Of Time and Topics

The nineteenth century was a period of dramatic changes in science and technology in the United States. The railroad ushered in a new phase of mobility, speeding up travel and shortening the distances between places. Advances in printing technology and the use of the railroad to distribute publications widely and cheaply supported the formation of a variety of “imagined communities,” including new religious movements. The sciences offered new accounts of time and history, with evolution calling into question the biblical narrative of human creation and geology indicating a much longer history of the world than previously imagined. With time in flux in nearly every area of life, it is not surprising that the organization of time would feature as one of the defining characteristics of the emerging Seventh-day Adventist denomination. Holding to a literal six-day understanding of creation, and reclaiming what they understood to be the biblical organization of the week and the day, Seventh-day Adventists created and inhabited an alternative temporal imaginary. In so doing, they created space for resisting the prevailing shifts in understandings of gender, holding to a cooperative rather than individualistic social structure where both men and women were expected to work to spread the gospel within the home and the world.

In this dissertation I focused on time in the formation of Seventh-day Adventism. Using topic modeling to surface patterns across seventy years of the denomination’s periodical literature, I examined the yearly prevalence of the language of end-times expectation, found in interpretations of prophecy, descriptions of the second coming, interpretations of current and historical events in light of prophecy, and concerns about religious liberty. This approach enabled me to examine the denomination’s development over the long life of its founding prophet Ellen White. Looking beyond the denomination’s initial progression from revivalism to formal organization, the topic model of the periodical literature revealed repeating cycles of end-times expectation that shaped the rhetoric of the denomination and the development of gender norms. Repeatedly finding themselves at the edge of time, Seventh-day Adventists created a culture at odds with their surroundings, maintaining a communal understanding of salvation and an expansive understanding of those called to share the message. Throughout these cycles, Ellen White repeatedly articulated a vision of the religious life that situated women’s labor at the heart of the work of salvation, whether that was in giving testimony to the truth of the sabbath message, creating healthy homes and habits for children, sharing the adventist message through selling denominational literature, or working in the health institutions. However, as the prominence of end-times expectation waned, the language of domesticity and more male-centric understandings of the work of salvation came to dominate the denominational literature.

Additionally, I grappled with the application of current text analysis algorithms in historical research. Although the periodical literature of the denomination has been digitized and appears to be a source well suited for computational analysis, the idiosyncrasies of nineteenth-century print limited the effectiveness of OCR, resulting in low quality textual data, full of errors in both character and layout recognition. The messiness of the data limited the types of computational algorithms that could be used reliably with the denominations periodical data. Although itself sensitive to messy data, topic modeling with the cleaner texts in the corpus provided a way to generate a model of the patterns in the data despite the problems of word order created by errors in layout recognition. The topic model data opened opportunities for tracking the discourse of the denomination over time, as well as raised additional considerations. For topic modeling algorithms, such as Mallet, that do not include the relationship between time and word use as part of the model, how topics are aggregated shapes the resulting patterns. Algorithms that do include the relationship between time and content within the resulting model rely on assumptions regarding the nature of historical change to model shifts within topics over time, assumptions that may or may not match the understanding of time within the community studied. Consideration of the relationship between research questions, content, and algorithm is vital for the robust use of computational technology in humanities research.

Engaging these questions of computational methodology for historical research are increasingly important as the historical record becomes digital, both through digitization and as the percentage of “born digital” material increases. For historians of the late twentieth and twenty-first centuries, digital materials will be a necessary source for historical research, as cultural artifacts and institutional records are created, interacted with, and archived in exclusively digital forms. From problems of scale to issues of access, technical literacy is quickly becoming a required skill for working with the digital historical record.¹ Additionally, the growing ubiquity of data science is changing expectations around knowledge production and dissemination, changes that historians must grapple with for interpreting the past, and communicating in the present. As a result, theoretical work is needed to explore the relationship between these different forms of knowledge production and to propose methodologies for historical research that incorporate critical engagement with different computational techniques into the meaning-making process of historical scholarship.

