An art collage that measures public sentiment

Aleta Michaletos was interviewed on the Africa News Network last night about her collages and their unforeseen connection with Nelson Mandela. Michaletos began making the collages in Pretoria, SA in 1989.  Mandela was still in prison, and politics were tenuous. Clashes between supporters of the ANC and the National Party erupted in anonymous bombings throughout the country.

One of many fearful South Africans, Michaletos decided to address her rising cynicism by culling through daily newspapers and clipping out the positive words. She didn’t have an artistic goal in mind specifically at the time, but she began painstakingly fitting the words together into collages as part of a project she called Precious Circle.


According to Michaletos, it took six months to collect enough positive words for a single collage.  Flash forward to May 10, 1994, the day Mandela was inaugurated as the president of South Africa. The emotional tone in the media shifted drastically and immediately, and  Michaletos was able to make each new collage in a matter of days.

What’s cool  is that whether she knew it or not, by creating a rubric of specific positive words and measuring the frequency of their occurrence in the media over time, Michaletos was conducting a sociological study. You could analyze her collages in a bunch of different cross-sections: which types of articles contain the most and least positive words, (surprised to know that Sports contained the most positive words?), whether certain words trend, remain consistent over time, at what point they become buzzwords and how they spread (fiscal cliff, occupy, entitlements), and perhaps most awesomely, how Nelson Mandela’s release from prison caused mass euphoria in the media so that Michaletos could churn out a collage like it was nothing.


More than two decades later, automatic sentiment and subject analysis programs abound. If you have a million customer reviews about a vacuum cleaner and you want to know how many people said it sucked, or if you want to know whether Kim Kardashian’s popularity on Twitter is increasing or declining, you can either purchase usage of an existing API, or you can hire a pot-smoking engineer just out of college to build it for you in a week. This blog put together a list of 20 of them, each designed with specific sentiment analysis projects in mind, many of them geared toward harnessing the power of social media’s big data.

Or, if you’re artistically inclined, you can short out your printer and use scissors and paste to do it like Michaletos did in 1989.

Michaletos’ work is currently on display at the Pretoria Art Museum.


Photo credits: Aleta Michaletos

Leave a Reply