The Life Aquatic

Like digital humanists, amphibians lead divided lives. Most breathe air but can only reproduce in water. Some species resolve this tension by organizing their lives around a pilgrimage from one environment to the other. After toads find a pond to spawn in, their descendants spend part of their lives swimming before growing legs and lungs and climbing back into the air. But there are also other solutions. Certain species of aquatic frogs spend their entire life cycle in water, rising to the surface periodically to breathe.

Most digital humanists spend most of their lives in a humanities department; like toads (so to speak), we are basically at home on that ground. But to create new digital humanists, we also need other environments. A graduate student or assistant professor who wants to become a digital humanist often has to spend some time swimming around in another field to gain experience with digital media or computational methods before returning to the home discipline. At least, this seems to be the informal backstory behind many careers. Systematizing that metamorphic life cycle as a formal plan of graduate education remains a bit challenging.

I probably do not have to describe the challenges in detail because they have been explored in other essays. In chapter 6 of the present volume, Manfred Thaller’s reflections on “digital Groundhog Day” trace a long history of (mostly failed) attempts to institutionalize graduate training in digital humanities dating back to the 1970s. My own attempts failed, about one decade ago, in ways that closely resemble the story Andrew Goldstone tells in “Teaching Quantitative Methods: What Makes it Hard (in Literary Studies).” Like Goldstone, I started by trying to squeeze all aspects of DH into one graduate seminar in an English department. It seems insane now, but in 2012, there were many reasons to try to pack everything in. Above all, I could not assume that an English department would contain a large population of students committed to taking multiple DH courses. So any course I designed would need to be pitched for students new to the field and perhaps wary of it. In that context, it seemed to make sense to offer a general overview of DH, focused on theoretical debate with a light introduction to data analysis along the way.

Like Goldstone, I found that this sort of course struggled to achieve its stated goals. I think I provided a decent overview of theoretical debate, but I basically failed to prepare students to understand or evaluate the computational side of DH. Computational analysis is simply not something that can be taught well in a single course and certainly not “along the way” in a course where it competes with other topics. In 2012, it perhaps seemed plausible to address computation casually because conversations about DH pedagogy were then framing the computational part of the discipline as a collection of user-friendly “tools.” But as others have since pointed out, the tool metaphor was misleading (Tenen, “Blunt Instrumentalism”). Even if we skip coding and teach students to use GUIs, students will need to evaluate the results their GUIs produce. That means learning statistics. Because I could not squeeze a semester of statistics into the margins of a course on the theoretical debate in DH, my course was not really putting students in a position to critically evaluate computational research. Certainly, it was not teaching them how to do it on their own.

I now realize that getting students to the point where they can do computational research requires at least three or four semesters. Students need a semester or two of programming, a course in statistics, and perhaps a capstone course where they learn to apply these methods to the unstructured data and slippery historical questions that typify the humanities. And this still only gets them up to speed in one aspect of DH. For full preparation as a digital humanist, separate courses might be needed on topics like digital preservation and digital scholarly communication. If I had been warned about this in 2012, my answer would have been a sigh. I did not know how to make any of that possible inside an English department curriculum.

Like amphibians, digital humanists have evolved a variety of ways to address a constitutive tension between different environments. We can incorporate aspects of other subjects in the humanities, or we can build an interdisciplinary program, combining courses from several departments. Many of the contributions to this volume explain how to make those models work. But at my own university, there was another option. Just two blocks from the English department, the School of Information Sciences was already offering courses in programming, statistics, data ethics, book history, and a half-dozen other topics relevant to DH. Just as importantly, the value of those courses was clear in the context of information science: graduate students who invested in preparation for computational work would not feel isolated by a risky, controversial choice. (For a graduate student’s perspective on the tensions confronting DH students in English departments, see Sean Weidman’s contribution in chapter 16 in this volume.)

Because there seemed to be no point in duplicating an institution that already worked well, I asked whether it was possible to move half of my job into information science. The figure later became 75 percent. I have continued to teach courses in English, but most of my graduate training is now done in information science, and I train students mostly for jobs in that discipline. In other words, I changed my disciplinary identity, so that it would be possible to reproduce without a long metamorphic pilgrimage, and I became an aquatic frog.

