For University of Pennsylvania senior Konhee Chang, Kandinsky’s “Circles in a Circle” painting sings. Or at least it will soon.

Chang is about halfway through a project to turn the famous painter’s work into a musical score. Each of the 50 or so circles and lines will be assigned a musical note according to Kandinsky’s theories on sound and image, and the ultimate tune will depend on which order the viewer chooses to “play” those notes, similarly to how a painter selects the sequence to paint.

“I’m interested in this relationship between seeing and hearing,” Chang says. “I wanted to look at paintings and imagine a sound that could be a musical representation.”

Chang, a double major in visual studies and economics, began the work during a summer fellowship with the Price Lab for Digital Humanities, one of three such students. The other two, Rebecca Heilweil and Izzy Korostoff, studied media coverage of free speech and the design of Philadelphia streets in the 20th century, respectively. The trio made up the inaugural class of undergraduate Price Lab fellows.

“The independent projects this summer went encouragingly well,” says James English, the John Welsh Centennial Professor of English in the School of Arts & Sciences and the Price Lab’s faculty director. “The fellowship program is a success so far, but it’s still early days.” 

Unlike the research support through the center, which matches undergraduates with faculty already doing digital humanities work, these summer fellowships provide more autonomy for students with their own project ideas in mind. Price paired Heilweil, Korostoff and Chang each with an advisor and offered financial support, then set them up with a team from the Penn library and SAS computing.

“We start them off with a meeting … to go through the system they’re going to use to gather data and organize it, where they’re going to store it, what kind of encoding, how they will ensure the work they do doesn’t disappear behind a forgotten password,” English says. “There are certain basic practices that people need to follow if they’re going to get involved in a data-gathering digital project.”

And these in particular required a large amount of data collection.

Heilweil, a junior who studies history and political science, wanted to delve into how the notion of free speech has changed in the past five decades, focusing specifically on media coverage dips and peaks.

“In the 1960s, the free speech movement is a far-left movement, at least a young leftist movement on college campuses,” she says. “Today when people talk about free speech, you normally expect that coming from conservatives or traditionalists or constitutionalists, which is the exact opposite image of the person you’d imagine in the 1960s.”

She started by testing key phrases like “free press” and “press freedom” to text-mine a database of New York Times articles. As her results appeared, she could better hone the phrases, paying careful attention to those like “First Amendment” that could produce false positives like articles about religious freedoms, not free speech. Heilweil says she read through about 1,000 documents during that trial phase. Once comfortable that she had the best prediction process and keywords in place, and with guidance from her advisor Marc Meredith in the Political Science department, she scaled up, increasing the number of documents she searched.

“We know this works 90 percent of the time with 1,000 articles. If we put in 20,000 or 30,000 articles, we can maybe assume that 80 to 90 percent of those are going to be right enough to give us a general sense of the history,” Heilweil says.

Summer’s end did not signify the finale of this project. Though Heilweil has made some observations, that high points of free speech coverage historically likely do not line up with important free speech jurisprudence in the United States, for instance, she has continued the research and analysis as an independent study to better understand why media companies make the decisions they do. It may become her senior thesis topic.

Korostoff has similar ideas about his project, envisioning the final product as an Urban Studies senior thesis.

The junior from Spring Mills, Pa., dedicated his summer to what he describes as a “digital investigation of hierarchies of streets in Philadelphia in the early 20th century,” he says. “As I worked through this project, it became evident that an architectural hierarchy of street types creates a distinct racial and socioeconomic hierarchy in turn.”

To draw this conclusion, Korostoff gathered three large datasets: architectural details from fire insurance maps, census data and appraised house values. Using his experience working with PennDesign professor Amy Hillier on The Ward Project, which reconstructs the 1896 survey of downtown Philadelphia taken by W.E.B. Du Bois, Korostoff set out to explain why well into the 20th century, Philly appeared to remain less segregated than similar cities. The answer seemed to come down to scales of measuring integration.

“Philadelphia has extraordinarily large blocks. They came to be cut up into smaller streets early in the city’s history, which produced interesting courts and alleyways. These irregular spaces have been looked down upon in the city’s history,” he says. “I made the case that they’re a valuable form of organic growth, and that this actually provided the city with the diversity of urban form that let a high level of racial integration persist.”

All ambitious projects from undergraduates without digital humanities backgrounds. But that’s part of the Price Lab’s aim: to provide new opportunities for such students.

“There’s been an explosion in the amount of data,” says Meredith, Heilweil’s advisor. “Students have capabilities to generate data that no one else had seen before and use it in a way that makes meaningful contributions to research.”

Which, in Chang’s case, means turning beautiful art into beautiful music.