There’s a Toy in My Essay: Problems with the Rhetoric of Text Analysis
There’s a Toy in My Essay: Problems with the Rhetoric of Text Analysis
Abstract and Keywords
Why do we fear computers in the humanities? Chapter 5 looks at the rhetoric around the use of computers in traditional activities like interpretation. A outline of how we interpret texts with computers is drawn from John B. Smith’s work on Joyce and “Computer Criticism.” The phases start with one of demarcating what is the text to be encoded, then analyzing it into parts, that can then be synthesized into new views. This leads to a discussion of new statistical and visual forms of synthesis and how they are useful when dealing with big text collections. An article from Etienne Brunet from 1989 on the exploitation of large corpora is used to look again at how visualizations and later hermeneutical interactives are appearing in humanities arguments. We return to how such toys can be embedded live into essays in way that allows readers to play with the interpretation without fear.
Do not trust the horse, Trojans! Whatever it is, I fear the Greeks, even bringing gifts.
Virgil, Aeneid, Book II
In Bruce McDonald’s “semiotic zombie movie” Pontypool, the protagonists theorize about what has infected the citizens.1 The local doctor, who has taken refuge with the aging talk-radio host and his producer in the sound booth of the radio station, speculates about the virus they have heard that friends have caught:
It is in words. Not all words, not all speaking, but in some. Some words are infected. And it spreads out when the contaminated word is spoken. Ohhhh. We are witnessing the emergence of a new arrangement for life … and our language is its host. It could have sprung spontaneously out of a perception. If it found its way into language it could leap into reality itself, changing everything. It may be boundless. It may be a God bug.2
Of course, in a movie that takes place almost entirely in a radio station in a small Ontario town, with the protagonists reporting mysterious events outside, we never really know what is going on. The doctor’s semiotic zombie theory could be just a wild interpretation by a madman. The doctor may be infected and therefore be infecting listeners as he theorizes on air. The irony of the movie is that it treats the interpretation of the word as infectious, as if ideas conveyed by a radio show could actually change things. The title of the novel on which the movie is based says it all: Pontypool Changes Everything.
The idea that words could be vectors for a plague sounds unusual, even for a zombie movie. But why should we be sceptical of the idea that words can change people? What scholar doesn’t hope that her words will in some way affect her readers or infect change? Why not, by extension, imagine that words could infect readers? If we can talk seriously about malicious (p.84) code as a “virus,” or propaganda as causing mass hysteria, then why not imagine malicious language as a parasitic organism? This idea is not new to Pontypool, something the movie itself acknowledges; one scene features a copy of Neal Stephenson’s 1992 novel Snow Crash. Rather, the idea goes back to antiquity. Titus Lucretius Carus’ De Rerum Natura (circa 50 BCE) describes a plague sweeping Athens, choking and silencing people. In the context of Lucretius’ exposition of Epicurean philosophy, this is a warning about the influence of words. More recently, Stephen Greenblatt, in The Swerve (2011), has argued that the rediscovery of De Rerum Natura by the Italian humanist Poggio Bracciolini in the fifteenth century put Epicurean ideas back into circulation, infecting modern thought so that it swerved toward science.
In previous chapters we discussed what text analysis is, presented an example, and discussed how it plagues websites. In this chapter, we will look at the analytical toys that are infecting online discourses, at how interpretive toys work within and with words, and at the relationship between hermeneutica and that which they interpret. We have noted the emergence of tools woven into websites; now we want to ask about hermeneutical widgets woven into the prose of online essays. We see this as an issue of the rhetoric of digital analytics such as interactive panels, or hermeneutica: How do they convey information in a context?
The online version of this book at hermeneuti.ca has examples of hermeneutica embedded in the prose. Did you try exploring the panels? How did you interpret a word cloud rendering of a chapter? What do you think of having text toys infect an essay? How do these hermeneutica present themselves? How do authors hide or show their analyses? We will conclude the chapter by revisiting the unease some humanists have with computing toys as potential vectors of analytical dis-ease.
Examples from Print Rhetoric
In considering the challenge of articulating analytical methods in the rhetoric of the humanities, we will look at an early example by a computing humanist who struggled with how computers could be used to convey both technique and interpretation. Our example is John B. Smith’s 1973 article “Image and Imagery in Joyce’s Portrait: A Computer-Assisted Analysis,” which presents a thesis about the use of imagery by James Joyce demonstrated through a computer-generated distribution graph of image words.