Changing disciplines felt like a risky move in 2016, but I am increasingly grateful I had the opportunity. At the School of Information Sciences, I found collaborators from a wide range of disciplinary backgrounds and an institution that was committed to helping students turn a mixture of humanistic and technical interests into a career. In 2012, I could say that DH added value to an English major, but my promises probably didn’t sound confident because they weren’t. Now it is easy to point to students who have gone on to work as librarians, research data engineers, or professors of information science. I work in a context where those career paths are familiar: the faculty and staff of the school know how to support them.

I am now sure that a school of information science can be a good place to train digital humanists. But the purpose of this chapter is not to convince anyone to make the same choice. That would be absurd because few of us really have a choice. The institutional options for digital humanists at a given college or university are constrained by local history. At most institutions, schools of information science do not even exist. One does exist at the University of Illinois Urbana-Champaign (UIUC) because we are a large public research university. But information science is rarely found at smaller private universities and colleges.

So I am not advocating for a mass migration to information science. Most digital humanists will continue to work in other disciplines. I just want to describe the way DH training works in my discipline to enlarge our sense of the field’s diversity. Information science is different from the core humanities departments that dominate conversation about digital humanities. As a result, I often hear claims about the necessary shape of graduate study that sound to me unconsciously parochial. “It would be unrealistic to expect students to do X,” for instance, when I know a context where X is already the norm.

First, a bit of historical background. Many programs in information science have roots in libraries. My school was founded in 1893 to staff the multiplying libraries of the midwestern and western United States. It remained a school of “library science” for most of the twentieth century.1 But the rise of information technology in the 1960s posed a challenge and an opportunity for a discipline centered on the management of paper codices. Questions about library organization now clearly overlapped with the new field of information retrieval. By the 1970s, many library schools were adding “Information Science” to their name or simply adopting the latter phrase (Olson and Grudin, “The Information School Phenomenon”). The iSchools movement, which crystallized at the beginning of the twenty-first century, explicitly aimed to harmonize the social ideals and service tradition of librarianship with a broader educational mission that prepares students to manage and analyze information in the private sector as well.

The task of theoretically justifying this synthesis has produced a great deal of discussion (Bawden and Robinson, Introduction to Information Science, 37–60; Zhang and Benjamin, “Understanding Information Related Fields”). My task here is not to prove the unity of information science but simply to observe that it is in practice a diverse tradition. Some observers describe the discipline as a social science, but large parts of it could just as easily be called “humanistic” or “computational” (Cibangu, “Information Science as a Social Science”). At UIUC, for instance, the study of children’s literature is centered in the iSchool at the Center for Children’s Books. But I also have colleagues who teach computational subjects like text mining and bibliometrics. Between those two poles, the center of the institution is occupied by social-scientific research, especially on the social implications of information technology. Because information technology has had ample social implications lately, it is unsurprising that several leading public intellectuals have worked in iSchools (Zeynep Tufekci) or were trained there (Safiya Umoja Noble).

I have already confessed that my immediate motive for moving into a new discipline was a recognition that at UIUC, the computational part of DH graduate training would inevitably be centered in the iSchool. Because my own work is computational, that was decisive. But many other aspects of digital humanities are an equally good fit for information science. Questions about access to cultural heritage, and about the ethical and political risks of new technologies, for instance, are central to the mission of an iSchool. More broadly, Marcia Bates has argued that information science is a “meta-science” (“Invisible Substrate,” 1043). Like librarians themselves, information scientists try to understand how other academic fields do their work, and they try to help them do it better. Reflection of this kind has long been central to DH, and several early DH pioneers spent part of their careers at iSchools (see Drucker, Graphesis; Renear et al., “Refining our Notion”; Unsworth, “Scholarly Primitives”).

There are also pragmatic reasons why the collaborative culture of a (largely) social-scientific field can be a congenial place to train digital humanists. For one thing, it addresses the awkward problem that DH projects are usually too big to make good dissertations. If a student had to build everything themselves—as is the norm in English—their six years could easily be up before they even had a corpus to study. In information science, by contrast, it is normal for many chapters to emerge from coauthored articles. The dissertation is still fundamentally an individual project, but it can use the ongoing collective work of a lab as a starting point that saves the author from having to reinvent every wheel.

