A Digital Humanist-Informatician Review of Jim Ridolfo and William Hart-Davidson, editors, Rhetoric and the Digital Humanities

Rhetoricians can boast working in a more or less unbroken conversation stretching back to before Corax and Tisias, two millennia before modern English was spoken. Of course, humans have been speaking, reasoning, and arguing long before that, and all over the planet. The rhetorical tradition starting in the Mediterranean is generally considered unique because it both produced and theorized rhetoric at the same time. And in the present, the teaching and study of rhetoric is consistently theoretically self-aware. Surveying the methodologies of presenters at the 2015 Indiana Digital Rhetoric Symposium and also reflecting on the state of the field, Crystal VanKooten assigns theory more than half of a pie chart and concludes “Most of us are theorizing digital rhetoric”  [VanKooten 2015].

That is one reason for reading this book. It has been argued (by, for example, Drucker and Liu in the landmark collection Debates in the Digital Humanities) that digital humanities is strong on method and short on self-examination, criticism, and theorizing. Although most of the 23 chapters in this collection are applied studies of one form or another, they are all theoretically informed. As their authors – who are mostly faculty in rhetoric programs – join this conversation on the digital humanities, they bring a thoroughgoing theoretical examination.

Drucker and Liu’s stance toward DH is reinforced when contributors McNely and Teston assert: “Rhetorically-informed digital humanists should proceed with caution — doing DH is not as simple as choosing a digital tool and then combining that tool and tactic with a given methodological approach; indeed, a given tactic may be at odds with one's strategy”  [Ridolfo and Hart-Davidson 2015, 112]. Their observation might seem self-evident, an enthymeme grounded in how critics have been trained for decades: we always have a theory and some narrative of literary history in mind when we read. This may not be so self-evident in a time when it is not surprising to see a digital humanist pick up the latest technology and run with it just to see what happens. We might throw fifty novels at an algorithm to see what sticks, and think about interpretation and narrative after we have some results — and many might argue that this is an entirely valid methodology. But there is more at stake here. This book is published and I write this review in a moment when some argue that theory is no longer necessary — see for example Chris Anderson's essay “The End of Theory” and the cluster of essays it introduces in Wired Magazine [Anderson 2008], and the collection of essays published by Microsoft, The Fourth Paradigm [Hey et al. 2009]: machines process mountains of data and find patterns and correlations with more predictive power than our theories can. Beyond truisms, what we need is an examination of how the use of theory has and is changing. Digital tools are (or should be – provided one's tools are not a black box) manifestations of theory, such as Bayes' theorem or other statistical models, along with some basic structuralist assumptions.

Below I account for some of the contributions to this thoughtful collection.

The positionality of the digital humanities, and how it might disrupt established disciplinary lines, is much discussed at the moment (for that matter, so is the institutional status of English studies generally). Chapters take on these disciplinary issues, and these are all distinguished by self-awareness. Reid argues that just as digital humanities disrupts the humanities, digital rhetorics are post-human, post-structuralist disruptions of humanist traditions in rhetoric. Carter, Jones, and Hamcumpai argue for viewing disciplinary lines like rhetoric and DH not as territory to be seized and defended but rather as kairos (situations, occasions of opportunity and need for rhetoric). Walls builds a theory of interdisciplinarity (a “rhetoric of alliance”), based on histories of native peoples in North America.

Two theoretically-aware case studies of editing projects demonstrate how we might proceed on practical projects while trying to maintain theoretical awareness. Eyman and Ball narrate their efforts to put in place a new online publishing system for the online journal Kairos: A Journal of Rhetoric, Technology and Pedagogy. This effort at digital production included rhetorical self-analysis at every step. Along similar lines, Boyle asks what makes an online archive of Quintilian's works genuinely rhetorical.

