis that humanists possess in abundance.

Scholars from numerous fields outside the humanities are already using text mining to track the responses or lack thereof in institutions such as corporations and civic authorities. In one climate-oriented project, researchers created a tool for mining climate news stories, the results of which were then used as indicators of how to hedge the stocks of companies that were linked negatively to climate events (Engle). The authors of the report advertise their tool as one that could automatically create “green” investment portfolios, but one could imagine the same analysis of climate-related news being used for monitoring governance, political discourse, and philanthropy from below. Text mining has also shown the effectiveness of clarity in the writing of environmental regulation; one text analysis study showed that more readable and precise directives about landfill design were better predictors of efficient waste management than spending (Richter, Ng, and Fallah). Examples of this kind demonstrate that text mining has pragmatic uses. Assessments of text can offer ratings of how well institutions are doing at responding to climate and where they fail. It can allow the close comparison and scrutiny of institutions.

Elsewhere, I have proposed that the creation of real-time data for actionable response grafts onto the skill set of “critique” the timely review of the “audit,” which is iterative by nature. Successive audits of institutions taken on a regular basis give researchers important information about collective progress over time (Guldi, Scholarly Infrastructure). Text mining offers many opportunities for “auditing” the way institutions are adapting to the climate emergency. Applied to the discourse of institutions, text mining allows us to ask: Has my state legislature or city council started talking more about solar panels and trees over the last year? Has my newspaper covered climate-related news every week this year? How does the language, human detail, and geographic specificity of that coverage compare with the reports of other states, cities, or newspapers?

Certain kinds of auditing might require other kinds of checks—for example, the ability to critically ascertain whether an institution is acting in harmony with its published climate action plan. It is not clear that every kind of audit is a good fit for text mining, but there are indications in the existing work that automatic text mining can be used as a tool for holding institutions accountable.

Text mining can support the assessment of an institution’s published reports over time. Scholars have begun to analyze the discourse, history, and political rhetoric of international organizations, drawing on the readily accessible transcripts of the World Bank and United Nations (Moretti and Pestre; Kentikelenis and Voeten). Applied to the published reports of institutions, text mining can provide important information about an institution’s bias, data that raises the possibility of comparing institutions and their peers and tracing institutional orientation over time. A study of the political bias of at the BBC tracked journalists’ “follows” on Twitter and reached the conclusion that “the BBC leans to the centre right” (Mills). Text mining provided an abstract metric for gauging the bias of the BBC; the metric’s importance is that it allows the comparison of the BBC with other news sources, thus allowing BBC journalists and readers to make appropriate decisions about how their content might change in the future. Ratings of each newspaper’s coverage of climate news in terms of political orientation, racial bias, and geographical scope—something that text mining could in theory support—would offer a valuable “report card” for auditing the progress of newspapers and other information sources, ranking each in their success at adapting to an age of climate change.

In fact, several text-mining studies have attempted to monitor responses to the climate emergency in the news media and the public at large. Faculty in forest management have used text mining to generate just-in-time histories of contemporary conflict over water, for example (Herrera et al.). A communications scholar has used text mining to analyze the effectiveness of political communication about climate (Majdik). Additional opportunities for a broad climate-emergency audit of communities and institutions might build on this work, although again the specific techniques would need to be refashioned to fit the specific questions raised by an audit that engages with questions about science, policy, and the public sphere—for example, understanding and interpreting the consequence of a right-leaning tilt in media coverage.