Smith’s work on the ARchive Retrieval and Analysis System (ARRAS), discussed above in chapter 3, demonstrates that Smith had the technical (p.85)
(p.86) background for such work. H. Van Dyke Parunak called Smith a pioneer in “displaying and analyzing such distributional problems.”3 Smith later argued that “diachronic distributions, Fourier Analysis, Principal State Component Analysis, diagrams, and CGAMS are all models that may be used to explore thematic structures and relations.”4 He was one of the first to think about how to use visualization to interpret a text and how to use it to convey an interpretation. For a number of reasons, “Image and Imagery in Joyce’s Portrait” is a great example of an early hybrid essay to study:
• It was written to argue a thesis about Joyce, not about computing. There are a lot of articles in Computers and the Humanities from the 1960s on that discuss technology, but far fewer that use computing primarily to make claims about a work of literature.
• It was written for an audience not familiar with computing and was published in a collection (Directions in Literary Criticism) intended for literary specialists, and it tries to illustrate what interpretive work can be done with computers.
• It is relatively short and tightly organized, and it illustrates the compromises one makes when introducing computed results in an interpretive essay.
• It is nicely mirrored by Smith’s 1978 article in the journal Style, “Computer Criticism,” in which he theorizes about the use of computer-assisted analysis for a literary audience. We can thus compare what Smith does against what he says about what can be done.
In “Image and Imagery” Smith sets out to demonstrate that the aesthetic theory discussed by Joyce’s protagonist, Stephen Dedalus, in the last chapter of A Portrait of an Artist as a Young Man informs the development of the novel and that it is “indeed, Joyce’s own.”5 To do this, Smith recapitulates Dedalus’ aesthetic theory of image. “The most important moments in the development of his personality,” Smith theorizes, “should be related to those passages in the text where the heaviest concentrations of important images or image components occur.”6 He “used a high-speed computer” to confirm that his hypothesis is correct.7 The resulting imagery distribution plot (figure 5.2 here) is then analyzed chapter by chapter.
How does Smith explain his use of computers to plot the richness of imagery across Joyce’s novel? To paraphrase Smith, he uses a “rather mechanical” series of operations designed to generate a computer readable list of some 1300 image words that have “sensory or thematic value.”8 Since his “emphasis is on interpretation not methodology,”9 he glosses over the tedious process used to generate the list and concludes simply that “this set (p.87) of words were selected as images represents an axiom on which the study is based.”10
Smith’s next step is an interpretive move that involves “the translation of the thesis into a form that the computer can apply to the novel.”11 He breaks the text into 500-word segments,12 then evaluates the intensity of the imagery in each segment by comparing the number of image words with their occurrences in the novel. The resulting graph is a page-by-page plot of the richness of imagery—pages dense with imagery show as peaks, pages with less imagery as valleys. Smith’s explicit instructions to readers assume that they have little knowledge of visualization:
To read the graph, assume that the novel runs from left to right (first word, Once, occurs at the extreme left side and the last word, stead, occurs at the extreme right). The richness of imagery rises and falls as one proceeds through the novel.13
The rhetorical effect of the computer-assisted analysis does not, however, lie in the methodological paragraphs, but in the graph. It is a type of graph that should be quite familiar to Smith’s readers: a distribution graph or histogram. The reader can explore the graph as they read the chapter-by-chapter exposition at the heart of the essay, comparing their sense of novel’s epiphany against the graph.14 This is the rhetorical power of visualizations: they encourage exploration, showing you something you are drawn to interpret rather than telling you how the computer may have assisted in the interpretation. These visualizations involve you in the interpretation rather than telling you how to interpret.
In his article “Computer Criticism” Smith completes the move from imagery, as it is manifest in computer visualizations of various kinds, to a critical theory that opens the way for computational methods. What makes this hermeneutical circle even more interesting is that the aesthetic theory voiced by Stephen Dedalus nicely captures the encounter with the visualization. Dedalus has developed an idea from the medieval philosopher Thomas Aquinas. The encounter with art, Dedalus argues, passes through three phases:
1. Integritas: First, the aesthetic image (visual, audible, or literate) is encountered in its wholeness as set off from the rest of the visible universe.
But, temporal or spatial, the esthetic image is first luminously apprehended as self-bounded and selfcontained upon the immeasurable background of space or time which is not it.15
2. Consonantia: Then one apprehends the harmony of the whole and the parts.
(p.88) Then said Stephen, you pass from point to point, led by its formal lines; you apprehend it as balanced part against part within its limits; you feel the rhythm of its structure. In other words the synthesis of immediate perception is followed by the analysis of apprehension.16
Dedalus’ argument echoes the process of reading the visual plot in Smith’s “Image and Imagery”; following the rhythm of lines, you pass from point to point.
