Introduction

Much has been written of what lies inside (and outside) the digital humanities (DH). A fitting example might be the annual Day of DH, when hundreds of “DHers” (digital humanists) write about what they do and how they define the field (see https://twitter.com/dayofdh). Read enough of their stories and certain themes and patterns may emerge, but difference and pluralism will abound. More formal attempts to define the field are not hard to find—there is an entire anthology devoted to the subject (Terras, Nyhan, and Vanhoutte 2013)—and others have approached DH by studying its locations (Zorich 2008; Prescott 2016), its members (Grandjean 2014a, 2014b, 2015), their communication patterns (Ross et al. 2011; Quan-Haase, Martin, and McCay-Peet 2015), conference submissions (Weingart 2016), and so forth.

A small but important subset of research looks at teaching and learning as a lens through which to view the field. Existing studies have examined course syllabi (Terras 2006; Spiro 2011) and the development of specific programs and curricula (Rockwell 1999; Siemens 2001; Sinclair 2001; Unsworth 2001; Unsworth and Butler 2001; Drucker, Unsworth, and Laue 2002; Sinclair & Gouglas 2002; McCarty 2012; Smith 2014). In addition, there are pedagogical discussions about what should be taught in DH (Hockey 1986, 2001; Mahony & Pierazzo 2002; Clement 2012) and its broader relationship to technology, the humanities, and higher education (Brier 2012; Liu 2012; Waltzer 2012).

This study adds to the literature on teaching and learning by presenting a survey of existing degree and certificate programs in DH. While these programs are only part of the activities that make up the broader world of DH, they provide a formal view of training in the field and, by extension, of the field itself. Additionally, they reflect the public face of DH at their institutions, both to potential students and to faculty and administrators outside of DH. By studying the requirements of these programs (especially required coursework), we explore the activities that make up DH, at least to the extent that they are systematically taught and represented to students during admissions and recruitment, as well as where DH programs position themselves within and across the subject boundaries of their institutions. These activities speak to broader skills and methods at play in DH, as well as some important silences. They also provide an empirical perspective on pedagogical debates, particularly the attention paid to theory and critical reflection

Background

Melissa Terras (2006) was the first to point to the utility of education studies in approaching the digital humanities (or what she then called “humanities computing”). In the broadest sense, Terras distinguishes between subjects, which are usually associated with academic departments and defined by “a set of core theories and techniques to be taught” (230), and disciplines, which lack departmental status yet still have their own identities, cultural attributes, communities of practice, heroes, idols, and mythology. After analyzing four university courses in humanities computing, Terras examines other aspects of the community such as its associations, journals, discussion groups, and conference submissions. She concludes that humanities computing is a discipline, although not yet a subject: “the community exists, and functions, and has found a way to continue disseminating its knowledge and encouraging others into the community without the institutionalization of the subject” (242). Terras notes that humanities computing scholars, lacking prescribed activities, have freedom in developing their own research and career paths. She remains curious, however, about the “hidden curriculum” of the field at a time when few formal programs yet existed.

Following Terras, Lisa Spiro (2011) takes up this study of the “hidden curriculum” by collecting and analyzing 134 English-language syllabi from DH courses offered between 2006–2011. While some of these courses were offered in DH departments (16, 11.9%), most were drawn from other disciplines, including English, history, media studies, interdisciplinary studies, library and information science, computer science, rhetoric and composition, visual studies, communication, anthropology, and philosophy. Classics, linguistics, and other languages were missing. Spiro analyzes the assignments, readings, media types, key concepts, and technologies covered in these courses, finding (among other things) that DH courses often link theory to practice; involve collaborative work on projects; engage in social media such as blogging or Twitter; focus not only on text but also on video, audio, images, games, maps, simulation, and 3D modeling; and reflect contemporary issues such as data and databases, openness and copyright, networks and networking, and interaction. Finally, Spiro presents a list of terms she expected to see more often in these syllabi, including “argument,” “statistics,” “programming,” “representation,” “interpretation,” “accessibility,” “sustainability,” and “algorithmic.”

These two studies form the broad picture of DH education. More recent studies have taken up DH teaching and learning within particular contexts, such as community colleges (McGrail 2016), colleges of liberal arts and science (Alexander & Davis 2012; Buurma & Levine 2016), graduate education (Selisker 2016), libraries (Rosenblum, et al., 2016; Varner 2016; Vedantham & Porter 2016) and library and information science education (Senchyne 2016), and the public sphere (Brennan 2016; Hsu 2016). These accounts stress common structural challenges and opportunities across these contexts. In particular, many underscore assumptions made about and within DH, including access to technology, institutional resources, and background literacies. In addition, many activities in these contexts fall outside of formal degrees and programs or even classroom learning, demonstrating the variety of spaces in which DH may be taught and trained.

