Reading Level is a powerful and oft-misunderstood tool for writers. Since the upcoming (and exciting!) Draft Map relies on this technique, I thought I'd explain usage here. I'll be using the venerable, old-standby Flesch–Kincaid as the default test in this post, but alternatives exist.
Reading Level approximates the minimum completed US grade level required to comprehend the work.
Note this does not imply the text is inherently appropriate for a child at that grade level. This means they could probably understand with all of the tools at their disposal. They would tire with the sentence complexity and have trouble spending energy on the wide plot, descriptions, and implications.
When analyzing a full work with this formula, this should give you a good guess as to both who can understand the text and those who would do so without effort. While a few lines can transcend the formula, the law of averages will bring the calculation closer to reality. When reading a physical book, jump to ~6 random places, count sentences, words and syllables, and extrapolate from there. The formula will work just as well on 40 sentences as 15,000. A computer analysis does not need to approximate. The methods I've used to counting sentences/words/syllables probably aren't perfect, but they are very close.
When writing, especially with an awesome piece of software, keep in mind the reading level of your audience. Targeting middle grade fiction (8-12 years old) corresponds to ~4-7 grade, so you likely want a text level of ~3-5. Sounds low, but this means your readers won't struggle with long, complex sentences. When producing mass-market fiction, consider 7-9 grade as an ideal range. This permits some denser structure while remaining accessible to a wide audience. Most modern best-sellers conform to this level.
DraftMap highlights individual sentences that break the 12 and 16 grade level barriers. Use this to spot incongruous sentences mixed in with lighter fare. Chopping these into digestible tokens often improves the flow of the piece.
If you've read this far, a brief explanation of how Flesch–Kincaid Grade Level is computed is in order. The base formula is defined thusly:
|words||Total # of word in body|
|sentences||Total # of sentences in body|
|syllables||Sum of all syllables of all words in body|
DraftMap uses a simplified formula to analyze sentences in isolation:
Virtually all reduction of complex structure (i.e. text) to a number can fail to capture reality. As with all automated suggestions, use your best judgment.
Reading Level breaks down with sentences at the extreme ends of the spectrum, packed with heavy words. Longer sentences have a harder time tricking the formula. A long sequence (that still makes sense) is likely tougher for novice readers, regardless of vocabulary.
These examples are taken from Pride and Prejudice by Jane Austen, available to the public domain at Project Gutenberg.
This is an example of a false positive. Owing to the dialog tag and semicolon, the system guesses this to be a college level sentence. In reality, this is three chunks of ~4 grade.
"They have none of them much to recommend them," replied he; "they are all silly and ignorant like other girls; but Lizzy has something more of quickness than her sisters."
This next excerpt is borderline, since the lack of full stops mean an otherwise grade 7 passage is rated at post-college complexity.
However little known the feelings or views of such a man may be on his first entering a neighbourhood, this truth is so well fixed in the minds of the surrounding families, that he is considered the rightful property of some one or other of their daughters.
Consider Reading Level to be yet another wrench in your writers' toolbox. Pull it out when analyzing a block of text and put it away when common sense overrides a simple formula.