What would George Orwell think? While the origins of textual analysis dates back to ancient times, the use of computers to analyze texts is a recent development. A team of researchers, led by social psychologist James Pennebaker of U.T. at Austin, have set up a blog named Wordwatchers, which details their interpretations of data mined from politicians’ speeches. First they use a computer program to mine the speeches for certain elements of grammar and vocabulary. Then they interpret the data. For instance, the computer program can measure the frequency of politicians’ use of personal pronouns (I, me, we, your, etc.) during a debate. The candidate found to use personal pronouns more frequently is then inferred by the researchers to be “more personal.”
While this methodology is far from perfect (like any quantitative study it tends toward reductionism), the results of the research are nonetheless fascinating. Check out Pennebaker’s analysis of the final 2008 Obama-McCain debate or his study of U.S. presidents’ States of Union addresses from Truman to Obama. What do you make of the results of these studies? And the methodologies employed? I’m interested in reading any comments you might have.