Other accounts have drawn the deep picture of DH education by examining the development of programs and courses at specific institutions, such as McMaster University (Rockwell 1999), University of Virginia (Unsworth 2001; Unsworth and Butler 2001; Drucker, Unsworth, and Laue 2002), University of Alberta (Sinclair & Gouglas 2002), King’s College London (McCarty 2012), and Wilfrid Laurier University (Smith 2014), among others. Abstracts from “The Humanities Computing Curriculum / The Computing Curriculum in the Arts and Humanities” Conference in 2001 contain references to various institutions (Siemens 2001), as does a subsequent report on the conference (Sinclair 2001). Not surprisingly, these accounts often focus on the histories and peculiarities of each institution, a “localization” that Knight (2011) regards as necessary in DH.

Our study takes a program-based approach to studying teaching and learning in DH. While formal programs represent only a portion of the entire DH curricula, they are important in several respects: First, they reflect intentional groupings of courses, concepts, skills, methods, techniques, and so on. As such, they purport to represent the field in its broadest strokes rather than more specialized portions of it (with the exception of programs offered in specific areas, such as book history and DH). Second, these programs, under the aegis of awarding institutions and larger accrediting bodies, are responsible for declaring explicit learning outcomes of their graduates, often including required courses. These requirements form one picture of what all DHers are expected to know upon graduation (at a certain level), and this changing spectrum of competencies presumably reflects corresponding changes in the field over time. Third, formal DH programs organize teaching, research, and professional development in the field; they are channels through which material and symbolic capital flow, making them responsible, in no small part, for shaping the field itself. Finally, these programs, their requirements, and coursework are one way—perhaps the primary way—in which prospective students encounter the field and make choices about whether to enroll in a DH program and, if so, which one. These programs are also consulted by faculty and administrators developing new programs at their own institutions, both for common competencies and for distinguishing features of particular programs.

In addition to helping define the field, a study of formal DH programs also contributes to the dialogue around pedagogy in the field. Hockey, for example, has long wondered whether programming should be taught (1986) and asks, “How far can the need for analytical and critical thinking in the humanities be reconciled with the practical orientation of much work in humanities computing?” (2001). Also skeptical of mere technological skills, Simon Mahony and Elena Pierazzo (2002) argue for teaching methodologies or “ways of thinking” in DH. Tanya Clement examines multiliteracies in DH (e.g., critical thinking, commitment, community, and play), which help to push the field beyond “training” to “a pursuit that enables all students to ask valuable and productive questions that make for ‘a life worth living’” (2012, 372).

Others have called on DH to engage more fully in critical reflection, especially in relation to technology and the role of the humanities in higher education. Alan Liu notes that much DH work has failed to consider “the relation of the whole digital juggernaut to the new world order,” eschewing even clichéd topics such as “the digital divide,” “surveillance,” “privacy,” and “copyright” (2012, 491). Steve Brier (2012) points out that teaching and learning are an afterthought to many DHers, a lacuna that misses the radical potential of DH for transforming teaching and professional development. Luke Walzer (2012) observes that DH has done little to help protect and reconceptualize the role of the humanities in higher education, long under threat from austerity measures and perceived uselessness in the neoliberal academy (Mowitt 2012).

These and other concerns point to longstanding questions about the proper balance of technological skills and critical reflection in DH. While a study of existing DH programs cannot address the value of critical reflection, it can report on the presence (or absence) of such reflection in required coursework and program outcomes. Thus, it is part of a critical reflection on the field as it stands now, how it is taught to current students, and how such training will shape the future of the field. It can also speak to common learning experiences within DH (e.g., fieldwork, capstones), as well as disciplinary connections, particularly in program electives. These findings, together with our more general findings about DH activities, give pause to consider what is represented in, emphasized by, and omitted from the field at its most explicit levels of educational training.

Methods

This study involved collection of data about DH programs, coding descriptions of programs and courses using a controlled vocabulary, and analysis and visualization.

Data collection

We compiled a list of 37 DH programs active in 2015 (see Appendix A), drawn from listings in the field (UCLA Center for Digital Humanities 2015; Clement 2015), background literature, and web searches (e.g., “digital humanities masters”). In addition to degrees and certificates, we included minors and concentrations that have formal requirements and coursework, since these programs can be seen as co-issuing degrees with major areas of study and as inflecting those areas in significant ways. We did not include digital arts or emerging media programs in which humanities content was not the central focus of inquiry. In a few cases, the listings or literature mentioned programs that could not be found online, but we determined that these instances were not extant programs—some were initiatives or centers misdescribed, others were programs in planning or simply collections of courses with no formal requirements—and thus fell outside the scope of this study. We also asked for the names of additional programs at a conference presentation, in personal emails, and on Twitter. Because our sources and searches are all English-language, the list of programs we collected are all programs taught in Anglophone countries. This limits what we can say about global DH.

For each program, we made a PDF of the webpage on which its description appears, along with a plain text file of the description. We recorded the URL of each program and information about its title; description; institution; school, division, or department; level (graduate or undergraduate); type (degree or otherwise); year founded; curriculum (total credits, number and list of required and elective courses); and references to independent research, fieldwork, and final deliverables. After identifying any required courses for each program, we looked up descriptions of those courses in the institution’s course catalog and recorded them in a spreadsheet.