Other contributors explore new methods to conduct research in writing and communication. Johnson suggests ways we might model scholarly influence, beginning with citation data. These include factor analysis, grouping by institutions, and geographic mapping. How he would get all the information he plans to model is not always clear, but his arguments could point to future studies. The primary counter-argument Johnson sets himself against is that many humanists are skeptical of numbers and suspicious of reducing their own influence to numbers (particularly as these numbers could be used for tenure) and he says, “Scholars who are hesitant to adopt numerical values as a way to bargain for tenure and promotions are giving up one of the more powerful argumentative rhetorics of the twenty-first century — the use of mathematics”  [Ridolfo and Hart-Davidson 2015, 100]. While the data available are far from complete,[1] this does not mean that studies that map what we can see, like what Johnson suggests, cannot lead to important results. If, say, we can see a third or half of all citations or some other fraction, then generalizations we make about influence could still be worth investigating in the ways that Johnson suggests, even if the rate of recall would be unacceptable for matters of tenure and promotion.

Kennedy and Long use online interaction to view the writing process. They argue that sources such as the version histories of wiki pages (where every single edit is recorded), can be an excellent place to study authorship: “these writerly moves and practices reveal the authorial life of the writer . . . as well as the life of the document”  [Ridolfo and Hart-Davidson 2015, 148]. They provide a short primer on qualitative approaches to online behavior, and discuss how the data can be collected, managed, and coded — as well as ethical considerations. This chapter would serve as an excellent starting point for research projects investigating digital authorship and online behavior more generally, as would McNely and Teston's chapter.

Hart describes his effort running a commercial program, DICTION, to perform sentiment analysis of the words in a corpus of texts with predefined genres. The corpus is described only as “sixteen thousand contemporary texts spanning the rhetorical universe”  [Ridolfo and Hart-Davidson 2015, 161] and I wish we were told what the texts were and how they were assigned to genres. He observes that “DICTION cannot distinguish between a sentence like ‘the dog bit the man’ and ‘the man bit the dog’”  [Ridolfo and Hart-Davidson 2015, 158].

To Hart’s credit, this is the only reference in the collection to the concept of the “bag of words,” so it was disappointing how quickly he moved past this important point. In topic modeling, a very popular method of text mining at the moment, a topic is defined as a distribution of words over a fixed vocabulary [Blei 2012] and should not be conflated with a rhetorical topos. Hart says that the software supposes “that audiences depend on human understandings of proportionality when responding to a text”  [Ridolfo and Hart-Davidson 2015, 158], that is, simply: the more optimistic words are contained in a text, the more optimistic a text is. This methodology holds up and shakes documents so that the words fall into a bag, defining as a text a set of probabilities of words. This general method, combined with the scale of corpora, would seem to flatten rhetorical situations and empty individual texts of their intentional (if we might otherwise be inclined to look for intention) and suasive content. I hope to hear more speculation by rhetoricians of what happens to rhetorical texts when they are viewed at the scale at which Franco Morretti and many other digital humanists prefer to operate. We can argue about whether Hart has adequately theorized his digital tools here, but this piece deserves credit for spurring exactly that argument.

Three thoughtful chapters on archives and digitization should be read by the broad community of researchers concerned with digital archives and libraries. Hoffman and Waisanen's chapter provides a survey of algorithmic tools for analyzing texts and corpora. Like Hart, they theorize these methods rhetorically. They bring in the theory of the “ideograph” as formulated by Michael McGee, referring to specific words that carry ideological weight in a given historic moment [Ridolfo and Hart-Davidson 2015, 180]. The idea here is similar to Raymond Williams' “keywords,” and several other formulations linking words with ideologies. When in DH we count words and find patterns, we hope that the words will serve as surrogates for something significant. But so far this is a blunt instrument and we cannot generalize from one case to another very well. This activity would benefit from more general theorizing, and contributors to this book offer some points of entry.

Samra Graban, Ramsey-Tobienne, and Myers do something that should be done more often. They survey theoretical issues – “invigorating epistemological dilemmas”  [Ridolfo and Hart-Davidson 2015, 235] – raised by digital archives: location, access, authorization, digitizations, publics, and metadata (along with Eyman and Ball’s, this essay stands out for addressing issues of metadata). This is a starting point, and deeper exploration is still needed.

