I can't get a higher-than-12 Flesch-Kincaid score no matter how verbose I
get. Also, does the number 12 mean 12th grade (i.e. high school senior year)
ot the 12th year of formal schooling (i.e. 4th year of college)?
Readability Statistics
The readability statistics are based on research by Rudolf Flesch. He
found that words-per-sentence and syllables-per-word are good
predictors of readability. The idea is that longer complex sentence
structures and polysyllabic words require the reader to keep more
temporary state to be able to sort out the meaning of the sentence
when he finally hits a period. (Counting prefixes and suffixes works
just as well as counting syllables. In his book "The Art of Plain
Talk" of 1946 he used affix count.)
The Real Flesch Test
Dr. Flesch has produced several variants of his test over the years.
The genuine Flesch tests all work correctly, because he does careful
studies to make sure that they work. Imitators and copiers of his
ideas do not necessarily take the same pains over accuracy.
One key measure is sentence length. Longer (and generally also more
complex) sentence structures require the reader to keep more temporary
state in his head to be able to sort out the meaning of the sentence
when he finally hits a period.
Another key measure is word complexity. This can be measured either by
counting affixes (prefixes and suffixes other than plural markers) or
by counting syllables; either will work as long as you use the correct
multiplier. Complex words require more mental effort for the user to
understand their meaning.
Some versions of his test also measure use of personal names. (See his
book "The Art of Plain Talk", published in 1946.) Personal names,
pronouns, or words for blood relationship (brother, mother…) (but not
impersonal names like "the doctor") increase the effort that the
reader is willing to expend on understanding a sentence. People are
better at reading about people than they are at reading about other
topics.
Names have the same weight as affixes but in the opposite direction.
The average number of names per sentence is about one for 60-point
writing; you can change the score about 5 point by using zero or two.
For 90-point writing the average is about two names per sentence; you
lose ten points if you don't use names.
One version of Flesch's test can be found at:
http://www.mang.canterbury.ac.nz/courseinfo/AcademicWriting/Flesch.htm
This version of the test reduces the raw data to a 0 - 100 scale using
this formula:
206.835 - 1.015 x words-per-sentence - 84.6 x syllables-per-word
A sentence scoring 100 is dead easy to understand; a sentence scoring
zero is unreasonably difficult. Anything below 60 should be considered
as suitable only for a special audience. Comics score above 90,
advertising is typically above 80, Reader's Digest and Sports
Illustrated above 60, Time and Newsweek about 50, Harvard Law Review
32, a typical auto insurance policy about 10, and the Internal Revenue
Code gets negative 6.
Flesch scores have in turn been correlated to the educational grade
level required for easy reading. Here is a rough table of
equivalences.
Score Grade
90 - 100
80 - 90 6th
70 - 80 7th
60 - 70 8th and 9th
50 - 60 10th to 12th (high school)
30 - 50 Some college
0 - 30 college graduate
To improve your writing's score, break sentences into shorter
sentences and use shorter words. When you split up sentences
containing subordinate clauses you will get shorter sentences starting
with the words "and", "or", or "but". The grammar checker won't like
it, but your readers probably will. When looking for shorter word,
look particularly for words with fewer prefixes and suffixes. Use a
thesaurus if necessary.
Word's Flesch Test
A real Flesch test counts only running text; it skips headings,
captions, and the like. Word seems to ignore things that are not
sentences; i.e. strings of words that don't end in sentence-ending
punctuation. Headings generally should not end in punctuation, so they
will generally be ignored. However, if your text includes tables or
other non-running text that include sentence punctuation, Word may
erroneously include them in the computation.
A real Flesch test counts syllables (or affixes). Since Word does not
display counts of syllables or affixes, one might assume that it does
not actually count them. However, it appears that Word actually does
measure syllables (not affixes) for computing readability.
Word apparently caps the words-per-sentence metric in its formula at
about 50, longer sentences don't get worse scores in Word, though they
would in a real Flesch test.
Word counts passive sentences, but apparently does not use this number
in computing readability.
Word uses an unspecified formula, but it is probably much like the one
on the web site mentioned above.
Word displays both the raw score and the grade. Word's help file used
to suggest aiming at eighth grade. In Word 2002, the suggestion has
diminished to "7 to 8". Note that Word never displays anything more
discouraging than 12th grade, no matter how badly you write. It also
doesn't display negative raw scores. Don't be complacent about a score
of 12th grade; you may be doing much worse, but Word is shielding you
from the awful truth.
Bob S
The above is supposedly about ninth grade level. Did it seem easy
enough?
