There was some mild buzz over the announcement recently that Wolfram Alpha had uploaded the works of Shakespeare into its database. If you’ve never heard of Wolfram Alpha, you’re hardly alone—but you must not be not a math student. Wolfram Alpha is a “computational knowledge engine” designed to answer questions that require computation posed in natural language. Reportedly it works a treat at solving equations (hence its popularity with math students) and at answering questions that require manipulating numerical data (want to know the coal production in 1978 of the ten countries with the lowest GDP? Here you are). And even though Wolfram Alpha has made virtually no progress toward its announced “long-term goal [of making] all systematic knowledge immediately computable and accessible to everyone” (don’t get me started on its results for “Doctor Who”), its announcement couldn’t help but raise hopes for a new tool in Shakespeare studies, especially the ever more important field of stylometric analysis.
Bloggers such as Shakespeare Geek and Stuart Ian Burns of The Hamlet Weblog have already weighed in with qualified interest. I’m much more skeptical. I believe that because it’s purely computational, Wolfram Alpha is useless for Shakespeare research, specifically for stylometric analysis.
The fundamental problem is that Wolfram Alpha only supplies sophisticated variations on word counts. But counting words isn’t analysis, or even the starting point of analysis; it’s a preliminary to analysis. Yes, it’s an irresistible bit of trivia to know that Lear’s part consists of 5625 words, but then what? As it happens, my very favorite Monty Python sketch addresses this subject—an interview with the “Great Actor” Sir Edwin:
Alan: How many words did you have to say as King Lear at the Aldwitch in ’52?
Sir Edwin: Ah, well, I don’t want you to get the impression it’s just a question of the number of words… um… I mean, getting them in the right order is just as important. Old Peter Hall used to say to me, ‘They’re all there Eddie, now we’ve got to get them in the right order.’
Other results are too idiosyncratic to be useful. Consider the result of searching for “Troilus and Cressida,” which is typical. Would even the most desperate student with a deadline want to know the most capitalized words (incredibly enough, they are almost all proper names) or the most common two-word combinations in the play? Two-word combinations would be interesting if they reflected something characteristic about Shakespeare’s style—for example, if they never appeared in other writers—but you won’t learn that from Wolfram Alpha.
Second, Wolfram Alpha’s search engine is inflexible and its database too impoverished to be useful. It can handle a simple query such as “occurrences of ‘dog’ in Hamlet” (two), but it won’t even understand a query that doesn’t involve counting words, such as “how many feminine endings in Hamlet.”
Third, serious computerized analysis can’t be limited to Shakespeare’s texts. Stylometric analysis is comparative.
I’ll grant that one Wolfram Alpha statistic looks interesting at first—the list of words per character. I’ve always wanted a reference for the size of Hamlet’s part, and now I know that it’s 11,631 words.