Coding and intercoder agreement

To analyze the topics covered by programs and required courses, we applied the Taxonomy of Digital Research Activities in the Humanities (TaDiRAH 2014a), which attempts to capture the “scholarly primitives” of the field (Perkins et al. 2014). Unsworth (2000) describes these primitives as “basic functions common to scholarly activities across disciplines, over time, and independent of theoretical orientation,” obvious enough to be “self-understood,” and his preliminary list includes ‘Discovering’, ‘Annotating’, ‘Comparing’, ‘Referring’, ‘Sampling’, ‘Illustrating’, and ‘Representing’.

We doubt that any word—or classification system—works in this way. Language is always a reflection of culture and society, and with that comes questions of power, discipline/ing, and field background. Moreover, term meaning shifts over time and across locations. Nevertheless, we believe classification schema can be useful in organizing and analyzing information, and that is the spirit in which we employ TaDiRAH here.

TaDiRAH is one of several classification schema in DH and is itself based on three prior sources: the arts-humanities.net taxonomy of DH projects, tools, centers, and other resources; the categories and tags originally used by the DiRT (Digital Research Tools) Directory (2014); and headings from “Doing Digital Humanities,” a Zotero bibliography of DH literature (2014) created by the Digital Research Infrastructure for Arts and Humanities (DARIAH). The TaDiRAH version used in this study (v. 0.5.1) also included two rounds of community feedback and subsequent revisions (Dombrowski and Perkins 2014). TaDiRAH’s controlled vocabulary terms are arranged into three broad categories: activities, objects, and techniques. Only activities terms were used in this study because the other terms lack definitions, making them subject to greater variance in interpretation. TaDiRAH contains forty activities terms organized into eight parent terms (‘Capture’, ‘Creation’, ‘Enrichment’, ‘Analysis’, ‘Interpretation’, ‘Storage’, ‘Dissemination’, and ‘Meta-Activities’).

TaDiRAH was built in conversation with a similar project at DARIAH called the Network for Digital Methods in the Arts and Humanities (NeDiMAH) and later incorporated into that project (2015). NeDiMAH’s Methods Ontology (NeMO) contains 160 activities terms organized into five broad categories (‘Acquiring’, ‘Communicating’, ‘Conceiving’, ‘Processing’, ‘Seeking’) and is often more granular than TaDiRAH (e.g., ‘Curating’, ‘Emulating’, ‘Migrating’, ‘Storing’, and ‘Versioning’ rather than simply ‘Preservation’). While NeMO may have other applications, we believe it is too large to be used in this study. There are many cases in which programs or even course descriptions are not as detailed as NeMO in their language, and even the forty-eight TaDiRAH terms proved difficult to apply because of their number and complexity. In addition, TaDiRAH has been applied in DARIAH’s DH Course Registry of European programs, permitting some comparisons between those programs and the ones studied here.

In this study, a term was applied to a program/course description whenever explicit evidence was found that students completing the program or course would be guaranteed to undertake the activities explicitly described in that term’s definition. In other words, we coded for minimum competencies that someone would have after completing a program or course. The narrowest term was applied whenever possible, and multiple terms could be applied to the same description (and, in most cases, were). For example, a reference to book digitization would be coded as ‘Imaging’:

Imaging refers to the capture of texts, images, artefacts or spatial formations using optical means of capture. Imaging can be made in 2D or 3D, using various means (light, laser, infrared, ultrasound). Imaging usually does not lead to the identification of discrete semantic or structural units in the data, such as words or musical notes, which is something DataRecognition accomplishes. Imaging also includes scanning and digital photography.

If there was further mention of OCR (optical character recognition), that would be coded as ‘DataRecognition’ and so on. To take another example, a reference to visualization and other forms of analysis would be coded both as ‘Visualization’ and as its parent term, ‘Analysis’, if no more specific child terms could be identified.

In some cases, descriptions would provide a broad list of activities happening somewhere across a program or course but not guaranteed for all students completing that program or course (e.g., “Through our practicum component, students can acquire hands-on experience with innovative tools for the computational analysis of cultural texts, and gain exposure to new methods for analyzing social movements and communities enabled by new media networks.”). In these cases, we looked for further evidence before applying a term to that description.

Students may also acquire specialty in a variety of areas, but this study is focused on what is learned in common by any student who completes a specific DH program or course; as such, we coded only cases of requirements and common experiences. For the same reason, we coded only required courses, not electives. Finally, we coded programs and required courses separately to analyze whether there was any difference in stated activities at these two levels.

To test intercoder agreement, we selected three program descriptions at random and applied TaDiRAH terms to each. In only a handful of cases did all three of us agree on our term assignments. We attribute this low level of agreement to the large number of activities terms in TaDiRAH, the complexity of program/course descriptions, questions of scope (whether to use a broader or narrower term), and general vagueness. For example, a program description might allude to work with texts at some point, yet not explicitly state text analysis until later, only once, when it is embedded in a list of other examples (e.g., GIS, text mining, network analysis), with a reference to sentiment analysis elsewhere. Since texts could involve digitization, publishing, or other activities, we would not code ‘Text analysis’ immediately, and we would only code it if students would were be guaranteed exposure to this such methods in the program. To complicate matters further, there is no single term for text analysis in TaDiRAH—it spans across four (‘Content analysis’, ‘Relational analysis’, ‘Structural analysis’, and ‘Stylistic analysis’)—and one coder might apply all four terms, another only some, and the third might use the parent term ‘Analysis’, which also includes spatial analysis, network analysis, and visualization.