3. Claritas: And finally one sees the work in its particular radiance as a unique thing, bounded from the rest but with harmonious parts. To quote Dedalus on claritas, and uses of the basket of a butcher’s boy nearby.
When you have apprehended that basket as one thing and have then analysed it according to its form and apprehended it as a thing you make the only synthesis which is logically and aesthetically permissible. You see that it is that thing which it is and no other thing.17
These three phases or this aesthetic encounter make a pattern that we can use for the interpretive encounter. The sequence might be something like the following steps, the first two having been outlined in chapter 2 and the third covering how results are returned to you (the subject of this chapter).
First, one has to identify and demarcate the text (or corpus) one is going to interpret. In that choice one is defining the boundary of a work or collection from the rest for the purposes of interpretation. Pragmatically this takes the form of choosing a digital text and preparing it for study.
Second, one breaks the whole into parts and studies how those parts contribute to the form of the whole. Smith, in “Computer Criticism,” suggests that this is where the computer can assist in interpretation. Stephen Dedalus summarizes this as analyzing “according to its form.”18 The computer can automate the analysis of a text into smaller units (tokens). Even individual words can be identified and be used as the unit for analysis.
Third, there is a synthesis of the analytical insights and interpretive moves into a new form that attempts to explain the particular art of the work being studied. That is the synthesis into visualizations and then into an essay; a new work that attempts to clarify the interpreted work.19 These syntheses are hermeneutica.
When we return later to theorize computer-assisted interpretation, we will look more closely at Smith’s structuralist critical theory, the weakest (p.89) part of his triptych of illustrative essay, theory, and tools. Here we will close this review of “Image and Imagery” by recapitulating the aesthetic theory that frames the essay.
Stephen Dedalus provides a theory of interpretation drawn from literature and applied to computer-assisted interpretation. In the circular spirit of a hermeneutic that draws from the work a guide to the interpretation of that very work, Dedalus’ theory describes how we might approach a text with, or without, a computer to assist in the analysis. That aesthetic theory grounds Smith’s rhetorical approach to deploying computer criticism.
In the past few decades, numerous computing humanists have likewise used visualizations in their arguments. Etienne Brunet’s 1989 paper “L’exploitation des grand corpus: Le bestiaire de la littérature française” is a rich and early example illustrating what can be done with visualization when applied to a larger corpus. Brunet’s argument is still current. He argues for analyzing a large corpus (a database of French literature) with multi-variant techniques that generate visualizations.20
Brunet wrote his article for a Humanities Computing audience. In it he doesn’t focus on discussing the implications of the “bestiary of French literature”; rather, he uses the analysis of animal words to demonstrate the effectiveness of statistical techniques and large-scale analysis in generating composite visualizations.21 Nonetheless, he has to face the problem of how to explain the statistical techniques behind what he wants to show visually. Using a series of funny puns on cooking, he opts for the rhetorical power of the visual. He wants to
Count words rather than weigh them! To count beasts rather than caress them! … To reassure Colette [the author he will focus on] and to save her ghost from the nightmare of numbers the cooking of numbers won’t be shown. It is enough to know that the curves which we will produce are obtained with cross multiplication, square roots, and many other ingredients whose names alone could spoil your appetite, though the computer can digest them without trouble.22
What are the ingredients of Brunet’s cooking of visualizations? He provides the tables of data used. His endnotes include formulas for some of the calculations. However, in the body of the article he explicitly turns our attention to the lines (curves) of visualization.
Brunet begins by showing how statistics can confirm what most readers would consider obvious about Colette’s fondness for cats. His first graph shows which animal words Colette is statistically more likely to use than the authors of the comparison corpus of nineteenth-century and twentieth-century texts from the full database. Not surprisingly, “chat” stands out, (p.90)
but many other animals have a positive deviation from the norm. Such analysis and visualization require a large corpus of both Colette’s works (five of which are in the database) and comparable authors with which to establish the norm.
From bar graphs to scatter plots based on factor analysis, Brunet goes wild with visualizations as he looks at a corpus of comparable authors from the same period. Many of his graphs are comparative or are arranged to enable comparison with other graphs. (See figure 5.4 here.) Other graphs, such as that shown here as figure 5.5, are designed to compare the animals that, in the writings of a certain author, appear more frequently than the norm (on the right) against those that appear less frequently (on the left). In graphs such as that shown here as figure 5.6, readers are encouraged to compare authors—in this case Chateaubriand (empty circles) and Hugo (p.91)
The build-up of visualizations culminates in two scatter plots, generated by factor analysis, that map selected authors and animals into a two-dimensional space that shows how they cluster. Each of these plots takes up more than half of a page, overwhelming the narrative. The final one (figure 5.7 here) brings us back to the beginning—it shows Colette off in the lower right-hand quadrant, where she is the author closest to “chat.” (The arrows drawing attention to Colette and “chat” are in the original figure.)
