Empowering Play, Experimenting with Poems: Disciplinary Values and Visualization Development

At DH2013 we presented our initial approach to poetry visualization, which undertakes to support active, spontaneous, and unique close reading experiences by emphasizing temporality and reader-engagement in exploring over twenty dimensions of poetic sound (Abdul-Rahman et al., 2013a; Abdul-Rahman et al., 2013b). PoemViewer, the software we discussed last year, makes important advances toward that aim; some literary colleagues already proficient with digital tools have noted how they might use that program in their own research and classrooms. But many still wonder, Why not just read the poems? Sometimes that question indicates a simple lack of digital literacy or a defensive response to a perceived technological threat. But it also points to ongoing tensions between core values and practices in the humanities (like close reading) and those implicit in the engineering that makes digital tools (like data visualization) possible.

People will always read and write poems without computers. Still, poets like Nick Montfort and Stephanie Strickland compellingly bridge literary composition and computational techniques. And DH research has powerfully aided traditional humanities scholarship in archival work, mapping, etc., that would also have continued, if differently, without computational tools. We too strive to design software that can support existing humanities practices while also suggesting new ones—a dual goal we believe is served by honoring different disciplinary values: how might we in practice value multiplicity and openness as well as exactitude and decision, for instance? Rather than endeavoring to quell such tensions, we argue they should be embraced and productively leveraged through play—not only in the capabilities a tool can ultimately deliver, but also in the conception and experimentation that inform each stage of its development.

This poster will share images and anecdotes from the technology probes—what we’re calling ‘experiments’—the rapid prototyping informing and guiding the design of our new close reading tool, Poemage. We will explain how these first development stages already begin to answer the call from DH scholars like Johanna Drucker and Stephen Ramsay for digital tools more imbued with and suited to the humanities—without sacrificing scientific meticulousness and rigor. Like PoemViewer, Poemage will invite users to visualize poems of their own choosing, rather than rely on existing visualizations of poems chosen and manually coded by others. By using automated techniques to reveal various nuanced kinds and patterns of sonic ambiguity (or play) within poems, Poemage promises to advance the fields of computer science, literary study, and digital humanities. By embracing play in its development and prioritizing play as a condition of use, it seeks to empower a greater range of scholars and readers actually to engage and benefit from those advancements.

Poems are not reducible to their constituent pieces. Sound affects tone; lineation expands and turns syntax; meaning is multiple, building on ambiguities and multiplicities of diction and rhyme and so forth—and each reader adds to the complexity already inherent in the words and their arrangement every time she reads a poem. Moreover, all of this happens through many dimensions and in multidirectional time, a factor usually overlooked by other poetry visualization software but crucial to our efforts. We believe that poems are not only animated but created and recreated through each act of reading, and therefore that our tools should foreground this act and the relationship that develops between the reader and the poem. Rather than simplifying poetic complexity or stifling ambiguity, then, we accentuate and explore them. This approach to poetry visualization does not save a reader’s time, but we do hope our approach enriches time spent. We want users’ interaction with our software and visualizations to continually lead them to new ways in and through a given poem. We want them to experience the pleasurable mental focus and the intellectual and emotional expansion that happen through play: the play that inheres in particular poems and the opportunities for play that emerge as the reader explores a poem with our software.

These goals require Poemage to offer an unprecedented degree of fluidity, precision, and responsiveness. In the belief that these complementary values are shared by literary and computer science disciplines alike, our team is endeavoring to build them into Poemage by layers and degrees via strategies like rapid prototyping, user-centered design, and agile programming. To that end, we are experimenting with different algorithms for detecting and classifying sonic elements in a poem, as well as quantifying the uncertainty in these computational results, all within the model of play. In our work, much as in close reading, play entails testing the tool in multiple ways through query and response and through continually looking at and through new approaches, even when existing approaches may seem to yield satisfactory answers, with an emphasis as much on experience and surprise as on hypothesis and (inherently uncertain) results.

Our poster will share visualizations and screenshots and discuss what we have learned both from our experiments and from some of the practical challenges we have faced so far. For example, our efforts to capture patterns of morphing sonic clusters (like the e/s/t group recurring in different orders and in different syllables in Louise Bogan’s [1968] poem “Night”—“estuaries,” “restless,” “inlets,” “set,” “itself,” “reflects,” “firmament’s,” and “setting”) via data sets and adaptable, customizable queries have raised questions about how much preprocessing is possible and desirable, as well as how much complexity to expose to the user. How do we balance support for open and spontaneous exploration with the computational limitations of computer memory, real-time processing speed, and algorithmic definitiveness? One way is to offer multiple exploratory strategies. Poemage will show readers the set of words in a poem participating in a particular aspect of traditional rhyme, such as a specific vowel recurrence (see Fig.1). We are also exploring how to show “sets of sets,” displaying for instance every distinct assonance pattern (Fig. 2), or the variety of sonic patterns overlapping in a chosen word (Fig. 3). We are also developing a character cluster search, unique to Poemage, which will reveal sets of words containing letter units of varying length in any order (Fig. 4)—but whose algorithm as yet does not stretch to reach past intervening phonemes. Such gaps, though, invite reader participation to extend or reframe computational results.