Even after reviewing these examples and the definitions of specific TaDiRAH terms, we could not reach a high level of intercoder agreement. However, we did find comparing our term assignments to be useful, and we were able to reach consensus in discussion. Based on this experience, we decided that each of us would code every program/course description and then discuss our codings together until we reached a final agreement. Before starting our preliminary codings, we discussed our understanding of each TaDiRAH term (in case it had not come up already in the exercise). We reviewed our preliminary codings using a visualization showing whether one, two, or three coders applied a term to a program/course description. In an effort to reduce bias, especially framing effects (cognitive biases that result from the order in which information is presented), the visualization did not display who had coded which terms. If two coders agreed on a term, they explained their codings to the third and all three came to an agreement. If only one coder applied a term, the other two explained why they did not code for that term and all three came to an agreement. Put another way, we considered every term that anyone applied, and we considered it under the presumption that it would be applied until proven otherwise. Frequently, our discussions involved pointing to specific locations in the program/course descriptions and referencing TaDiRAH definitions or notes from previous meetings when interpretations were discussed.

In analyzing our final codings, we used absolute term frequencies (the number of times a term was applied in general) and weighted frequencies (a proxy for relative frequency and here a measure of individual programs and courses). To compute weighted frequencies, each of the eight parent terms were given a weight of 1, which was divided equally among their subterms. For example, the parent term ‘Dissemination’ has six subterms, so each of those were assigned an equal weight of one-sixth, whereas ‘Enrichment’ has three subterms, each assigned a weight of one-third. These weights were summed by area to show how much of an area (relatively speaking) is represented in program/course descriptions, regardless of area size. If all the subterms in an area are present, that entire area is present—just as it would be if we had applied only the broader term in the first place. These weighted frequencies are used only where programs are displayed individually.

Initially, we had thought about comparing differences in stated activities between programs and required courses. While we found some variations (e.g., a program would be coded for one area of activities but not its courses and vice versa), we also noticed cases in which the language used to describe programs was too vague to code for activities that were borne out in required course descriptions. For this reason and to be as inclusive as possible with our relatively conservative codings, we compared program and course data simultaneously in our final analysis. Future studies may address the way in which program descriptions connect to particular coursework, and articulating such connections may help reveal the ways in which DH is taught (in terms of pedagogy) rather than only its formal structure (as presented here).

Analysis and visualization

In analyzing program data, we examined the overall character of each program (its title), its structure (whether it grants degrees and, if so, at what level), special requirements (independent study, final deliverables, fieldwork), and its location, both in terms of institutional structure (e.g., departments, labs, centers) and discipline(s). We intended to analyze more thoroughly the number of required courses as compared to electives, the variety of choice students have in electives, and the range of departments in which electives are offered. These comparisons proved difficult: even within an American context, institutions vary in their credit hours and the formality of their requirements (e.g., choosing from a menu of specific electives, as opposed to any course from a department or “with permission”). These inconsistencies multiply greatly in an international context, and so we did not undertake a quantitative study of the number or range of required and elective courses.

Program data and codings were visualized using the free software Tableau Public. All images included in this article are available in a public workbook at https://public.tableau.com/views/DigitalHumanitiesProgramsSurvey/Combined. As we discuss in the final section, we are also building a public-facing version of the data and visualizations, which may be updated by members of the DH community. Thus, the data presented here can and should change over time, making these results only a snapshot of DH in some locations at the present.

Anglophone programs

The number of DH programs in Anglophone countries has risen sharply over time, beginning in 1991 and growing steadily by several programs each year since 2008 (see Figure 1). This growth speaks to increased capacity in the field, not just by means of centers, journals, conferences, and other professional infrastructure, but also through formal education. Since 2008, there has been a steady addition of several programs each year, and based on informal observation since our data collection ended, we believe this trend continues.

Program titles

Most of the programs in our collected data (22, 59%) are titled simply “Digital Humanities,” along with a few variations, such as “Book History and Digital Humanities” and “Digital Humanities Research” (see Figure 2). A handful of programs are named for particular areas of DH or related topics (e.g., “Digital Culture,” “Public Scholarship”), and only a fraction (3 programs, 8%) are called “Humanities Computing.” We did not investigate changes in program names over time, although this might be worthwhile in the future.

Structure

Less than half of DH programs in our collected data grant degrees: some at the level of bachelor’s (8%), most at the level of master’s (22%), and some at the doctoral (8%) level (Figure 3). The majority of DH programs are certificates, minors, specializations, and concentrations—certificates being much more common at the graduate level and nearly one-third of all programs in our collected data. The handful of doctoral programs are all located in the UK and Ireland.