To recapitulate: Brunet begins with a serious discussion of the value of large-scale analytics, then proceeds to a gallery of playful visualizations that overcome the argument, or perhaps become the argument. The second half of the article, with its twenty graphs, encourages visual comparisons. A visualization attached to the article’s last endnote at the end of the article shows the popularity of cats and dogs over time and across genres of writing. Concluding the narrative, Brunet writes: “As for dog his fate runs parallel to that of cat, and the two lines are twisted together like a garland. Visibly the cats and the dogs were created to understand each other, like Toby-Dog and Kiki-The-Sweet.”23 The article closes, not with discourse, or narrative, but with Brunet’s final visualization of cat and dog (and horse), shown here as figure 5.8.
One wonders if this is a discourse with embedded visualizations or a comic strip of visualizations connected by text. Visibly, in Brunet, the text and the graphics are woven together in more ways than one. The question is whether can we weave them together in the new online and interactive publishing medium of the Web.
But here we need to look at how interactive visualization, which has emerged as a broadly accessible form of analytic, has been used rhetorically. How have interactive analytics been integrated into texts online? How can toys be embedded in texts? In chapter 3 we looked at a number of examples. Here we will look more closely at one of the first tools that could be embedded in a webpage: TACTWeb.
TACTWeb, developed from TACT by John Bradley and Geoffrey Rockwell, was one of the first mature text-analysis environments to be available on the Web. To teach text analysis, and to teach how to use the TACTWeb, a workbook was developed that provided a narrative with embedded interactive fields and buttons.24 The example provided here in figure 5.9 shows (p.93)
how a user can learn about distribution graphs by trying TACTWeb. The workbook followed a model of general explanation, showing examples, providing narrated “Try It” panels, and then leading to the analytical environment, in which there is no explanatory text woven around the fields and buttons.25
It was the possibilities of Common Gateway Interface (CGI) server-side processing offered by HTML Form elements that, in 1996, made it possible to imagine alternative configurations of text and analytical results. On the one hand, we could now weave an analytical tool directly into text so that a reader could try it in the tutorial. On the other hand, we could also show the same functionality in different ways on different webpages, thereby erasing the lines between instructions, essays, and tools and extending the tools’ pedagogical and rhetorical uses. Because we didn’t have to clearly demarcate the interactive components from associated text on the web-page, we were able to present the same components in different explanatory (p.95)
contexts, ranging from the heavily narrated tutorial to the cleanly labeled Query screen shown here in figure 5.10.
Being able to interlace components with text and to launch interactive examples has obvious pedagogical uses. This model can be found in other online tutorials, including John Maeda’s Design By Numbers (dbn.media.mit.edu) and the tutorials available at www.w3schools.com. However, the rhetorical opportunities for this intertwining have not been exploited in interpretive work, mainly because interpretations are usually published in print and because those available online are in online journals that enforce a common format.
Examples of Tool Rhetoric
Though Smith and Brunet hoped that computer-assisted analysis would be an effective critical method when supported by visualization, their studies did not have the hoped-for effect. Smith, in “Computer Criticism,” clearly hoped that the shift in perspective brought about through computer-assisted interpretation would become a school of criticism of its own, rather than a specialized extension of structuralist/formalist theory. He hoped for a Kuhnian paradigm shift that would introduce an “altered concept of proof and what constitutes demonstration of a literary hypothesis.”26 That was not to be. Perhaps literary theory got tired of structuralism just at the moment Smith was trying to use it to argue for computer criticism, or perhaps Smith was ahead of his time. We also think Smith was too worried about proof, but that is another issue. Only now on the Web are Smith’s ideas about (p.96) interactive analytics playing out. Similarly, Brunet’s point about the value of large-scale text analysis would have to wait until the scale of text collections for other languages caught up. Only since Franco Moretti’s Graphs, Maps, Trees (2005) have we seen significant uptake by literary historians.27
Earlier in this book we reflected on the rhetorical problem of engaging a humanities audience with research assisted by text analysis. The challenges with the integration of analytics in interpretation include the following:
1. Results consisting of long lists of words often take up a lot of precious publication space (even digital space is valuable if users have to scroll a lot).
