| 
theory,
tests and tools
This
page considers online readability tests.
It covers -
readability tests
What is a test of readability? William DuBay commented
that by
the 1980s, there were readability 200 formulas and over
a thousand studies regarding their statistical and theoretical
validity.
Basic measures of readability include the Flesch Reading
Ease and Flesch-Kincaid tests (featured on Microsoft Word),
the Fry
Formula, SMOG
(Simple Measure of Gobbledegook) test, Gunning Fog
index and Cloze Procedure.
They are indicative only (readability online can be significantly
affected by placement and format of the text) and as Rudolf
Flesch commented in his The Art of Readable Writing
(New York: Wiley 1994) -
Some
readers, I am afraid, will expect a magic formula for
good writing and will be disappointed with my little
yardstick... What I hope for are readers who won't take
the formula too seriously and won't expect from it more
than a rough estimate.
It
is generally accepted that simply writing for a low score
will not, in itself, improve the comprehensibility of
a page. Substantial empirical research suggests that users
have a 'preferred reading level' online and offline, influenced
by
Interest
- as Keith Rayner & Alexander Pollatsek note in
The Psychology of Reading (Englewood Cliffs:
Prentice-Hall 1989) users will read at a higher level
than normal if interested in the subject matter.
Readability of the type - a user's preferred
reading level decreases when type is too small, too
dense, too faint or reversed out of a colour (showing
up as white on a colored or black background). There's
a detailed discussion in Miles Tinker's lucid Legibility
of Print (Ames: Iowa State Uni Press 1963)
Sentence length and redundancy - short sentences
and reinforcement of ideas help people to read at higher
levels than they're used to.
Format - white space and pertinent illustrations
provide visual breaks that encourage the reader to keep
going and "make the overall appearance of the material".
Conceptual density - too many new concepts or
excessive condensation in a given number of words tend
to deter users.
Flesch and Flesch-Kincaid measures
The Flesch formulas measure reading ease and 'human interest',
based on four principles:
1.
the more syllables in a word, the harder it is to read
and understand that word.
2. the more words in a sentence, the harder it is to
read and understand.
3. the more words about people there are in a passage,
the more "interesting" it is to read.
4. the more sentences "addressed to an audience",
the more interesting that passage is to read.
The
Flesch Reading Ease Score identifies readability based
on the average number of syllables per word (ASL) and
the average number of words per sentence (AWS). The formula
is -
206.835
- (1.015 X ASL) - (84.6 X ASW)
The
formula results in a reproducible and predictable score
in the range from zero to 100, with higher scores signifying
greater reading ease. The average Reading Ease score for
comic strips is around 90, with the Australian Financial
Review in the 40s, non-specialist legal journals
around 30, standard insurance policies around 15 and some
of the wilder deconstructionist journals around 10.
The formula for the Flesch-Kincaid Grade Level score,
an indicator of the level of education (K-12 and beyond)
is -
(.39
x ASL) + (11.8 x ASW) - 15.59
where
ASL is average sentence length (number of words divided
by number of sentences) and ASW is average number of syllables
per word (number of syllables divided by number of words).
Ben Yagoda comments that the appeal of Flesch's work -
such as his landmark 1946 The Art of Plain Talk
- is that it
broke down the issue of writing and style into a formula
and thus made it seem scientific. Flesch and his progeny
also expressed themselves with a Dale Carnegie, Kiwanis
Club breeziness, full of italics and direct address,
that made achieving a good style seem nothing fancy,
just good business sense. It was so simple! One could
analyze every piece of writing to find its "reading
ease" score, and the easier the better. "The
simple style -- the style that meets the scientific
tests of readability -- is the classic style of
great literature," Flesch wrote. "If you start
to analyze what style is, the only possible general
rule is that the reader must be able to understand what
the writer says; and the surest way to that is simplicity."
Flesch was silent on the nature of the general rule
when the writer is trying to say something subtle and
complicated.
Automated
versions of Flesch-Kincaid should be used with care. There
have been suggestions that Microsoft's version of the
Flesch-Kincaid does not score above grade 12, ie understating
the difficulty of some documents.
Gunning Fog Index
The Gunning Fog Index is a similar measure of reading
ease and comprehension, using the formula -
.4
(ASL + Wpoly ) = Fog Index
Typically
those using the Index select a sample (of say one hundred
words), determine the average number of words per sentence,
determine the percentage of 'hard' words (those with more
than two syllables), add the two figures and multiply
by 0.4. The 'ideal' Index score is 7 or 8, with a score
above 12 suggesting that the text would present difficulties
for most people.
FORCAST
Another indicator of readbility is provided by the US
FORCAST formula developed for technical material.
| Readability
=
|
20
- number of one-syllable words
10
|
1.
count the number of one syllable words in a 150 word
passage
2.
divide that number by 10
3. subtract the answer from 20.
The
Cloze Procedure is another readability formula based on
the deletion of every fifth word. The reader's ability
to fill in the blanks becomes the measure of the text's
readability
next page
(fonts)
|
