A computational approach to Eng Lit

Updated with a few links to projects concerned with using computers in the context of literature.

I really like the way the writer portrays something or other, and how nuanced his use of blah blah blah. However, I am surprised that such an eminent personage as Dr Bunkoff would dismiss Jenkins’ theory so lightly....
— EXCERPT FROM A TYPICAL BOOK REVIEW
Poetry generator, by Terry Freedman

This is my random poetry, or book review, or literary essay generator. I think size might be an issue, but I’m working on it. Credit: Poetry generator, by Terry Freedman

Literary criticism is all so vague. I mean, it’s like trying to grasp a cloud.

How about a more mathematical approach? I tried this out, using a chapter from Clive James’ Unreliable Memoirs. I noted how many examples of humour there were on each page, and how many examples of pathos. I entered all these numbers in a spreadsheet, and turned them into a graph. You can see the result below:

Unreliable Memoirs Analysis, by Terry Freedman

Analysis of a chapter from Unreliable Memoirs. Terry Freedman

Now that’s what I call literary criticism. None of this airy-fairy rubbish, but good solid numbers, and a graph to boot. If one carried on applying this approach to the whole book, and taking into consideration more criteria, you would be able to see the shape of the entire narrative at a glance. You wouldn’t even have to read anything.

In fact, you wouldn’t even need to read the book if you didn’t like the graph — because you would know in advance exactly what kind of a book it was going to be.

I shall be writing to the Education Secretary to suggest that this approach be adopted in schools. Why should kids have to wade through stuff like this…

Why, thou clay brained guts, thou knotty pated fool, thou whoreson obscene greasy tallow catch!
— HENRY IV, PART 1

… when they could simply look at a graph instead?

If book blurb writers had any sense, they wouldn’t put wordy descriptions on the back cover of books. No, they would put a graph there, or perhaps a sort of nutrition label in which elements of the book are colour-coded, and given percentages:

Ingredients:

Humour 5%

Horribly gruesome stuff 17.5%

and so on.

Bottom line: the trouble with Eng Lit specialists is that they put too much store by words alone. Come on, let’s have some hard facts in the form of numbers as well.

Update

This article was inspired by the response I received when I presented the Clive James analysis to a creative writing class. It was received with interest. That approach was, in turn, inspired by an article I read circa 1990, in which an English teacher had his students analyse Shakespeare’s plays in much the same way.

I’m writing this now because in the most recent edition of Hello World magazine, Tom Liam Lynch proposes exactly the same thing, with some minor variations. He suggests, quite rightly, that such an approach can lead to students asking or considering some interesting questions. That, to me, is the most important aspect of this kind of approach. Some people commented, when I posted this originally, that such analysis did not convey the richness of the work under scrutiny. But in a way, that isn’t the point. The purpose of the computational approach is, surely, to suggest questions to ask. As Stephen Downes said when commenting on my original article:

I’m honestly not sure whether this post is serious or satire, but it doesn’t really matter, because the idea of using graphs instead of text for literary criticism is satisfying either way.
— Stephen Downes

My article was written a little tongue-in-cheek because many book reviews and book blurbs suggest that the reviewer has not actually read the book, but is simply plucking words and phrases from a bank of words and phrases. How many book reviews have described the tome in question as “breathtaking”, or a “worthy addition to our knowledge of…”, or “a brilliant newcomer into the world of literary fiction”?

(That’s why a kind of “ingredients label” and/or a graph would be a much better indicator of whether the book is worth buying. And much better than a star rating which, from the reading I’ve done on this subject, doesn’t have much effect on would-be readers anyway.)

I often wonder if the most arduous part of the work of the people who write this sort of tripe is to ensure they don’t use the same words more than once for the same editor.

It seems to me that if you’re going to encourage students to use computer programming in literature studies, one very worthwhile project would be to get them to create a random book review or literary essay generator.

Incidentally, on the subject of random generators, there is an Ada Lovelace Poetry Generator, a Scratch program, that children can try for themselves and change.

Similarly, Giulia Carla Rossi has been working on using the Python programming language to generate random poetry from text, and also to analyse the structure of poems.

In conclusion, what a great idea it would be to write a computer program to analyse a piece of literature, and then automatically write up the analysis in the form of a literary essay. We have the technology, so why not cut out the middleman?

See more of my writing on my Eclecticism website and newsletter.

Copyright Terry Freedman. All rights reserved.