2. Results have to be explained. Because a list of words by itself doesn’t mean much, such lists are often accompanied by long technical discussions explaining how they were generated.
3. Long discussions of method distract from interpretation.
4. We do not generally write about methods in the humanities, because the proof is in the interpretation, not in how one arrived at the interpretation.
Hermeneutica presents the interactive panel or hermeneutical toy as an alternative way of weaving analytics into discourse so as to engage a larger audience. The interludes show how embedded toys can demonstrate analytical results in a way that avoids many of the rhetorical problems of reporting analyses in print. The long lists are within an embedded scrolling window, so they do not take up the same visual space as the rest of the text, but can still be explored. The tools are explained through exploration and online documentation.
Hermeneutica invite play rather than distracting from interpretation, especially when there is a responsive interface to the toy. An embeddable toy such as TAToo substitutes interaction for documentation as a way of understanding method. Rather than tell you about it, we let you try it as a way of figuring it out. You could call it fiddleware, though there are usually help screens and documentation for those who don’t like to discover technique through use. Finally, this model allows interpretive narrative and interpretive toys to be tightly interwoven, giving the author more flexibility in the pacing of an essay. You are not forced to have a whole section on technology; instead the toys can be side panels placed where you want. You can create a palimpsest of texts and interactives.
Of course, this chapter, located at the center of the book, is also another way of interpreting technique through the doubling of essays mirrored with reflective chapters such as this one. The interludes or case studies are paired not just because it is a convenient way of moving technical reflection out of (p.97) interpretation but also because the doubling reflects the thinking through of Agile Hermeneutics, in which we worked as a pair and paired programming with hermeneutics. All this doubling may seem cute, but it is in the game of trying technique through real questions that you find more to reflect on than just the question at hand—a reflective surplus about technique. Just as we worked in a pair with our toys, this book, with its combination of interludes and chapters, emerged from a series of experiments trying to do text analysis as a process from playing with toys to producing an essay.28 In experimenting we found ourselves struggling with how to report what we thought we saw. Willard McCarty put it this way in a Humanist post:
So let me ask: how better might we talk about the research we do (and that we avoid doing) than always to be rattling on about pure vs. applied, or curiosity-motivated vs mission-orientated or whatever?29
This is a problem faced by all of us who try to explain computer-assisted text analysis: there are very few examples of concrete humanities research essays based on computer-assisted techniques that are not about the techniques per se. There are papers that respond in traditional ways to original literary, historical, or philosophical problems; however, they do not describe how computing methods modulated the response, and thus they hide their method. There are papers about tools and methods, but they don’t give us concrete examples of techniques in use. No, we are tempted to conclude that computer-assisted tools are like Wittgenstein’s ladder: they are discarded once one climbs up them to purchase a new view. We are tempted to go further and speculate that method has become invisible to the humanities, and that there is something fundamental to the rhetoric of the humanities that is antithetical to method, though not to reflection about method. You can have theory (including theory about method) or method, but not both. Perhaps the humanities are embarrassed by methods because they look like toys.
This is a problem for computer-assisted text analysis. As we develop methods and apply them to problems in the humanities, we also have to figure out how to talk about the insights that emerge without either boring our audience with technical details or hiding the methods. It is possible that we may never grasp the “holy grail” of writing works that engage our colleagues while documenting new methods used rigorously. Hermeneutica presents a model of online writing in which the results of text analysis are woven into a hybrid form that is not an academic essay or a tool, but rather is a weaving of text and toy that invites exploration. We imagine (p.98) an academic rhetoric that resembles the emerging public rhetoric—an academic rhetoric in which things such as visualizations are woven into text. The online essay can now be an interactive hybrid that can sustain two threads, showing both conclusions and the analytics used to reach them in a fashion that can be recapitulated. As Brunet imagined, the online essay can be a satura—a comedic mixture of ingredients.
What Are These Things?
This brings us back to the hermeneutical things we want to mix into our work. How are they different from the texts we want to weave them into? They obviously aren’t texts in the way a book is a text, though they may contain text. So what are they? First of all, they are software things that run within more complex systems, such as Web browsers. Things in general seem to be the dumb other for the disciplines of discourse of the humanities. Though things don’t talk the way discourses do, they seem to be real in a way that talk isn’t. Things, even software things, seem to persist longer than talk and seem to resist us the way talk doesn’t. You can interact with things, touch them or click on them, but they are not of us. For these reasons they are studied by practical disciplines such as carpentry, industrial design, engineering, or, in the case of software things, computing science.
