General Inquirer applied to blogs
Berkman’s Henrick Schneider at the Tuesday morning informal get-together talks about some quick research he’s done using General Inquirer, “A computer-assisted approach for content analyses of textual data” by Philip Stone. It’s a dictionary-based approach with over 10,000 words and 180 categories. (GI has a blog.) E.g., it found a strong correlation between the optimism in the first speech given by presidential candidates and the outcome of the elections. Also, the pessimism in popular songs and newsmagazzins predicted decreased consumer optimism and economic recession.
Henrik did a quick study feeding in blogs about the Eason Jordan case, using just six blogs (or other statements) and only one text from each source, so this is more of a test of whether there’s a reason to go ahead with a statistically significant study. The content came from Eason Jordan, Rebecca MacKinnon, LaShawn Barber, Rony Abovitz, Richard Sambrook, Brian Ericksen.
Ethan points out that GI is sometimes used to note rhetorical signatures. Henrik’s mini-study shows, for example, that Jordan and Sambrook use the most “negative” terms, Abovitz is distinctively positive, and Ericksen (an Army guy) uses the most political terms.
Applied more broadly — Ethan suggests looking at the top 100 technorati-ranked bloggers — this could be quite interesting. We kicked around other ideas, e.g., looking at the deviation among US MSM, foreign media and bloggers on a particular topic. [Technorati tags: blogs berkman]
