My project for Summer 2013 is to design a text-analysis algorithm capable of recognizing Shakespeare’s rhetorical figures. For instance, this repetition of “farewell” in Othello is called an anaphora:
[1]
That’s a pretty straightforward anaphora, and is just the kind of linguistic feature that a pattern-recognizing algorithm could detect. I could show you more complicated examples, but first let’s imagine the higher-order interpretations that this algorithm would enable.
Start with that idea of recognition. The average schoolboy in Renaissance England was rigorously trained to recognize rhetorical figures, in order to imitate them in his own writing.[2] This was formalized analysis according to classical precepts, adapted for the English language. John Milton’s copy of Harington’s translation of Ariosto’s Orlando Furioso is thick with marginal annotations on its rhetoric — recognizing not only its argumentative stages but also “the fineness of speech in the Rhetoricall ornaments, as comely tropes, pleasant figures,” and the like. The purpose of recognition was imitation. To quote another contemporary pedagogue, the pupil learned “to flourish and adorn it [his own writing] neatly with Rhetorical Tropes and Figures, always regarding the composure of words.”[3]
So if recognition served imitation, then systematic recognition of verbal patterns of repetition and variation — or in other words, of rhetorical figures — will enable textual interpreters to recognize broader systems of imitation and influence. What would that enable?
John Lyly (in)famously wrote Euphues (1578) with as many figures as possible; see Wikipedia’s entry on euphuism for examples. Imagine that you could place an anaphora filter over Lyly’s text, highlighting each repeated word in green, as I did in the image above. Then imagine a similar filter for anthimeria, antimetabole, apostrophe — all the way through the alphabet to zeugma.
[4]
With zeugma we’ve moved beyond mere repetitions, into more complicated natural-language processing that reliably recognizes early modern parts of speech; in 2013 we’re not quite able to do that yet, as I’ve written elsewhere. But because I’m an optimist, and because I want to claim this very cool-sounding word before someone else does, I’m calling this future algorithm the Zeugmatic.
When complete, the Zeugmatic will automate the detection of rhetorical figures from anaphoras to zeugmas. It will map a single text’s figures, and compare them to other texts’. It will identify usage patterns from models to imitations, helping us to quantify claims of imitation (like Milton’s of Ariosto) and to identify imitations we haven’t yet recognized. It will compare the figural habits of two authors, and of any permutation of authors. It will identify the figural signatures of students of influential teachers or schools, students who read the same texts and recognized the same figures. It will point us toward the texts and writers that had a measurable influence on the figural habits of other texts and writers–say preachers on playwrights, or Lyly on Burton.
But not yet, not in 2013. Phase 1 begins in May, when my research assistants Sarah Hertz and Maria Jaramillo and I start to identify the low-hanging rhetorical fruit like anaphoras and other repetitions, and learning more about SEASR and the other text-analysis tools that look most promising. (Thanks to the University of Calgary’s PURE Awards for initial funding, and to SSHRC for an Insight Development Grant to 2015.) Our Basecamp page for project management is busy with meeting notes and to-dos, and we’ll post updates and queries throughout the summer. In the meantime, you can follow the project on Twitter.
++ Footnotes ++
[1] Wales, Katie. “An A-Z of Rhetorical Terms.” in Sylvia Adamson, Lynette Hunter, et al. (eds.) Reading Shakespeare’s Dramatic Language: A Guide. London: Arden Shakespeare, 2001. 271-301; 278. Photo with Crossprocess (app) by the author.
[2] Vickers, Brian. In Defence of Rhetoric. Oxford: Clarendon Press, 1988. 254-293; esp. 259-61.
[3] Cit. Vickers, 261.
[4] Vickers, 498.
