The Student Perspectives category collects posts written by current CityLIS students.
Mahmud El-Shafey considers the uses of text analysis applications for creative writing.
You can follow Mahmud on Twitter @MahmudElShafey.
Over the past few months, I have balanced my LIS studies with a Faber Academy course in fiction writing where I am trying to develop a sci-fi novel (elevator pitch: The Arab Spring… in space!). One might think that the two would have very little overlap, but I have experienced an interesting cross pollination of ideas (I’m looking at you, new A.I. character Otlet). Most recently, I have seen how text analysis can serve as a useful tool for creative writing, and particularly the editing process.
Running the first 30,000 words of my first draft through Voyant, an open-source web-based application for performing text analysis, helped clarify a number of things for me. First, a quick glance at the word cloud feature provided immediate revelations. Some obvious. Yes, my main character’s name (Iskander) is the most used term, followed closely by other character names (Damietta, Lebanon, Taki). Some less obvious. Why is “greeted” so high? And “nodded?” Who are these characters I’ve written that are spending all their time greeting and nodding at each other while on the other hand, there are only two instances of “goodbye?”
More importantly, some words that I felt certain would be among the most used terms were completely absent. These obviously loomed larger in my imagination than the physical text.
In addition to the word cloud feature, Voyant includes a number of other tools that I found useful:
◾“Phrases” revealed that on three occasions, Taki performed an action “with a flourish” and that on two occasions Iskander “made a warding gesture.”
◾‘TermsBerry’ was useful in seeing what characters are particularly associated with what words. So Iskander was most closely linked to the characters he interacted with, as well as words like “knew” and “thought” whereas Damietta, a non-POV character, enjoyed a far broader range of observed actions.
◾“Trends” was particularly helpful in tracking characters’ appearances and viewing what is happening in the background of each section of the text.
These insights will certainly prove helpful in the editing and re-writing process. In a final draft, I will make sure that my characters nod less, and pour coffee with less flourishes. However, not all of my fellow students agreed with me on the use of text analysis.
I shared an image of my word cloud with my fellow Faber students, sparking off a spate of text analyses. Some who tried it for themselves came away feeling unsatisfied, viewing the process as “mechanical” and potentially stifling creativity. They also pointed out that the analysis could not take into account authorial intent, such as when particular words or colours are purposefully associated with a character, or when a phrase might be repeated for effect.
While there have been many examples of how text analysis can be used by students of a text, there is no reason why it can’t also be used in the same vein by authors during the writing and editing process. But beware, your mileage may vary.