The reorganization of time and society brought about by advances in transportation and communication technology in the nineteeth century created space for the rise of new religious movements and opportunities for envisioning alternative social structures. Seventh-day Adventism emerged out of that period of change, and offered members alternative ways of being in the world, even as they preached that the known world was coming to a close. In the present, computational technologies are opening up new ways of constructing knowledge and changing again notions of space and time. The internet has made it possible to communicate around the world seemingly instantaneously, and has created space for new communities to develop. But as has become increasingly apparent, not all community formation is positive. The same tools used for fostering connection between people have also become tools for disinformation and radicalization.

The development of Seventh-day Adventism draws attention to the ongoing influence of beliefs on the cultural norms of a group. For the denominational leaders, their beliefs about the second coming and their embrace of Ellen White’s visionary leadership created space for alternative constructions of gender and alternative social organization. Their rejection of the growing paradigm of individual economic success as the primary measure of worth provided opportunity for a more comprehensive and equitable understanding of gender and human flourishing to initially develop. While the disruptions of the early nineteenth century provided the space for these beliefs, the ongoing commitment to the nearness of the second coming incentivized the creation of a collaborative culture, one focused on fulfilling the Seventh-day Adventist calling to spread the message, first to fellow Adventists, then within the United States, and finally to the world.

Conflicts between religion and science due to beliefs about time, and particularly the length of history, are not just an artifact of the nineteenth century. Rather a range of current political debates, including approaches to climate change and international politics, are rooted in temporal disagreements. As with evolution, the science grounding our current understanding of climate change assumes a long history of the planet, a sense of historic time in direct conflict with the beliefs of a significant percentage of Protestant Americans. Similarly, expectations of cataclysmic events ahead of the second coming, strong beliefs in divine providence, and an emphasis on individual salvation that are the hallmarks of evangelical Protestantism decrease the effectiveness of calls for communal action to avert environmental or political catastrophe.²

More than a matter of personal conscience, beliefs give the world shape and inform how individuals interpret their experiences therein. In studying religious history, taking beliefs seriously is vital to understanding the development of religious groups and interpreting their actions in the world. While periods of religious revival may be linked to periods of economic or social disruption and explained through the opportunities created by the collapse of older social structures, the content of the beliefs also shapes how a community responds and the long-term effects of revival periods.³ For early Seventh-day Adventists, the belief that they were living at the edge of time motivated their actions in the world — their use of publishing to reach the Adventist community; their embrace of health to prepare themselves for salvation; and their political work on behalf of religious liberty.

Telling the history of Seventh-day Adventism requires wrestling with their beliefs about time, whether that work is done through traditional narratives or through data analysis and visualization. Doing so also requires taking into consideration the ways our modern organization of time and understanding of history shapes the research process.⁴ Using computational methods requires a careful consideration of the assumptions about the world that are built into the models and whether those assumptions illuminate or conceal what is being studied. The humanities offer strategies for examining the ways beliefs shape knowledge formation, whether within religious movements or within the scientific and technical community. Importantly, humanities scholarship reminds us that our current understanding of the world is historically and culturally contingent, that even something as seemingly stable as time has a history. In bringing together historical scholarship and computational methods, scholars have the opportunity to contribute to both our understanding of the past and our understanding of the present, making apparent the assumptions embedded in our technological infrastructure and offering a different vision of what that infrastructure could be.

Digital Dissertations in History

Despite the Internet being part of the modern information infrastructure for nearly thirty years, dissertations in history that leverage digital interfaces are still rare. This is in part because the place of digital scholarship in the larger ecosystem of scholarly publishing is unclear. In her digital dissertation, the first such project for the department of History and Art History at George Mason in 2016, Dr. Celeste Sharpe notes that digital dissertations fall into a gray area in the current publishing and analysis landscape.⁵ The format of a digital dissertation puts the work outside of the normal structures of submission, archiving, and publication. This is true for the full range of potential digital dissertations in history, from digital public history websites to interactive games and software.

For history, the dissertation has come in recent years to serve as the prototype or first draft of the scholar’s first monograph. As such, the required format and pacing of the dissertation has shifted to ease that transition from dissertation to book, reducing the requirements for historiographical surveys and emphasizing tight analysis and timely topics. For a digital dissertation, by contrast, the trajectory to the first book is less clear. While scholarly presses are increasingly interested in digital monographs and investing in platforms for their publication, these platforms in general envision a narrative text supplemented by multimedia elements, such as annotated images, video files, or interactive visualizations.⁶ While such a vision captures much of the work being done, these environments also constrain the possibilities for innovative work that pushes beyond multimedia enriched text. As survival in the academic environment depends on the ability to secure publication for scholarly work, finding ways to publish and archive a wider variety of digital work is necessary for this form of scholarship to thrive.