To understand hermeneutica, therefore, we need to understand the difficult relationship the humanities have had with things. We can do that by returning to Descartes’ Discourse and his autobiographical story of doubting every thing. At the beginning of part 4 of the Discourse, Descartes tells us about doubting everything in order to see if there was anything that could not be doubted. Because the senses can deceive us, he rejected the things they present to us; he also rejected as possibly doubtful the abstract objects presented to thought in everyday life. He found reason to doubt the existence of everything, either sensual or abstract, with one exception: the thinking self that is conscious of its thinking and therefore must exist. (“I think therefore I am.”) Actually, that exception is not just a thinking self, but a self thinking in discourse. The Cogito is a thinking through of discourse that then privileges thinking discourse over other things, whether perceived or thought.
Things, in this story, will always be dependent on discourse for their truth, and it isn’t clear that the humanities have totally freed themselves from this logocentricity. Idealism, the philosophical view that things are just bundles of perceptions called ideas, has taken many forms before and since Descartes’ Discourse. Once you start doubting things, it seems obvious (p.99) that the thinking self (consciousness) is more real than, and different from, the objects it thinks about. But where do the ideas of things come from? Bishop Berkeley suggested a simple solution; that God broadcasts everything to our minds. More recently, the 1999 movie The Matrix presented an artificial reality (the Matrix) generated by artificial intelligences to pacify humans.
Perhaps the most influential response to Descartes’ doubt came from Immanuel Kant, who argued that reality is how we structure experience. Whatever the state of things-in-themselves, we wouldn’t have any experience without structuring it with categories of time, space, and so on. For Kant there were things-in-themselves, but those weren’t knowable by humans except through the structuring that made thinking possible. Things were things again because otherwise there would be no way to think about them. David Hume’s more pragmatic approach was to point out that, whatever the ontological status of things, one still must act as if they are there, and, as we will see in chapter 10, leave a room through the door, not the window. Hume’s Dialogues Concerning Natural Religion shows in dialogue how common sense about things is consistent with doubting skepticism, especially if you are skeptical about skepticism. Recently, Ian Bogost (2012) has taken skepticism of the importance of the human perspective even further by asking what it would be like to remove ourselves entirely from the understanding of things. He points out how the humanities are still blindly committed to our own discourse and thus blind to the way things can bear knowledge.
Part of the issue with theorizing things is setting things against human knowledge. As Bogost points out in Alien Phenomenology, all these discussions of the being of things depend on us and on our knowing of the thing. If you start by questioning how we know about beings, then all being tends to end up being defined by our thinking. You then end up with a separation between mind and things (e.g., bodies), between subject and object, between mind and nature, or between text and toy. These fundamental categories then structure our thinking in all sorts of ways, including an artificial division of the disciplines of the mind and its discourses in opposition to the disciplines of things. The Cogito, a brilliant idea, becomes a tool that makes sense of some things but fails to make sense of other things.
Martin Heidegger, in Being and Time (1953), questioned the Cartesian assumptions and restored thingness to the thing. Heidegger flipped the philosophical fascination with the question of the reality of the thing-in-itself to argue that it was actually as everyday stuff that we experience things. In living we encounter things as ready-to-hand tools that we use (p.100) to achieve our various desires and purposes. A tool such as a hammer is a paradigmatic thing of everyday life. It is equipment, and as equipment it doesn’t stand apart from us or alone in itself. In fact, in the hammering of a project there is no subject or object, just the carpentry. Tools and equipment, including everyday language, are transparent in use, which is why all the discussion about doubting minds and things-in-themselves doesn’t tell us about things so much as it tells us about philosophical minds.
Further, any piece of equipment is part of a system or network of equipment. A hammer goes with nails, wood, hardware stores, and weekend projects such as building a bookcase for all the stuff you read. We don’t notice a hammer as a mere thing until is breaks, or until we stand back from it and begin doing ontology. Then it becomes present-at-hand—that is, becomes the sort of thing, present before us, that is deliberately thought about. Only when it is present-at-hand does it stand apart such that we could think about doubting it as a philosopher would, or studying it as an engineer would.
We can see now how hermeneutica are things. When used in a news website, they are transparent in the way a telescope is transparent in use. As things for research, however, they are meant to be seen and noticed as things during their use. Research instruments have to be fiddled with in order to work, and in the fiddling they can draw attention to themselves. They are thus meant to resist completely transparent use and to be alternatively used through manipulation and manipulated to be useful. Research things thus are meant to be both ready-to-hand and present-at-hand so as to avoid uninterpreted use. They are meant to be a little out of place, as is a toy in a serious essay. One might even go so far as to suggest that in their playfulness they undermine other things that are around, but only when they work—and often they don’t work.