Perhaps the most sophisticated climate-related applications of text mining to date are those that attempt to model the varieties of causal understanding in the public realm. Here there are extremely promising techniques for the purposes of a cultural or institutional audit. A team of anthropologists and economists used diffusion models—typically used in marketing to capture expectations of dissemination and response—to measure the relative distribution of words collocated with a discourse of changing climate (Bentley et al.). The process allowed them to date moments of peak diffusion relative not to word count or proportion but rather the momentum of word dissemination. They traced what they believe to be a bias against discourse about the “global” and “adaptation” during recent years and argued that advocates of environmental policy work to define and disseminate the language of environmental awareness with an intensity that promotes “social learning” and allows certain patterns of speech and thought “to spread.” (Bentley et al., 7). Meanwhile, a sociologist has used social network analysis to identify communities associated with “misinformation” about climate change in philanthropy circles over the past twenty years (Farrell). Other scholars have applied the word-counting techniques to the problem of understanding resistance to climate change (Houser, Infowhelm). Studies of this kind offer a diagnostic of the spread and containment of misinformation about climate, thus adding to the potentially automated tools for auditing corrupted institutions and communities using their written discourse.

As we can see, across academia, multiple disciplines are turning toward text mining as a tool for making sense of climate change, for assessing public discourse, and for providing pragmatic responses. In disciplines ranging from civil engineering to literature, text mining has contributed to the cause of helping individuals and institutions to reckon with real-time opportunities for action around climate change. But the vast majority of studies of this kind come from computer science, data science, and information science.

Not all of the text-mining examples given here would meet the standards of rigorous textual analysis in the digital humanities in terms of content, rhetoric, and authorship. The landfill study (Richter, Ng, and Fallah) ranks all writing under an artificial metric of “readability” based on scores similar to those generated in the Microsoft Word grammar check. Readability scores make a poor substitute for discourse analysis, let alone a careful examination of the life of the data or a slow-grown examination of when and how a delay occurred in a political process—as DH approaches might explore. More generally, many of these text-mining studies demonstrate a weak engagement with the discursive elements of text. Such work is often characterized by simplistic measures (“readability scores” and “sentiment analysis” as opposed to deeper and historical discourse analysis), motivated by naive questions (How to save money on landfills? What would the public support, as opposed to how and when do cultures change? What is the work of science?), and informed by weak models of how concepts or science work in the world (through public opinion alone versus through a detailed inspection of the life of competing institutions, science, models, and data).

The relative lack of sophistication of many text-mining projects with respect to linguistic rhetoric and cultural expression reflects the strong need for interdisciplinary work in which scholars of discourse, history, and culture collaborate with information scientists. Digital humanists have much to contribute—indeed, humanists in general do. Humanistic skills of text mining, inspecting discourses over time, and comparing different styles of engagement are precisely the skills that are presently lacking in efforts to apply text mining to climate change. Indeed, recent publications by digital humanists demonstrate ample skill sets at analyzing and comparing the bias of textual data over time. In the disciplines of sociology and literature, scholars have applied text mining to group change in discourse about gender, race, and class over the twentieth century (Evans and Aceves; Kraicer and Piper; So, Long, and Zhuet). The skillful analysis of discourse in recent DH publications could usefully be applied to text mining the reports of public institutions about climate change with the consequence of creating data of genuine importance to political action.

Participatory Infrastructure

An additional opportunity for pragmatic engagement comes from the world of building critical infrastructure for members of the public to interact with the data generated by any text-mining projects that model the changing discourse of climate change. Researchers in various subjects have, in recent decades, created web portals to data that allow ordinary people outside the academy to investigate data, as rendered through a model endorsed by the scholar. In a recent paper, I suggest that such exploratory open datasets exemplify a range of scholarly, humanistic, and democratic values about replicability of research, transparency with respect to the origin of documentation, and accessibility of scholarship for a range of public purposes. Infrastructure building of this kind thus constitutes a mode of scholarly argumentation in which the work of scholarship is potentially replicated across hundreds of thousands of other uses (Guldi, “Scholarly Infrastructure”).

Infrastructures of this kind abound in certain domains of the academy—especially in the digital humanities, where literary scholars have focused on making tools for basic text mining of the literary canon accessible to all, among them Voyant Tools, the Hathi Trust, and multiple JSTOR Labs projects. By creating data infrastructures where citizens can explore data about texts for themselves, builders of sites for topic modeling and other forms of textual analysis not only analyze the institutions around them, but they also make it possible for ordinary citizens to replicate that analysis at a detailed level. By expanding access to the tools of analysis from the few to the many, infrastructure extends the possibility that activists, journalists, and other citizens will use text mining analytics to make sense of how democratic institutions work, bringing a multitude of silences and omissions to light.

