“Borrowing Back into the Body,” my workshop for Forms & Features Online, asked participants to use digital techniques of borrowing to find new entry points into their own creative practice. Importantly, rather than seeing appropriative or computer-based techniques as depersonalized conceptual gestures, I hoped to recenter our own bodies in this act. The following exercises (which work either sequentially or individually) use found text and generative procedures to create starting points for poetic work.

1. Freewriting in Voices

This is not quite freewriting in costume. My notion of “voice” is expansive and does not necessarily imply a speaking body. There is a voice to the manual that comes with a pressure cooker as much as there is a voice to the narrator of a novel. So, find any piece of text, and locate a few sentences that intrigue you. Don’t just copy and paste these sentences into a document or bring them near your notebook, but actually begin to rewrite the lines word by word. This initial language will provide a sort of runway at the end of which you will take off into your own freewrite inhabiting the voice of your source material. Try to write for at least 5—10 minutes without stopping. Get lost in the voice, but don’t get too caught up in the content. This exercise is more about embodying the rhythm, cadence, and tone of the language. That being said, if your voice begins to bend, break, or evolve, that’s even better!

2. Generating Your Own Voice

I find generative text algorithms are best used as something initial ideas can pass through before they are sculpted into a final product. One of the simplest and most effective techniques involves using Markov chains. These text generation tools work on a letter by letter basis and generate a probabilistic sequence based on a specific body of text. Importantly, they can quickly generate surprisingly coherent language without a particularly large source text.

Hay Kranen, a “freelance creative coder with a focus on projects in the art, heritage and media sector,” has a great online tool for exploring Markov chain text generation. I invite you to take your own language (hopefully your freewrite from above) and feed it into this tool. Simply paste your writing into the text box and hit go! The more text you input, the more interesting and varied your output should be. If your output too much resembles gibberish, make the “order” higher by one or two. If your output seems too similar to what you entered, lower the “order” by one or two.

Here’s some output I generated using “Text to Complete a Text,” by Bhanu Kapil, one of the poets we looked at in my workshop:

I came here to complete a thing I began in the rain. It was impossibly far. A beautiful night of shells and pebbles. A beautiful hazard: to go to the hospital? This is an example of departure in another time. I came here to come down in another place. It was difficult to ask my body a question about the beginning because it seemed somewhat reticent . . . no, I was. My sexual experience considers the limbs of the car a boring sentence in a dinged-up Datsun Cherry.

I encourage you to save the bits of your own output that you like and then generate more, as the results will be different each time. While these pieces could be a poem on their own, I think they are best served as a jumping off point for new work. Sculpt, delete, rearrange, and then consider adding more writing of your own.