Additionally, relying on computational analysis, as I do in A Gospel of Health and Salvation, raises important questions about what is required of scholarship that uses computation and statistical modeling. A version of this dissertation could be submitted as a textual object with supporting visualizations. However, part of my argument in the dissertation is that the computational work is also part of the final dissertation object — the analysis of the dissertation cannot be fully evaluated without access to the computational work in a format where that work can be executed.⁷

That said, the experience of re-examining the format and methods of historical analysis through a study of Seventh-day Adventism, while productive, has been challenging. The range of questions and problems involved in creating a digital dissertation such as this pushes the boundaries of what can, or should, be tackled within the scope of a dissertation, a credentialing work. This is in large part because working in digital mediums and using computational algorithms blurs the boundaries between existing disciplinary areas. As a result, the criteria for what constitutes a scholarly production are unclear, as is the academic home for such research.

While reconsidering the format for dissertations is a useful first step in encouraging scholarship to embrace the opportunities made available with the rise of computational systems, the full embrace of these technologies is limited by the model of single-author scholarship still dominant in the humanities. A project such as mine could be the site of multiple dissertation-level projects in digital humanities, focused on different aspects from data collection and the uses of computational algorithms in analysis to interface design for interweaving the various modes of analysis. Such projects would be grounded in the research questions of the humanities field, all working to engage a larger motivating question in different formats: how the data represents the subject matter, how the algorithms model the subject matter, how the interface illuminates or obscures the argument. Such work requires more of a lab model for dissertation research, where all involved parties share the overall agenda of the lab and the elements of the project form the basis for multiple dissertation projects. Such a model would improve the overall quality of the final projects, would increase the opportunities for mentorship, and would provide training in the collaborative work that takes place for professional academics, public scholarship, and in “alt-ac” jobs.

Additionally, a full embrace of “digital humanities” research in history should involve supporting a wider range of research agendas than is currently the case. The requirements for digital dissertation at George Mason require first and foremost the construction of a historical argument, an intervention within the existing historical literature. It is worth considering whether this requirement unduly constrains the types of scholarship that can, and need to be, productively undertaken within the discipline of history, especially in terms of the study and development of methods. There is a range of research needed investigating the methodological and epistemological consequences of digital technology for historical knowledge production, including considerations of sources, the consequences of adopting the methods of data science for historical inquiry, and the relationship between interface and argument. To also require such work to make a substantive intervention into a particular historical debate is to over-task, and thereby limit, the work that can be done by any individual researcher. One reason for this dual approach is the realities of the academic job market, where content area jobs dominate. Looking ahead, however, there is potential for “digital history” jobs to become more prevalent, especially with growing institutional support for data science initiatives. Expanding the model of historical scholarship to include multi-authored and methodology-focused work is the first step to making the complex research that could be done under the umbrella of “digital history” possible. This would also reflect a return to older, more expansive, models of the “historical enterprise,” where the boundaries between scholars, public historians, librarians, and archivists were less clearly defined.⁸

These shifts would require even more institutional and cultural change than that needed for allowing multimedia dissertations. There are real questions, however, at the intersection of the humanities and technology and it behooves academic departments to innovate in terms of the model of scholarship in order to engage those questions. From exploring the rise of our current data-drive culture to evaluating the epistemological and ethical consequences of the growing reliance on data science methodologies in policy creation, social organization, and decision-making, scholars trained at the intersection of historical inquiry and data science methodologies are uniquely able to contribute to current discussions and to prepare students for a wide range of roles in the current information economy. My work in this dissertation gestures toward that future, attempting to engage the epistemological shifts embedded within computational technologies and methods, as well as to make an intervention into the existing literature in religious history. My hope, however, is that the field continues to expand in the future and takes on the challenge of supporting historical research focused on the questions of methods and methodology made pressing in the digital age.