Another way in which hermeneutica are things is that they are part of systems of things, such as the World Wide Web and all the associated technologies. They may be toys, but they are toys in a network of technologies that includes everything down to the level of tubes through which data flows. Hermeneutica, however, have some special features as things for interpretation. The system of things they are part of includes the original text, other essays of interpretation, and the other things of interpretation. Hermeneutica don’t hide these things; they are designed to show them. Further, hermeneutica are meant to be embedded in specific interpretive (p.101) things, such as online essays. They are not meant to stand alone or to the side. They are not interpretations that replace the interpreted, but toys that augment the interpretation, though sometimes they interrupt the flow of thought.
In a later essay, “The Thing” (1971), Heidegger plays with the etymology of the Old High German word “thing” and suggests that a thing is a “gathering.” Bruno Latour (2000) picks up on this suggestion of how a thing gathers different expectations and perspectives. The same thing that is developed and debugged by a software weaver is used by a student to interpret a text and then is embedded in an essay for use by another reader. It is used transparently by someone browsing the Web, and it is fiddled with by someone interested in the toy and how it ended up in an essay. No one way of approaching hermeneutica as things is more real than another. Hermeneutica as things are a gathering of expectations in networks of use.
The second thing to say about hermeneutica is that they are follies—not in sense of foolishness, but in the architectural sense of decorative buildings not designed to be used so much as viewed. The faux Roman ruins and Medieval castles with which certain gardens were decorated were architectural follies. They were designed to remind people of another time or another place. They weren’t simulacra, meant to replace the originals; they weren’t simulations, meant to represent the originals faithfully. They were designed so that we would know that they aren’t the originals, but that they point through to an interpreted original.
You can see another way in which hermeneutica are interpretive. An interpretive thing has a special relationship to other things in that it announces itself not as something brand new, or as a replacement, but as an interpretation of something else. An interpretive thing is like a folly that is not a new building, even if it is built new, but a building that is meant to look as if it had been assembled from the parts of the old—a small ornamental building drawing attention, but part of a garden for which it is an ornament. In this new webbed world, in which we are surrounded by neat gadgets and new interactive games, some of them present themselves as being generated from other things. Those are hermeneutica.
Why use ornaments to draw attention to something in the past? The folly seems purposely anachronistic, something that runs counter to the spirit of the new-new modernity of computing. Of course, re-searching what has come before and re-interpreting it is the way the interpretive (p.102) humanities bear gifts. By ornamenting, framing, and curating the works of human expression, we animate them so that they can be relevant again and bear them forward for further use. By ornamenting them and providing new ways into interpreting them, we bring them to life for another generation. We try to make them infectious again by trying to bring forward that which is uniquely human in each work interpreted. Whether we are doing this with students in a class or in writing for a public, we are treating things as special, as if they mattered again today as they did in their time. Our interpretations may be follies, but they forge a way back to something idiosyncratic. Our hermeneutica may appear to be fiddly little toys in the serious work of the humanities, but they too point back to the ideographic.
The third thing to be said about hermeneutica is that they operate on surrogates. This is true of all computer-assisted text-analysis tools, though it is true in different ways for different tools. What do we mean by “surrogates”? As we mentioned above, the first steps in analyzing text are demarcating, digitizing, and then encoding the text to be studied. We are never operating on texts directly, as might be thought. With computers we are developing stand-ins or surrogates that can be processed as strings. This is doubly true of some of the work done by Franco Moretti and Matt Jockers: they don’t always operate on the full texts; sometimes they operate on metadata about the texts. Moretti, in “Network Theory, Plot Analysis” (2013b), makes this point about working with metadata: “once you make a network of a play, you stop working on the play proper, and work on a model instead. You reduce the text to characters and interactions, abstract them from everything else.” J. B. Smith illustrated this in his 1973 paper “Image and Imagery in Joyce’s Portrait” and made it the central insight of his 1978 paper “Computer Criticism.” The computer can be used to identify different types of structure in a text and to create a surrogate for visualization. By breaking Joyce’s text into 500-word chunks and measuring the density of image words in those chunks, he created a new “layer” or surrogate that showed something he wanted to study in the text. That surrogate was then represented by another surrogate: the graph that shows the visualization. What Smith realized in “Computer Criticism” is how many different ways we can use the computer to structure layers and how these layers can then be compared. Where we differ from Smith is that we don’t think of these layers as representing something “in” the text so much as we think of them as interpretations of the text in layers of interpretations over time. We see (p.103) hermeneutica not as microscopes revealing the inner structure, but as augmentations adding to a history of interpretation.