In addition to degree-granting status, we also examined special requirements for the 37 DH programs in our study. Half of those programs require some form of independent research (see Figure 4). All doctoral programs require such research; most master’s programs do as well. Again, we only looked for cases of explicit requirements; it seems likely that research of some variety is conducted within all the programs analyzed here. However, we focus this study on explicit statements of academic activity in order to separate the assumptions of practitioners of DH about its activities from what appears in public-facing descriptions of the field.
Half of DH programs in our collected data require a final deliverable, referred to variously as a capstone, dissertation, portfolio, or thesis (see Figure 5). Again, discrepancies between written and unwritten expectations in degree programs abound—and are certainly not limited to DH—and some programs may have not explicitly stated this requirement, so deliverables may be undercounted. That said, most graduate programs require some kind of final deliverable, and most undergraduate and non-degree-granting programs (e.g., minors, specializations) do not.

Finally, about one-quarter of programs require fieldwork, often in the form of an internship (see Figure 6). This fieldwork requirement is spread across degree types and levels.

Location and disciplinarity

About one-third of the DH programs in our dataset are offered outside of academic schools/departments (in centers, initiatives, and, in one case, jointly with the library), and most issue from colleges/schools of arts and humanities (see Figure 7). Although much DH work occurs outside of traditional departments (Zorich 2008), formal training in Anglophone countries remains tied to them. Most DH concentrations and specializations are located within English departments, evidence for Kirschenbaum’s claim that DH’s “professional apparatus…is probably more rooted in English than any other departmental home” (2010, 55).

The elective courses of DH programs span myriad departments and disciplines. The familiar humanities departments are well represented (art history, classics, history, philosophy, religion, and various languages), along with computer science, design, media, and technology. Several programs include electives drawn from education departments and information and library science. More surprising departments (and courses) include anthropology (“Anthropological Knowledge in the Museum”), geography (“Urban GIS”), political science (“New Media and Politics”), psychology (“Affective Interaction”), sociology (“Social and Historical Study of Information, Software, and Networks”), even criminology (“Cyber Crime”).

The number of electives required by each program and the pool from which they may be drawn varies greatly among programs, and in some cases it is so open-ended that it is nearly impossible to document thoroughly. Some programs have no elective courses and focus only on shared, required coursework. Others list dozens of potential elective courses as suggestions, rather than an exhaustive list. Because course offerings, especially in cross-disciplinary areas, change from term to term and different courses may be offered under a single, general course listing such as “Special Topics,” the list of elective course we have collected is only a sample of the type of courses students in DH programs may take, and we do not analyze them quantitatively here.

Theory and critical reflection

To analyze the role of theory and critical reflection in DH programs, we focused our analysis on two TaDiRAH terms: ‘Theorizing’,

a method which aims to relate a number of elements or ideas into a coherent system based on some general principles and capable of explaining relevant phenomena or observations. Theorizing relies on techniques such as reasoning, abstract thinking, conceptualizing and defining. A theory may be implemented in the form of a model, or a model may give rise to formulating a theory.

and ‘Meta: GiveOverview’, which

refers to the activity of providing information which is relatively general or provides a historical or systematic overview of a given topic. Nevertheless, it can be aimed at experts or beginners in a field, subfield or specialty.

In most cases, we used ‘Meta: GiveOverview’ to code theoretical or historical introductions to DH itself, though any explicit mention of theory was coded (or also coded) as ‘Theorizing’. We found that all DH programs, whether in program descriptions or required courses, included some mention of theory or historical/systematic overview (see Figure 8).

Accordingly, we might say that each program, according to its local interpretation, engages in some type of theoretical or critical reflection. We cannot, of course, say much more about the character of this reflection, whether it is the type of critical reflection called for in the pedagogical literature, or how this reflection interfaces with the teaching of skills and techniques in these programs. We hope someone studies this aspect of programs, but it is also worth noting that only 6 of the 37 programs here were coded for ‘Teaching/Learning’ (see Figure 12). Presumably, most programs do not engage theoretically with issues of pedagogy or the relationship between DH and higher education, commensurate with Brier’s claim that these areas are often overlooked (2012). Such engagement may occur in elective courses or perhaps nowhere in these programs.

European programs

All of the 37 programs discussed above are located in Anglophone countries, most of them in the United States (22 programs, 60%). We note that TaDiRAH, too, originates in this context, as does our English-language web searches for DH programs. While this data is certainly in dialogue with the many discussions of DH education cited above, it limits what we can say about DH from a global perspective. It is important to understand the various ways DH manifests around the globe, both to raise awareness of these approaches and to compare the ways in which DH education converges and diverges across these contexts. To that end, we gathered existing data on European programs by scraping DARIAH’s Digital Humanities Course Registry (DARIAH-EU 2014a) and consulting the European Association for Digital Humanities’ (EADH) education resources webpage (2016). This DARIAH/EADH data is not intended to stand in for the entirety of global DH, as it looks exclusively at European programs (and even then it is limited in interpretation by our own language barriers). DH is happening outside of this scope (e.g., Gil 2017), and we hope that future initiatives can expand the conversation about DH programs worldwide—possibly as part of our plans for data publication, which we address at the end of this article.