Digital humanists have beautifully demonstrated the applicability of their toolsets to traditional questions in the humanities. They have striven to make their tools accessible to students and colleagues with little technical training. They have worked hard to tailor their toolkits to humanistic questions—for instance, the identification of “discourses” and “genres.” And they have begun the question of bridging DH toolsets with the leading political questions of the day—for example, the enduring bias toward whiteness and patriarchy of the American publishing industry. But a pragmatic tilt holds up a higher bar still. Where is the publicly accessible tool for tracking the discourse of climate science and climate action in the media or law? Were digital humanists to take up this question, we would contribute significant talents of discourse analysis to the public debate.

Drawing on decades of “participatory research” in the postcolonial world, citizen-science infrastructure is sometimes interwoven with workshops (called “barn raisings” at Public Lab) that specifically invite minoritized and other underprivileged populations to actively design research programs around a community’s specific needs and to collect data according to that program (Guldi, “A History of the Participatory Map”). In our era, the least powerful groups in society often occupy the least habitable landscapes for reasons of poverty, access, and power; they are doomed to inhabit terrain made toxic because of the limited inquiry of science and the regulatory failures of government (Pellow; Checker; Taylor; Waldron; Gilio-Whitaker; Washington; Cooper and Aronson). Citizens’ groups have used participatory maps as the basis for organizing and pursuing lawsuits against local polluters, as in the case of fishermen in southern India suing the local tannery over effluents (Narayanasamy).

In the university, however, climate science rarely supports or interfaces with these groups, although a small number of scholars, many of them in relationship with Indigenous or ethnic groups, have begun to highlight the strength of person-to-person relationships in these communities as a source for a larger civil politics capable of countering threats to the environment (Golumbia and Koh; Senier).

One important source of theory on how research and infrastructure can be bridged is the “critical infrastructure” discourse associated with Alan Liu and other digital humanists, whose community has routinely engaged in building websites and communities for the sharing of knowledge and whose ideas about web infrastructure align well with the task of pragmatic criticism. Digital humanists have much to offer in this space, having launched projects that are inclusive along the lines of gender and race and that question the legacy of empire using data-driven means (Despain; So, Long, and Zhuet; Risam). They engage in the creation of datasets, the programming of infrastructure, and the writing of algorithms, forms of praxis that resemble poiesis and criticism in their ability to be reproduced. In their instance as tools, algorithmic criticism also offers an ability to affect the world, rendering action at a distance, as Heidegger might argue, as of the hammer as an extension of the arm. In Margaret Linley’s phrase, tools such as data and infrastructure “shape human perception and cognition, forms of discourse, and patterns of social behavior” (413). But data, infrastructure, and algorithms can also reshape models, redounding into worlds of expert consensus. They can create new publics or commons where information is shared and exchanged.

Toward a Pragmatic Tilt

Can a pragmatic tilt across the university as a whole—deploying the skills of digital humanists and information scientists together to analyze public discourse in real time—meet the demands of the climate emergency that confront us? Pragmatic precedents in the digital humanities can usefully guide us toward those actions and spheres of intervention where our work would be most expedient, given the urgent need for political action to limit carbon emissions. There are opportunities here where digital humanists—and their allies in design, data analysis, and the humanities more generally—can usefully apply their talents in data gathering, narrative, and critical analysis to engage with local communities in such a way as to bear actionable witness to the reforms that are needed. Pragmatic work requires networks of builders from a range of disciplines, including the analysis of data, the building and design of interfaces, as well as a rich understanding of the contemporary desires of a variety of engaged communities on the ground who might offer their energies as participants if the data and its analysis were tailored to meet their needs.

Bibliography