On the other hand, I want to acknowledge that information science does not excel at all aspects of DH graduate training. There are important things—even central things—we cannot provide. The humanities differ from the social sciences, after all, partly by insisting on the specificity of historical and cultural contexts. Social scientists can limit their inquiries to a particular nation or period, but many social science disciplines perceive that sort of specificity as a constraint, whereas in the humanities, it is often the whole point. If this conception of the humanities is embraced, it will be clear that digital humanities can never become a purely methodological project. It is inseparably bound up with the specificity of languages, periods, and cultures.

Although many students and faculty in information science have acquired that sort of knowledge, it is not usually the mission of the school to transmit it. We do not teach courses in, say, Chinese history or nineteenth-century British poetry. To be sure, there are exceptions to the rule. Many iSchools have domain expertise in children’s literature, for instance, because libraries have historically been central to its transmission. The history of science and technology is another place where iSchools often develop domain knowledge. But these are special cases. Usually our contribution lies in methodological reflection rather than historical specificity. This is appropriate because most of the graduate students we serve are master’s students who will get jobs as librarians or data scientists—in other words, as reflective experts in method who contribute across a wide range of domains.

We also teach undergraduates and doctoral students. But the emblematic doctoral thesis in information science is expected to advance understanding at a higher level of generality than the typical history or English dissertation. It may explain how other disciplines organize knowledge, or illuminate the social implications of technology, or show how new algorithms can be applied more effectively. It is possible to do those things while contributing to historical understanding of a particular cultural context. But students who want to make that sort of historical contribution would be well advised to build a network in a second discipline as well, presenting at its conferences, publishing in its venues, and connecting with mentors who can provide an informed critique of their domain-specific argument.

In short, I see no way for graduate education in DH to avoid interdisciplinarity. Graduate students studying DH in a humanities department will probably need to venture out into computer science, statistics, or quantitative social science if they want to use computational methods, or even critique them effectively. By the same token, students in a computational or social science discipline (like information science) will only fully understand the humanistic dimension of DH if they pause at some point in their careers to develop deep understanding of a particular historical context. Aspiring digital humanists will always need both kinds of training. But they still usually have to choose one department as an institutional home, and the choice is consequential because different parts of the scholarly life cycle become easier or harder in different locations.

That is why I introduced this chapter with a stretched zoological metaphor. There is no right way to be an amphibian or a digital humanist. The problems created by a complex life cycle are resolved in reality by ecological diversification, not by finding a single correct answer. At the same time, I want to be candid about the reasons why computational humanists, in particular, need information science as one possible option. Fitting computational training into a humanities degree program remains a Sisyphean task. Digital humanists have been trying to do it for decades, but it is not the path of least resistance for anyone, so there is a recurring temptation to abbreviate, downplay, or outsource the computational part of the training, as Thaller’s story about Groundhog Day makes clear. That may change eventually. But in the 2020s, I think computational humanists need to be able to point to other institutional worlds. We will not know what this project can be until we see it unfold in an institutional context where local incentives encourage full exploration of its challenges. For computational DH to seem worth all the hard work it requires, it especially needs to be clear how computational training opens new opportunities for students. That becomes easy to explain when DH is integrated into a larger curriculum where data curation, information policy, and data science have a central place.

So I expect graduate training in digital humanities to flourish not only in humanities departments but in information science (and perhaps in computer science as well; see Benjamin Charles Germain Lee’s intriguing report in chapter 30 in this volume). Training humanists in departments that have “science” in their names may create tension, but I think this tension is constructive. A version of DH fully captured by social science might lose its grounding in cultural specificity. A version fully contained by humanities departments might have difficulty sustaining methodological ambitions that do not seem to fit yet in the humanities curriculum. But the excitement of DH comes from a struggle to unite those two goals: maximally deep domain expertise and maximally adventurous methodological exploration. One way to guarantee that this dialectic remains lively is to keep the field divided across several disciplinary niches. Diversification will make the whole ecosystem stronger.