DARIAH’s database lists 102 degree programs, 77 of which were flagged in page markup as “outdated” with the note, “This record has not been revised for a year or longer.” While inspecting DARIAH data, we found 43 programs tagged with TaDiRAH terms, and we eliminated 17 entries that were duplicates, had broken URLs and could not be located through a web search, or appeared to be single courses or events rather than formal programs. We also updated information on a few programs (e.g., specializations classified as degrees). We then added 5 programs listed by EADH but not by DARIAH, for a grand total of 93 European DH programs (only 16 of which were listed jointly by both organizations). We refer to this dataset as “DARIAH/EADH data” in the remainder of this paper. A map of these locations is provided in Figure 9, and the full list of programs considered in this paper is given in Appendices.

The DARIAH/EADH data lists 93 programs spread across parts of Europe, with the highest concentration (33%) in Germany (see Table 1). We caution here and in subsequent discussions that DARIAH and EADH may not have applied the same criteria for including programs as we did in our data collection, so results are not directly comparable. Some programs in informatics or data asset management might have been ruled out using our data collection methods, which were focused on humanities content.

Program titles

A cursory examination of the DARIAH/EADH program title reveals more variety, including many programs in computer linguistics and informatics (see Appendix B). We did not analyze these titles further because of language barriers. And again, we caution that some of these programs might not have been included according to the criteria for our study, though the vast majority appear relevant.

Structure

Most programs in the DARIAH/EADH data are degree-granting at the level of master’s (61%) or bachelor’s (25%) (see Figure 10). While we are reasonably confident in these broad trends, we are skeptical of the exact totals for two reasons. In DARIAH’s Registry, we noticed several cases of specializations being labeled as degrees. Though we rectified these cases where possible, language barriers prevented us from more thoroughly researching each program—another challenge that a global study of DH would encounter. On the other hand, it’s also possible that non-degree programs were undercounted in general, given that the Registry was meant to list degrees and courses. Based on our inspection of each program, we do not believe these errors are widespread enough to change the general distribution of the data: more European programs issue degrees, mostly at the master’s level.

Location and disciplinarity

Most European programs are also located in academic divisions called colleges, departments, faculties, or schools (see Figure 11), depending on country. Only a handful of programs are located in institutes, centres, or labs, even less frequently than in our collected data.

We did not analyze disciplinarity in the DARIAH/EADH data because the programs span various countries, education systems, and languages—things we could not feasibly study here. However, 43 programs in the DARIAH/EADH data were tagged with TaDiRAH terms, allowing for comparison with programs in our collected data. These speak to what happens in DH programs in Europe, even if their disciplinary boundaries vary.

DH activities

To analyze the skills and methods at play in DH programs, we examined our TaDiRAH codings in terms of overall term frequency (see Figure 12) and weighted frequency across individual programs (see Figures 13 and 14). Several trends were apparent in our codings, as well as DARIAH-listed programs that were also tagged with TaDiRAH terms.

In our data on Anglophone programs of DH programs, analysis and meta-activities (e.g., ‘Community building’, ‘Project management’, ‘Teaching/Learning’) make up the largest share of activities, along with creation (e.g., ‘Designing’, ‘Programming’, ‘Writing’). This is apparent in absolute term frequencies (see Figure 12, excepting ‘Theorizing’ and ‘Meta: GiveOverview’) and in a heatmap comparison of programs (see Figure 13). Again, the heatmap used weighted frequencies to adjust for the fact that some areas have few terms, while others have more than double the smallest. It is worth noting that ‘Writing’ is one of the most frequent terms (11 programs), but this activity certainly occurs elsewhere and is probably undercounted because it was not explicitly mentioned in program descriptions. The same may be true for other activities.

Many program specializations seem to follow from the flavor of DH at particular institutions (e.g. the graduate certificate at Stanford’s Center for Spatial and Textual Analysis, University of Iowa’s emphasis on public engagement), commensurate with Knight’s (2011) call for “localization” in DH.

In contrast with the most frequent terms, some terms were never applied to program/course descriptions in our data, including ‘Translation’, ‘Cleanup’, ‘Editing’, and ‘Identifying’. Enrichment and storage activities (e.g., ‘Archiving’, ‘Organizing’, ‘Preservation’) were generally sparse (only 1.9% of all codings), even after compensating for the fact that these areas have fewer terms. We suspect that these activities do occur in DH programs and courses—in fact, they are assumed in broader activities such as thematic research collections, content management systems, and even dissemination. Their lack of inclusion in program/course descriptions seems constituent with claims made by librarians that their expertise in technology, information organization, and scholarly communication is undervalued in the field, whether instrumentalized as part a service model that excludes them from the academic rewards of and critical decision-making in DH work (Muñoz 2013; Posner 2013) or devalued as a form of feminized labor (Shirazi 2014). Ironically, these abilities are regarded as qualifications for academic librarian positions and as marketable job skills for humanities students and, at the same time, as a lesser form of academic work, often referred to as faculty “service” (Nowviskie 2012; Sample 2013; Takats 2013). We suspect that many program descriptions replicate this disconnect by de-emphasizing some activities (e.g., storage, enrichment) over others (e.g., analysis, project management).

