Re: Readability Indexes

[Bill Burns <WBURNS -at- VAX -dot- MICRON -dot- COM>]

Stephen Victor adds:

"P.S. - No personal flames intended here; I simply have a high regard for the
integrity of language and look on mechanical means of improving one's writing
with a great deal of suspicion. What can these "grammar checkers" provide that
a competent writer/editor cannot?"

I agree with the contention he makes that the methods these grammar-checking
applications derive from rather questionable premises.  In my own experience,
these applications rely mostly on probability.  If you notice, in Word's grammar
tool, most comments are posed as suggestions.  All constructions using a form of
"be" are flagged as potential passive locutions.  Whenever "that" appears, the
system suggests "which," and vice versa.  The system targets these problems
because they are statistically more prevalent in writing than other grammatical
issues.

I just ran a test using a very brief piece of text (yes, the old peanut-butter
and jelly sandwich creation procedure).  I intentionally added some mistakes to
see how the checker would respond.  Here's the test sample:

To make a peanut-butter and jelly sandwich, do the following:

1. Get bread, peanut butter, and jelly from the pantry.
2. Get a butter knife and cutting board from the shelf.
3. Laying two slices of bread, top-end up, on the cutting board.
4. Open the jar of jelly.
5. Using the knife, spread the jelly in an even layer on one slice of bread.
6. Wipe the knife blade on the other slice of bread.
7. The jar of peanut butter is opened.
8. Using the knife, spread an even layer of peanut butter on the second slice
of bread.
9. Stick the covered sides of both slices together.

So what happened when I ran the checker on this passage using the strict
formal rules that it uses to make its analyses?  On one pass, the checker
noted the use of "laying" instead of "lying" in 3 and suggested the latter
(WRONG!!!!).  In the next pass, the checker noted the passive construction in
7.  It did not catch the missing verb in 3, and it made no mention of parallel
construction.  It gave a Flesch-Kincaid readability level of 4.1, a Coleman-Liau
grade level of 6.2, and a Bormuth grade level of 8.2

This passage is only 101 words long.  I wouldn't rely on Word 6.0's grammar-
checking function on longer documents.  My experience suggests that it isn't
very useful or accurate.  I've tried Grammatik 5, and it seems much more
robust.  I prefer, however, to use my own knowledge and experience in grammar
and writing to find the errors.

Bill Burns                            *
Assm. Technical Writer/Editor         * LIBERTY, n. One of Imagination's most
Micron Technology, Inc.               * precious possessions.
Boise, ID                             *
WBURNS -at- VAX -dot- MICRON -dot- COM                 *                     Ambrose Bierce