Rice and Rice describe efforts at participatory archiving, similar to many projects in community-based archiving and community informatics (and they would probably benefit from studying this body of research). The archives Rice and Rice advocate are bottom-up and not necessarily built to last: “while its longevity may be useful for a number of reasons, we would not consider this archive a failure if it is erased tomorrow. The pop-up archive's focus is not in preservation but in the gesture and performance of archiving moments”  [Ridolfo and Hart-Davidson 2015, 251]. Like many contributions in this collection, I would love to take the essay into a seminar in informatics and ask graduate students to respond to this claim.

For some working in archiving and cultural heritage, this could sound like a shot across the bow. What I mean is: everything that Barack Obama and John Boehner say is carefully preserved with the intention of keeping it findable, usable, and trustworthy for as long as there might be persons to read it — because it is considered important. Many in the cultural heritage community are doing the same thing with the texts, artifacts, and stories left behind by members of subaltern groups which otherwise might be forgotten. On the one hand, building archives not defined by persistence is wonderfully provocative: it can prompt questions about the nature of memory and cultural production and narrative. On the other hand, the longevity of archives from Queen Elizabeth II or Donald Trump is in little doubt. We preserve what we value.

Potts's essay starts from a withering critique of digital archives: “these systems are clunky at best and irrelevant at worst”; they “prioritize data over experience”  [Ridolfo and Hart-Davidson 2015, 255]. She states “What these archives in practice and the digital humanities in general desperately need is a sense of audience, appeal, and interaction”  [Ridolfo and Hart-Davidson 2015, 255]. In fact, were Potts to walk into an iSchool she would find research and practitioners in whole subfields like knowledge organization, computer-supported cooperative work, community informatics, and even wonks in information retrieval who are constantly asking the same user-centered questions. My point here is not simply to refute a deficit Potts would fill, but rather to say that there are more people speaking her language than she seems to expect.

Several contributions cite Matthew Kirschenbaum's argument that students in the humanities should learn to code, and also Ian Bogost's formulation of “procedural literacy.” Two of the most interesting pieces, by Stolley and Ballentine, could be read as responding directly to Potts, arguing “Many of the discussions of making fetishize the concept of coding”  [Ridolfo and Hart-Davidson 2015, 257] — and this is another claim I would take into an information science seminar. This pair of essays by technical communication specialists makes fascinating arguments in favor of learning to program, and are among the most interesting in the volume (the editors did place them immediately following Potts' piece, and rhetoricians do love an argument).

Stolley discusses programming in Rails in an essay that, along with Hart's contribution, is probably the best read in the book. He argues that coding is an essentially writerly activity. In observing that Rails is installed and accessed entirely through the command line with no downloading, dragging, or double-clicking, he says, “Rails can be installed, invoked, and developed entirely though writing”  [Ridolfo and Hart-Davidson 2015, 265]. This is one of the most striking sentences, seeming self-evident once it is said yet thought-provoking, that I have read in a while.

Ballentine argues that the alternative to procedural literacy is “rapid obsolescence,” and “our future requires...collaboration with computer science and technical communication in order to not be shut out of important discussions (and our own interpretive practices) because we do not have the language to argue in these spaces”  [Ridolfo and Hart-Davidson 2015, 278–279]. He goes on to offer a case study of a technical writing project, providing documentation integrated into the display of X-ray, MRI, and CT scans. An intimate understanding of the code controlling how the documentation displays and interacts with the operation of the system is necessary for effective writing. Lack of such understanding would render the writing alone useless.

This volume should be read by digital humanists in their various disciplinary homes. It is a good introduction for those coming from a rhetoric background, and is of interest not only to those in English studies generally, but also to digital humanists in informatics programs. Of course, it is somewhat artificial to segregate rhetoricians, who swim in the same water as their colleagues in English and Communications departments, from other scholars with research agendas in digital humanities. But we can go further. Reading the essays in this book, I often found myself thinking that if these researchers were to walk into the nearest iSchool, and compare notes with researchers in informatics, both would benefit from the resulting conversations.