I was recently reading a blog about sentence length at Divertir Publishing. It mentioned reading level algorithms which I immediately researched. I was intrigued because, although I write every chance I get, I’ve never really thought about how it affects the reader.
Some people, like my wife, enjoy quick reads. If she comes across a really descriptive paragraph that is laying out the environment, she skips ahead. She wants the meat of the prose only; just the facts, Mam. Whereas some readers, like myself, can sit down with Proust or Dickens and be happy. If you want to tell me, in five pages or more, how you put on your pajamas last night, that’s fine. As long as the language is intoxicating and the words perform their ballet gracefully on the page. But what makes a quick read a quick read?
Sure, there’s the reader’s interest level. You love a book and so you stay up all night and finish it at 3:00 am. But sentence length plays an important role too. The shorter the sentences, the quicker your inner voice is going to read. And of course, since there are fewer words to read, it’s a quicker read by default anyway. Short sentences infer action, and maybe dialog. It races forward. Bullies its way along the pages.
Longer sentences command more attention. You can’t have ADD and read a Proustian sentence to the end. Once you reach the fourth aside, you start daydreaming and wondering what the first part of the sentence was even talking about to begin with. Longer sentences infer complexity. They are for setting scenes, backstories, character’s thoughts, philosophies, and description in general. They saunter, nonchalant across the page. They may require more introspection and brain power. I may read Dickens, but I don’t attempt to do so waiting in a doctor’s office.
So sentence length is important because it can affect how your work is read. And that means it can determine your audience. And knowing your audience is compulsory.
The Flesch Reading Ease and the Flesch–Kincaid Grade Level tests are used to determine ease of comprehension and reading difficulty. They are used by people like the U.S. Department of Defense, included in applications like Microsoft Word, and used as a gauge for certain standards on documents like insurance policies. And the algorithm used in these tests both use words per sentence as a variable. The other variable is syllables per word. The longer the sentences, the more syllables per words, the more complex the work. At least in general. Sure, there may be plenty two syllable words you’ll have to look up in a dictionary, and an algorithm like this doesn’t allow for obtuse metaphors, but overall it’s a good measuring stick.
So like any good geek, I felt the need to turn sentence length into a mathematical construct. I developed a program that takes anything written, measures the sentence length, and graphs it for a visual representation of the words per sentence. Using the graphs, you should be able to tell a few things about the work without actually seeing it. For this article, I’ve only graphed the first 50 sentences. Can you tell what kind of read the first graph would be? Don’t look at the X-Axis label.
It’s Janet Evanovich’s One for the Money. A quick read. I’m not grabbing the average length yet, but from the graph you can see the novel starts out with words/sentence in the 8 to 15 range, with only 3 sentences over 20 words.
What about this one?
It’s Proust. A brain bender and acquired taste. Easy to get lost. It can’t even be graphed using the 50 word max. Here’s the full graph.
The intro to Swann’s Way hardly ever dips below the 20 words/sentence mark. And a couple of paragraphs masquerading as sentences weigh in at over 160 words. I wonder if Proust talked like he wrote. You would have to have Steven Tyler lungs.
After a few different books, I fed it one of my blogs called Potential Energy. I think it looks like a healthy EKG. This looks to me like a diverse mix, enough to keep you on your toes so you don’t wander off on me.
Then I fed in the first part of one of my novellas called Sorry Charlie. It looks like someone should hit the nurse button. There seems to be a literary arrhythmia taking place.
And maybe I shouldn’t place to much import on these little graphs, but I’m very happy with Potential Energy and not so much with the start of Sorry Charlie. At any rate, I think these might be useful.
You could, for instance, feed in a book and see at a glance if it matched your reading level or preference. You could gather metrics over time to uncover patterns in the graphs. You might be able to tell a particular genre by the graph. Every book would have an EKG.
If you are a writer, you could use it as a tool to pick out the flow of certain sequences in your book. The graph above can be colored in under the line for specific ranges where there is considerable action, as noted by contiguous, short sentences, or where there is more thought provoking prose, as evidenced by a cluster of longer sentences. You could then visually ascertain whether your writing was leading the reader down the path you intended.
And of course, there are other chart styles that could be explored. Other variables you could take into account, so on and so forth. But aside from all the geekery, I will certainly give the overall sentence length variation a look.
If you would like to play with the already available algorithms online, Google ‘reading level algorithms’ and feed in some of your writings. Are you writing for your audience?
Pingback: A geek with a hat » I want to analyze your blog