May 21, 2016KR BlogBlogCurrent EventsEnthusiamsEthicsReadingWriting

The joy that has no stem nor core

The joy that has no stem nor core,
Nor seed that we can sow,
Is edible to longing,
But ablative to show.

By fundamental palates
Those products are preferred
Impregnable to transit
And patented by pod.

—Emily Dickinson


My previous post sent me down the rabbit hole of data visualization and poetry. Those two terms certainly don’t seem to belong together—we don’t like to think of poems as “data,” reducing poems to their parts or cooling creative energy to sets and points—but watching my students mapping and charting their poetic learning, and hearing the discussions coming out of that process of representation, got me wondering who else is out there engaged in that seemingly anathemic work with poetry.

Perhaps some of this interest is bubbling over from finding myself working on a campus so deeply identified with the STEM fields; perhaps some of this interest is in trying to revive my own somewhat dormant visual art practice and bringing it into conversation with my poetry (though mixed media, interdisciplinary collaboration, or illustration are slightly different conversations from visualization, I suppose). I was also blown away by the seeming ease with which my students created slideshows, videos, websites, and interactive online timelines and maps for their final poetry projects. I didn’t teach them any of these skills, but there they were at their fingertips as tools for critical synthesis. I don’t see myself catching up with my students’ generational technical aptitude anytime soon, but as someone who believes in the possibilities of different learning styles yielding, well, different learning, they’ve certainly got my attention.

If one considers poetry “hands-on,” a physical process (and if we paraphrase Dickinson and believe that a great poem is one that has the capacity to take the top of our head off, then we should probably be wearing hard hats or safety goggles when handling poetry), one could argue that the fairly recent pedagogical spotlight on “critical making” is simply tapping into a kind of process-focused thinking-through-making that writing workshops have engaged in for decades—and heuristic approaches in general have a far longer history than that.


Here are some trinkets from the rabbit hole. First, here’s Gertrude Stein, celebrating the diagram:

I really do not know that anything has ever been more exciting than diagramming sentences. I suppose other things may be more exciting to others when they are at school but to me undoubtedly when I was at school the really completely exciting thing was diagramming sentences and that has been to me ever since the one thing that has been completely exciting and completely completing. I like the feeling the everlasting feeling of sentences as they diagram themselves. In that way one is completely possessing something and incidentally one’s self.

—Gertrude Stein, “Poetry and Grammar” (1935)


I came across research far beyond my STEM skill set in “Rule-based Visual Mappings – with a Case Study on Poetry Visualization.” The paper’s authors introduce their study by writing:

A poem is a complex dynamic system. It features a variety of structural and relational information, including formal information (e.g., lines, stanzas), phonetic information (e.g., meter, intonation, timing), and semantic information (e.g., genres, words, repetition, sentiment). In addition, poems are studied from many different angles. Specific lenses or contexts for analysis might include the body of a poet’s work, a historical period, a nation or geographic location, a group or movement, and so on. Although a poem is not a big data set, poets frequently devote hours to a close reading of a poem. In many ways, a close reading is a literary form of “data ex- ploration”, in which scholars pay close attention simultaneously to various individual linguistic, literary and sociological features, as enumerated above, as well as to the interplay and relationships among these features . . .

To a computer scientist, a poem appears to be an ordered sequence of letters – and may be considered as a multivariate data set. From this point of view, it would seem that one might define a set of variables within the complex information space of a poem and then employ a multivariate visualization technique such as parallel coordi- nate plots to discover various structures. Poets, on the other hand, insist that finding and reasoning through structural and relational information in a poem is at the heart of their close reading practice. This process is complicated, time consuming, and inherently experiential; the result is that interpretations vary from one person (or one context) to another. As one of the poets in this project explained, “Close reading involves moment-by-moment choices, where every choice activates some possibilities and deactivates others” . . .

We realized our technical solution would need to accommodate two potentially conflicting requirements: it would need to address a large number of variables (poetic attributes) while providing readers with the ability to make “moment-by-moment choices” as they explored visualizations.


Herbert Tucker’s For Better For Verse is a contemporary interactive learning tool focused on the much older visualization tool of prosodic scansion:


Here’s Martin Wattenberg and Fernanda Viégas’s word tree visualization technique, with a link to their related academic paper “The Word Tree, An Interactive Visual Concordance”:


And Franco Moretti and Mark Algee Hewitt’s research collective, the Stanford Literary Lab, works with “experiments” like the Trans-Historical Poetry Project:

The goal of the Trans-Historical Poetry Project is to trace the variation of poetic form over a large corpus of English-language poetry, combining the insights of prosodic and metrical analysis with the methods of phonetics, natural language processing, and statistics. While using computational techniques for handling large corpora, we remain faithful to the aims and questions of traditional prosody: what kind of refinement, for instance, can we add to current theories of meter by being able to examine long historical series of poems? Can algorithms reliably recognize complex metrical schemes – and what patterns will emerge from the histories of those schemes? Our work on variation in line and poem length between 1500 and 1900 has already yielded results, and is now moving towards a more comprehensive analysis of poetic form that includes features such as stress, rhyme, and metrical form.


I’d love to find out about other projects or techniques in this conversation; please comment below if you have thoughts to add, though I’ll leave you (as, with Dickinson, I began) on a more skeptical note, with Adrienne Rich’s “Tonight No Poetry Will Serve”:

Saw you walking barefoot
taking a long look
at the new moon’s eyelid

later spread
sleep-fallen, naked in your dark hair
asleep but not oblivious
of the unslept unsleeping

Tonight I think
no poetry
will serve

Syntax of rendition:

verb pilots the plane
adverb modifies action

verb force-feeds noun
submerges the subject
noun is choking
verb     disgraced     goes on doing

now diagram the sentence