Generally, there seems to be less emphasis on content (‘Capture’, ‘Enrichment’, and ‘Storage’ terms) and more focus on platforms and tools (‘Analysis’ and ‘Meta-Activities’ terms) within programs in our collected data. In interpreting this disparity, we think it’s important to attend to the larger contexts surrounding education in various locations. The Anglophone programs we studied are mostly located in the United States, where “big data” drives many decisions, including those surrounding higher education. As boyd and Crawford note, this phenomenon rests on the interplay of technology, analysis, and “[m]ythology: the widespread belief that large data sets offer a higher form of intelligence and knowledge that can generate insights that were previously impossible, with the aura of truth, objectivity, and accuracy” (2013: 663). Within this context, programs advertising analysis, visualization, and project management may appear as more attractive to prospective students and supporting institutions, two important audiences of program webpages. This influence does not mean that such activities do not occur or are not important to DH, but it again turns attention to questions about the way in which these skills are developed and deployed and whether that occurs against a backdrop of critical reflection on methods and tools. How these broad program-level descriptions play out in the context of particular courses and instruction is beyond the scope of this program-level study, but we think that surfacing the way programs are described is an important first step to a deeper analysis of these questions.

When comparing our 37 programs to the 43 TaDiRAH-tagged European ones, several differences emerge—though we caution that these findings, in particular, may be less reliable than others presented here. In our study, we coded for guaranteed activities, explicit either in program descriptions or required course description. In DARIAH’s Registry, entries are submitted by users, who are given a link to another version of TaDiRAH (2014b) and instructed to code at least one activities keyword (DARIAH-EU 2014b). We do not know the criteria each submitter uses for applying terms, and it’s likely that intercoder agreement would be low in absence of pre-coordination. For example, programs in the Netherlands are noticeably sparser in their codings than programs elsewhere—perhaps submitted by the same coder, or coders with a shared understanding and different from the others (see Figure 14).

We tried to compare directly our codings with DARIAH data by looking at five programs listed in common. Only one of these programs had TaDiRAH terms in DARIAH data: specifically, all eight top-level terms. When examining other programs, we found several tagged with more than half of the top-level terms and one tagged with 40 of 48 activities terms. These examples alone suggest that DARIAH data may be maximally inclusive in its TaDiRAH codings. Nevertheless, we can treat this crowdsourced data as reflective of broad trends in the area and compare them, generally, to those found in our study. Moreover, there does not appear to be any geographic or degree-based bias in the DARIAH data: the 43 tagged programs span ten different countries and both graduate and undergraduate offerings, degree and non-degree programs.

Comparing term frequencies in our collected data and DARIAH/EADH data (see Figure 12), it appears that enrichment, capture, and storage activities are more prevalent in European programs, while analysis and meta-activities are relatively less common (see Table 2). While both datasets have roughly the same number of programs (37 and 43, respectively), the DARIAH data has over twice as many terms as our study. For this reason, we computed a relative expression of difference by dividing the total percent of a TaDiRAH area in DARIAH data by the total percent in our study. Viewed this way, ‘Enrichment’ has over five times as many weighted codings in DARIAH as our study, followed by ‘Capture’ with over twice as many; ‘Analysis’, ‘Interpretation’, and ‘Meta-activities’ are less common. Thus, Anglophone and European programs appear to focus on different areas, within the limitations mentioned above and while still overlapping in most areas. This difference might be caused by the inclusion of more programs related to informatics, digital asset management, and communication in the DARIAH data than in our collected data, or the presence of more extensive cultural heritage materials, support for them, and integration into European programs. At a deeper level, this difference may reflect a different way of thinking or talking about DH or the histories of European programs, many of which were established before programs in our collected data.

Reflections on TaDiRAH

Since TaDiRAH aims to be comprehensive of the field—even machine readable—we believe our challenges applying it may prove instructive to revising the taxonomy for wider application and for considering how DH is described more generally.
Most examples of hard-to-code language were technical (e.g., databases, content management systems, CSS, and XML) and blurred the lines between capture, creation, and storage and, at a narrower level, web development and programming. Given the rate at which technologies change, it may be difficult to come up with stable terms for DH. At the same time, we may need to recognize that some of the most ubiquitous technologies and platforms in the field (e.g., Omeka, WordPress) actually subsume over various activities and require myriad skills. This, in turn, might give attention to skills such as knowledge organization, which seem rarely taught or mentioned on an explicit basis.

A separate set of hard-to-code activities included gaming and user experience (UX). We suspect the list might grow as tangential fields intersect with DH. Arguably, UX falls under ‘Meta: Assessing’, but there are design and web development aspects of UX that distinguish it from other forms of assessment, aspects that probably belong better with ‘Creation’. Similarly, gaming might be encompassed by ‘Meta: Teaching/Learning’, which

involves one group of people interactively helping another group of people acquire and/or develop skills, competencies, and knowledge that lets them solve problems in a specific area of research,

but this broad definition omits distinctive aspects of gaming, such as play and enjoyment, that are central to the concept. Gaming and UX, much like the technical cases discussed earlier, draw on a range of different disciplines and methods, making them difficult to classify. Nevertheless, they appear in fieldwork and are even taught in certain programs/courses, making it important to represent them in the taxonomy of DH.

