I spent a couple of days at a techie ‘unconference’ 2 weeks ago where I met Jonathan Feinberg, the creator of Wordle, a program that is much beloved by us Monkey Cagers. Feinberg told me about two other initiatives of his that may be interesting to readers. First, a Java word cloud generator based on the underlying technology of Wordle, which allows you greater customizability through a CLI. Second, and more directly relevant to political scientists, a nice comparison of the language used in US Presidential inaugural addresses.
All visualizations feature a cloud that varies from gray to blue. In this cloud, the size of the word corresponds to the number of times the word was used in a given address. The word’s color depends on how statistically unlikely the word is in the normative text; in other words, a blue word was used more in the given speech than in the others it is compared to. … Some visualizations feature a rose-colored cloud, which features words that were conspicuously absent from a given speech, relative to the comparative corpus. The larger the word, in the rose cloud, the more unusual it is for the word to not have been used. Each inaugural address is compared to the five addresses nearest to it in time, the ten nearest addresses, and all addresses in U.S. history.
Pretty neat.
{ 2 comments }
Pretty neat, indeed.
A beautifully concise quasi-history of the zeitgeist of American politics…’quasi’ not for any methodological shortcomings, but for its indirect approach…you see the self-reflective relationship between the character of U.S. Presidents and the state of public sentiment at the time…without looking at them independently.
Elegant.
I’d love to see some comparison of this analysis to the relative acceptance of each speech as being great or not.
Comments on this entry are closed.