To be or not to be: Shakespeare’s ‘stylometrics’
This article was originally on a blog post platform and may be missing photos, graphics or links. See About archive blog posts.
Lit professors might be wary of the Shakespeare authorship question, as Ward Elliott writes in his review of James Shapiro’s ‘Contested Will,’ but he certainly hasn’t been: In the 1980s, this longtime professor of government at Claremont McKenna College plunged into the authorship question assisted by a form of number-crunching statistical analysis referred to as ‘stylometrics.’ Since then, Elliott and the Claremont Shakespeare Clinic have enlisted student-led teams to assess various claimants to Shakespeare’s work. The most recent clinic worked this spring. Elliott talked to Nick Owchar about his work.
Jacket Copy: What is the stylometric approach?
Ward Elliott: What you’re looking for are countable features — indications of a possible style. One traditional approach to iambic pentameter, for example, is to look at what kind of line-endings you have. Do they stop with punctuation, or does the line continue to the next without any kind of stop? In early Shakespeare and other playwrights in the 1590s, lines tend to stop at the end of each one. You find some kind of punctuation, giving the speaker time to gather his breath before continuing on to the next line. Late Shakespeare, however, has a much higher frequency of open lines. You look at such an aspect in Shakespeare and hold it up against the writing samples of possible claimants.
JC: What causes a professor of government to plunge into the question of Shakespearean authorship? How did you become involved?
WE: I remember running across an article in 1985 about a young scholar, Gary Taylor, who was editing a collection of Shakespeare’s plays and included a new poem he identified as being by Shakespeare, “Shall I die?” There was a huge ruckus over it in Lit departments. There were those who said the new poem wasn’t by him; there were others who said it sounded like Shakespeare on a bad day. It was all fascinating to me. I also remembered a 1976 article by Stanford statistician Brad Efron and his student Ronald Thisted in the journal Biometrika [“Estimating the number of unseen species: How many words did Shakespeare know?”]. It was a neat demonstration of a statistical approach to Shakespeare’s vocabulary and how it might be affected by the discovery of other works by him…like the poem Taylor had found. I contacted Thisted, who was a Pomona grad, and asked him if that methodology could be used not just for a single new poem, but for the entire authorship question. He suggested it could and, thanks to the Sloan Foundation, which gave The Claremont Colleges a big grant to fund applications of computers in the humanities, our Shakespeare clinic was able to set up.
JC: Where does the clinic get its list of claimants to test? What do you need from a claimant in order to analyze him?
WE: We started with the ‘Reader’s Encyclopedia of Shakespeare” list of 58 supposed claimants and co-authors: Our clinic started by looking at the poets on the list – those people we could get enough samples of poetry to test. You need enough language for a reliable stylometric test. At first, we didn’t know how much would be enough: At least 500 words seemed sufficient. Soon we realized it was much better to have between 1,500 and 3,000 words. Best of all was to have several whole plays.
More after the jump
JC: This is very different from previous ways of investigating Shakespearean authorship, isn’t it? What methods were used before?
WE: The traditional methods involved studying manuscripts closely and looking at documents and the history of the time. It also involves using one’s intuition to see if it sounds like Shakespeare and searching out parallels, images that are found elsewhere. There are some brilliant people out there who can tell you all kinds of things about watermarks or the compositors of the folios.
JC: Though your academic career has concentrated on the world of politics, it’s clear you’ve had a lifelong interest in the bard. Where did this come from?
WE: Partly from my father, William Y. Elliott. He was keenly interested in the authorship question. He was an amazing guy—a Rhodes scholar, counselor to six presidents, the grand old man of the Harvard history department: I’d say he was probably one of the highest-ranking doubters of his day.
JC: He was a doubter?
WE: Yes, he profoundly believed the real author of Shakespeare’s plays was the Earl of Oxford. He didn’t mind letting you know about it.
JC: Does that mean you’re also an Oxfordian?
WE: Oh no, I’m a Stratfordian.
JC: Why are you a Stratfordian?
WE: It’s clear from the Claremont students’ work that Oxford’s poems are in a different statistical galaxy from Shakespeare’s. The odds of that much stylistic discrepancy arising by chance are much lower than the odds of getting hit by lightning.
JC: And do you think the authorship question will ever be adequately resolved for all parties?
WE: I’m not holding my breath. Both sides are too deeply entrenched. But it was a huge step forward for James Shapiro to address the controversy in ‘Contested Will.’ And the Oxfordians have at least tried to rearrange the trenches a bit to shelter Oxford from his own gross mismatches with Shakespeare. Maybe, they argue, all his own poems were just crude learning experiences, written in his teens: The ones he wrote later (which are credited to Shakespeare), after a strange, unexplained silence of two or three decades, showed the polish of a mature writer. That’s hard to believe. I don’t know any author who could change his stylistic spots as suddenly and completely as Oxford would have had to do to become Shakespeare on their timetable, or whose production was as wildly erratic as Oxford’s would have to have been under the Oxfordian scenario - and there is still not a scrap of documentary evidence connecting Oxford or any other claimant with the canon. But even a token defense is better than none at all.
-- Nick Owchar