Unbreathed words: A Conversation with Lillian-Yvonne Bertram

In lillian-yvonne bertram’s Travesty Generator, their most recent book of poems, Afropessimistic thought finds expression through the brutal and precise procedures that take place on the inscriptive surface of silicon. Nearly all the poems in Travesty Generator have been composed in part by repurposed and modified computer programs, programs that were designed to splice and recombine language, often randomly, to forge new language, new poetry.

Bertram takes Travesty Generator’s title from “A Travesty Generator for Micros,” a technical essay by Hugh Kenner and Joseph O’Rourke originally published in the November 1984 issue of BYTE, a magazine for personal computer enthusiasts.¹ Comparing Kenner and O’Rourke’s description of TRAVESTY, their program for generating language by stochastically recombining letter sequences within a source text, with Bertram’s repurposing of this program helps to clarify what for Bertram are the stakes of manipulating text with computers. Kenner and O’Rourke seem most pleased with their program when it is most clever- seeming—when it invents almost-passable faux-James Joyce or humorous but recognizable names of English towns. Bertram’s poem “Incident” runs the TRAVESTY program on their own poem, adjusting the program’s parameters so that the text that it outputs is often nearly nonsensical, a jumble of almost-words. The notion of a text “generator” here is ironic; the algorithm is far better at destroying—scrambling carefully composed human language into an entropic slurry—than it is at forging anything new. But, since Bertram’s original poem recounts a moment of racial insult, is the violence wreaked on it by the TRAVESTY program an odd kind of mercy? If the signal that the world transmits is an oppressive one, is it better to drown this signal in noise?

Or are the capricious workings of the algorithm analogous to the capricious violence that Black life has faced and continues to face? In the cases of other poems in Travesty Generator, programs for randomly assembling text from a bank of lines have generated meditations on, or rehearsals of, the deaths of Trayvon Martin and Emmett Till. These programs can’t turn Bertram’s lines into something new, only something different; electricity is consumed, calculations are performed, memory is allocated, bits are flipped, lines are shuffled and reshuffled— but no epiphany is reached, no violence understood or avoided. We are living in a moment of great enthusiasm for the power of algorithms, including in the arts. Travesty Generator stands out from many contemporary examples of computer-generated poetry in the way that it pushes code toward moments of failure, confusion, and loss—moments that, in light of the book’s subject matter, become allegorical: racial violence isn’t a bug but a feature.

At the same time, Bertram’s poetic practice is open to the ways that algorithmic media can indeed work in the service of the endurance of Black life and thought. Their web-based project Forever Gwen Brooks generates variations on a poem by Gwendolyn Brooks.² Unlike Kenner and O’Rourke’s TRAVESTY program, Forever Gwen Brooks is not a general-purpose program for textual mimicry; rather, the code emerges from a deep understanding of and love for the poem it modifies. It is this sense of care that makes Forever Gwen Brooks (as Bertram describes it in a blog at the Poetry Foundation) both a way of constantly refreshing one’s remembrance of Brooks as well as a traversal through “the space of the could-have-been.”³

During the late fall and early winter of 2021, I had the opportunity to converse with Bertram over email about how they think about inviting algorithms into their poetic process.

—Kyle Booten

KYLE BOOTEN/

I’d like to begin by asking you to tell the story of how you came to integrate codework into your practice as a writer. You have published four full-length collections of poetry as well as several chapbooks and an artist book. What drew you to the experiments with code that we see in Travesty Generator (and a...