Katherine Parrish

Years ago, I began to play around with a simple but effective activity where students were to illustrate some of the metaphors and similes in Macbeth. I’d been thinking about the fact that a lot of struggling readers don’t habitually, actively imagine or visualize what they are reading, even in fiction. I was also embarking on teaching Michael Ondaatje’s In the Skin of a Lion for the first time, and was thinking of ways to help students track the images and motifs, through which so much of the story is told. I came up with “Picture This.”

It’s a group co-operative/competitive challenge, and an individual assignment. Students are given a line/sentence/short passage from the particular chunk of text we are reading at the time. They are to draw it with understanding, accuracy, and attention to detail. (Stick figures are welcomed and encouraged.) The next day, everyone puts their drawings on the board, with only their group names on the visible side of the page. The groups then have to identify the exact lines/sentence/short passage being illustrated by any group but their own, and claim it with a sticky note copying out that passage. If the claim is correct, the illustrator’s group and the identifier’s group each get a point. (points mean prizes)

I introduced this activity early on this year, given the weak skills and reluctant participation I’d been noticing from these post-pandemic students. They’d just read Lillana Heker’s “The Stolen Party,” and were showing themselves to be a little fuzzy on the details. I assigned “Picture This” for homework, and then, in a calculated whim, invited them to also use an AI image generator for bonus points for their group, and to e-mail me their results.

Only one student took me up on this. The results were fascinating. She used this passage as the prompt: “The mother swung around to take a good look at her child, and pompously put her hands on her hips.”

Her first result showed a Latinx woman, who didn’t so much look pompous as constipated, in a market that sold chillies and woven blankets. But at no time did my student give the AI any indication that the story was set in Latin America. She didn’t give any information about the story- just fed it the passage.

She used Ideogram, whose development team is based in Toronto. I figured that the AI must have known the story, and, recognizing the passage, predicted (reverse predicted?) the setting.

So I tried it myself, using a few different text to image generators.

I want to experiment more with this.