Building on Smith’s work, we contend that working with computers means working with computer surrogates for the things commonly thought of as texts. The materiality of computers is important to understanding what can be done with them. Some of these digital surrogates are corpora, some are electronic texts, some are metadata (data about data), and some are code. Our hermeneutica are surrogates that break apart and then resynthesize the original (if there is such a thing as an original). They invite exploration as a way of understanding the original. Word clouds and other hermeneutica are small models that represent some aspect of a much larger surrogate. You work with them, not with the text they are drawn from.
That said, one of the things we have tried to do with the hermeneutica we design is provide the humanist with a way back through to the text. An “interactive” may be a model of the original that invites exploration, but it should be possible to check one’s interpretation of the model against the electronic text. This is one of the features that distinguishes the text-analysis tools of the Digital Humanities from either the bibliometric tools of informatics or distant reading. We want to move from the small visualization to the text and back because we value the poetry of the text in and of itself. Ideally our tools would be like palimpsests that scrape the text and reuse it while still showing faint remains. Stefan Sinclair has, for this reason, theorized one of Voyant’s predecessors as a reading tool—a “hyper-reading” tool, but still a reading tool that doesn’t replace so much as remind. Voyant takes this further, giving you a way to generate little readings and embed them as ornaments in other interpretations. These little interpretations don’t pretend to replace that which they interpret the way big interpretations do; they are embedded toys in the tradition of follies.
The interludes in this book take advantage of the possibilities of the new online medium by proposing a new form of hybrid publication that builds on the print graphic visualization (Smith 1973) and the 1990s idea of the digital workbook. In a hybrid essay, the narrative is no longer pedagogical; it is interpretive, with interactive panels that enable readers to participate in the analysis. It was the pedagogical effectiveness of such interactive workbooks that led to their adaption in the form of “Recipes” for the TAPoR project, which, in turn, have been adapted for Hermeneutica. The embedded toy provides an interpretive thing in an essay. Hermeneutica embed analysis in interpretation, and they explain themselves by letting you try them.
But how do you feel about toys in your essay? Do you notice them because they are new? Are they Trojan horses bearing the virus of quantitative (p.104) methods, or can they provide new ornaments to interpretation? Do they demean the noble essay with technology, or do they re-interpret the essay as interactive? Will they corrupt the pure text of interpretation, or will they re-interpret computing for interpretation? We are reminded of the end of Lucretius’ The Way Things Are, where he leaves us with an image like that of Pontypool—an image of language, the first tool, as a plague on thinking:
- the tongue, that tries
- To be the mind’s interpreter, filled up
- Engorged with blood, became too hard to move,
- Seemed rough to the touch. From throat to chest and lungs
- The plague descended, thence assailed the heart,
- Battering all the bastions of life.
(10.) If one wanted to recapitulate his precise imagery plot, one would need access to this axiomatic list, but it isn’t provided and we don’t have much to go on other than the number of words.
(12.) Five hundred words is a commonly used chunk size corresponding roughly to the number of words on a page of the standard edition.
(14.) If you find yourself exploring the graph and checking your interpretation of it against Smith’s, you have already been sucked into the methodological assumptions giving them tacit assent.
(p.215) (19.) In “The Visual Concordance: The Design of Eye-ConTact” (1998), Rockwell argues that visualizations function like concordances that bring into a new synthesis a collection of passages broken out from the original.
(20.) The FRANTEXT database of the Trésor de la langue française is available in North America through ARTFL, http://artfl-project.uchicago.edu. Brunet discusses the advantage of using the Stella software that could be accessed over a terminal to analyze FRANTEXT. This article documents an important early example of a networked large-scale text database accessible through a suite of interactive text-analysis tools. There was probably nothing like it in the world at the time.
(21.) “The testimony of statistics isn’t solid except in large numbers. Its value is collective, like that of aggregated visualizations (portraits-robots), and not individual, like that of the finger-print.” (Brunet 1989, p. 121)
(23.) Toby-Dog and Kiki-the-Sweet are the interlocutors in Colette’s Dialogue of the Animals: The Cat and the Dog.
(27.) One can also see the NEH-SSHRC-JISC Digging into Data Challenge of 2009 as a form of validation of Brunet’s point about scale.
(28.) For more on these experiments, and for some of the raw materials, see http://tada.mcmaster.ca/Main/ExperimentsInTextAnalysis.