How Readable is Science Fiction?
a consideration by L J Hurst
The Flesch Test is a formula designed to indicate the readability of a text. It can be used to indicate whether a style is appropriate to a genre - you'd expect a textbook to be more difficult to read than a novel, but you might want to know whether your textbook was written in a difficult way. And, the test can be used to indicate the origin of a text, by relating the figures to other books written by the same author. It may be because of this that it has been one of the tools that have been used in analyses of alleged confessions etc, that have been brought back to the Court Of Appeal recently, as it indicates something of an author's individual style. I've come across references to it in several places, and been driven to try it out on a number of SF classics and not-so classics.
Flesch calculated Ease of Reading with this formula:
RE = 206.835 - (SYLL x 0.846) - (SL x 1.015)
SYLL is the number of syllables per 100 words of the text, and SL is the average sentence length (of words in the text divided by number of full stops).
This gives a number on the scale from 0 to 100, with anything less than 30 being considered difficult, and anything over 90 very easy.
In his 1987 book Diagnosis_And_Detection (Associated University Press) Pasquale Accardo gives an analysis of all the Sherlock Holmes stories and novels, and of the Holmes pastiches. He produces results like this:
He also gives the averages for the five collections: the Memoirs, the second volume, is lowest: and, the last collection, the Casebook is highest:
but the range overall is quite tight. Comparing the figures with those for other authors who have taken over the character is one indication of how close they were able to come in adopting the character and the style. If we look at four of the books, written with SF associations, we can get an idea of their verissimilitude:
The four books were, in order,
Adventures_Of_The_Peerless_Peer (* I calculate a figure of 70 for this book),
The four authors have written books that have a different Ease of Reading, or are more demanding on the reader, falling at the edge or outisde Conan Doyle's range. They use longer words in longer sentences. In other words, they have not managed to imitate Conan Doyle's style. Now to turn to SF itself. Here are my analyses of a number of SF novels and a few short stories. These are not based on the full text, but (following a statement of the Flesch formula in Writers_News) on a sample of 147 words. In each case the sample began at the fifth paragraph of the fifth chapter, or at just the fifth paragraph in the case of short stories. I have listed them in descending order of readability.
First, the novels:
The novels were, in order,
Now, the short stories:
The stories were, in order,
"The Nine Billion Names Of God",
"I'll Be Waiting For You When The Swimming Pool Is Empty",
"The Game Of Rat And Dragon".
The novels obviously fall into four groups, which I can't really explain. You can understand the classic professionals hitting the top two slots (and the short story, too). The Heinlein might be considered a juvenile and so aimed at a market demanding easier reading, and Blish would have been used to writing for a wide public at the end of the pulp era, but why the bottom of the list should be occupied by three or four dystopiae is not so clear. Although dystopiae are traditionally more demanding (though Aldous Huxley managed to avoid that - perhaps because he was writing satire), and all the four authors are usually regarded as intelligent, they are never written about as if they are hard or obscure. There is little to distinguish between the sexes, and not a lot between periods of writing (unless you suppose that things started difficult with Wells, grew better up to the fifties and then fell away in a sort of bell curve, except that people like Huxley and Van Vogt should be the other way round to really prove that).
A couple of things may challenge the idea that the Flesch test is objective: firstly, that the books cover such a wide range, and they don't seem that separate on reading them, and secondly, the list came out not as I expected before I began the calculations. Miscounting a syllable or too, would not account for the extent of the differences.
What I have not done is attempt to measure a number of books by one author as Accardo did for Conan Doyle, but as the examples from his analyses show, the attempted scientific examination of these texts can provide some thought-provoking results. And, it provides a way of emulating the writing style of an author you like. To have written more like Dr Watson, Philip Jose Farmer should have written longer sentences and more monosyllables.
This article first appeared in VECTOR The Critical Journal of the British Science Fiction Association
Return to Home Page