Interesting computation limitations also expose other kinds of uncertainty and ambiguity for the literary scholar to consider. For example, an early prototype assumed the pronunciation of “wind” in Bogan’s line, “The restless wind of the inlets,” was the verb “to wind.” It therefore linked that long “i” with words like “tide,” “lights, and “behind” elsewhere in Bogan’s poem (see Fig. 5). Results like this might be taken for meaningless, undesirable noise—except that “Night” enacts and meditates on various kinds of movement and so the notion of winding (of certain movements storing potential energies for other, differently expressed movements) is actually highly relevant, even though a reader might not have noticed it without a prompt. Poets are attracted to noise; part of their work is to render noise meaningful and aesthetically pleasing, and we hope in our visualizations not to eliminate but to capture and highlight just these kinds of uncertainty, which enhance rather than detract from the reading experience.

As their different names might suggest, Poemage—even more than PoemViewer—emphasizes the interpretive experience, oriented toward play rather than proof. Relying on but moving beyond the transformational power of PoemViewer’s “magic lens” that can reveal previously undetected but nevertheless unhidden poetic features useful for close reading (including the “surface reading” described by Best and Marcus [2009]), Poemage enhances the imaginative conjuring that can happen in readers’ interaction with poetic language and visualizations. It shares Drucker’s “performative materiality” ethos and works to support the critical and computational practices she outlined in DHQ(2013): “In place of transparency and clarity, [. . .] foreground[ing] ambiguity and uncertainty, unresolvable multiplicities in place of singularities and certainties. Sustained interpretive engagement, not efficient completion of tasks, [is] the desired outcome.” For, as Stephen Ramsay (2011) puts it:

Literary critical interpretation is not just a qualitative matter; it is also an insistently subjective mannerof engagement. . . . [C]onclusions are evaluated not in terms of what propositions the data allows, but in terms of the nature and depth of the discussions [—and we might add, experiences—] that result. . . . We are not trying to solve Woolf. We are trying to ensure that discussion ofThe Wavescontinues. (8-9, emphasis added; 15)

This is completely consistent with our efforts: we do not wish to “solve” the poem; if anything, we wish to open it to further and deeper questioning, to provide opportunities for continuous unsolving. It’s what our team is working toward both literally and metaphorically, as we seek to design poetry visualization empowering play.

Captions

Fig. 1: Rhyme set of words in “Night” containing “eh” assonance.

Fig. 2: Set of assonance rhyme sets in “Night.”

Fig. 3: Set of assonance, consonance, and slant rhyme sets overlapping in the word “estuaries” in Bogan’s poem.

Fig. 4: The E-S-T character cluster set, revealing each instance of these letters appearing together in any order and in any syllable within Bogan’s poem. Users can also explore other sets of clusters of varying length, including examples of anagram.

Fig. 5: Rhyme set of words in “Night” containing “ay” assonance.

References
Abdul-Rahman, A., K. Coles, J. Lein, M. Wynne. (2013a). Freedom and Flow: A New Approach to Visualizing Poetry. Paper presented at Digital Humanities 2013, University of Nebraska-Lincoln. July 2013.

Abdul-Rahman, A., J. Lein, K. Coles, E. Maguire, M. Meyer, M. Wynne, A.E. Trefethen, C. Johnson, and M. Chen. (2013b). Rule-based Visual Mappings—with a Case Study on Poetry Visualization. Computer Graphics Forum, 32: 381–390. doi: 10.1111/cgf.12125

Best, S. and S. Marcus.(2009). Surface Reading: An Introduction. Representations, 108: 1, pp. 1-21.

Bogan, L. (1968). Night. The Blue Estuaries: poems 1923-1968. New York: Farrar, Straus, and Giroux.

Drucker, J. (2013). Performative Materiality and Theoretical Approaches to Interface. Digital Humanities Quarterly 7:1 n. pag. www.digitalhumanities.org/dhq/vol/7/1/000143/000143.html (Accessed 26 October 2013).

Montfort, N. Computational Poems. nickm.com/poems (accessed 7 Feb 2014).

Ramsay, S. (2011). Reading Machines: Toward an Algorithmic Criticism. Chicago: University of Illinois Press.

Strickland, S. Stephanie Strickland. www.stephaniestrickland.com (accessed 7 Feb 2014).