With these examples in mind and considering the constantly evolving nature of DH and the language that surrounds it, it is difficult and perhaps counterproductive to suggest any concrete changes to TaDiRAH that would better represent the activities involved in “doing DH.” We present these findings as an empirical representation of what DH in certain parts of the world looks like now, with the hope that it will garner critical reflection from DH practitioners and teachers about how the next generation of students perceives our field and the skills that are taught and valued within it.

Conclusion and further directions

Our survey of DH programs in the Anglophone world may be summarized by the following points.

• The majority of Anglophone programs are not degree-granting; they are certificates, minors, specializations, and concentrations. By comparison, most European programs are degree-granting, often at the master’s level.
• About half of Anglophone programs require some form of independent research, and half require a final deliverable, referred to variously as a capstone, dissertation, portfolio, or thesis. About one-quarter of programs require fieldwork, often in the form of an internship.
• About one-third of Anglophone DH programs are offered outside of academic schools/departments (in centers, initiatives, and, in one case, jointly with the library). By comparison, most European programs are located in academic divisions; only a handful are offered in institutes, centres, or labs.
• Analysis and meta-activities (e.g., community building, project management) make up the largest share of activities in Anglophone programs, along with creation (e.g., designing, programming, writing). By contrast, activities such as enrichment, capture, and storage seem more prevalent in European programs. Some of these areas may be over- or under-represented for various cultural reasons we’ve discussed above.

As with any survey, there may be things uncounted, undercounted, or miscounted, and we have tried to note these limitations throughout this article.

One immediate application of this data is a resource for prospective students and those planning and revising formal programs. At minimum, this data provides general information about these 37 programs, along with some indication of special areas of emphasis—a compliment to DARIAH/EADH data. As we discussed earlier, this list should be more inclusive of DH throughout the globe, and that probably requires an international team fluent in the various languages of the programs. Following our inspection of DARIAH’s Registry, we believe it’s difficult to control the accuracy of such data in a centralized way. To address both of these challenges, we believe that updates to this data are best managed by the DH community, and to that end, we have created a GitHub repository at https://github.com/dhprograms/data where updates can be forked and pulled into a master branch. This branch will be connected to Tableau Public for live versions of visualizations similar to the ones included here. Beyond this technical infrastructure, our next steps include outreach to the community to ensure that listings are updated and inclusive in ways that go beyond our resources in this study.

Second, there are possibilities for studying program change over time using the archive of program webpages and course descriptions generated by this study. Capture of program and course information in the future might allow exploration of the growth of the field as well as changes in its activities. We believe that a different taxonomy or classification system might prove useful here, as well as a different method of coding. These are active considerations as we build the GitHub repository. We also note that this study may induce some effect (hopefully positive) in the way that programs and courses are described, perhaps pushing them to be more explicit about the nature and extent of DH activities.

Finally, we hope this study gives the community pause to consider how DH is described and represented, and how it is taught. If there are common expectations not reflected here, perhaps DHers could be more explicit about how we, as a community, describe the activities that make up DH work, at least in building our taxonomies and describing our formal programs and required courses. Conversely, if there are activities that seem overrepresented here, we might consider why those activities are prized in the field (and which are not) and whether this is the picture we wish to present publicly. We might further consider this picture in relationship to the cultural and political-economic contexts in which DH actually exists. Are we engaging with these larger structures? Do the activities of the field reflect this? Is it found in our teaching and learning, and in the ways that we describe those?

Acknowledgements

We are grateful to Allison Piazza for collecting initial data about some programs, as well as Craig MacDonald for advice on statistical analysis and coding methods. Attendees at the inaugural Keystone Digital Humanities Conference at the University of Pennsylvania Libraries provided helpful feedback on the ideas presented here. JITP reviewers Stewart Varner and Kathi Berens were helpful interlocutors for this draft, as were anonymous reviewers of a DH2017 conference proposal based on this work.

About the Authors

Chris Alen Sula is Associate Professor and Coordinator of Digital Humanities and the MS in Data Analytics & Visualization at Pratt Institute School of Information. His research applies visualization to humanities datasets, as well as exploring the ethics of data and visualization. He received his PhD in Philosophy from the City University of New York with a doctoral certificate in Interactive Technology and Pedagogy.

S.E. Hackney is a PhD student in Library and Information Science at the University of Pittsburgh. Their research looks at the documentation practices of online communities, and how identity, ideology, and the body get represented through the governance of digital spaces. They received their MSLIS with an Advanced Certificate in Digital Humanities from Pratt Institute School of Information in 2016.

Phillip Cunningham has been a reference assistant and cataloger with the Amistad Research Center since 2015. He received a BA in History from Kansas State University and MSLIS from Pratt Institute. He has interned at the Schomburg Center’s Jean Blackwell Hutson Research and Reference Division, the Gilder-Lehrman Institute for American History, and the Riley County (KS) Genealogical Society. His research has focused on local history, Kansas African-American history, and the use of digital humanities in public history.