Category : Tutorials + Patches
Archive   : STYLE.ZIP
Filename : STYLE.TXT
Output of file : STYLE.TXT contained in archive : STYLE.ZIP
From: david tomlinson
Date: 06 Feb 89 21:44
Subject: Report on Style Checkers, I
THE STYLE CHECKERS
By Professor David Tomlinson
and
Associate Professor Harriet Bergmann
PART I
Containing
Introduction to Style Checkers
A Letter From Willy: A Sample Used in Testing Style Checkers
As Well As Reviews of
Clearcut
Grammatik II
Part of the Instructional Development Project
A COMPREHENSIVE SURVEY OF SOFTWARE IN ENGLISH
United States Naval Academy
1988
Please distribute and reproduce freely.
This work has been done with public funds and is therefore
not copyrighted.
The line at the beginning of each review contains a left curly bracked ({).
Use the search mechanism of your word processor to take you quickly to the
file you wish to read by having it find the curly bracket.
{
INTRODUCTION TO STYLE CHECKERS
Style checkers are computer programs that tell as much as a computer
can tell about a piece of writing. A style checker reads a piece of prose
and analyzes it in two ways: by counting characters, words or sentences and
by recognizing verbal patterns. Style checkers cannot understand, so they
cannot evaluate prose. They can compare one text with another in terms of
numbers and patterns, and tell a writer about the comparison. When used
carefully and knowledgeably, a good style checker can reveal habits of
style, punctuation and composition which materially affect the quality of
the writing. Since such programs judge impartially, students often treat
the advice they give with great respect. We have watched students who
blithely disregard a teacher's comments about excessive use of the passive
voice, for example, take a style checker's relentless isolating of passives
quite seriously. Indeed, we have watched a number of students work
tirelessly to achieve a clear, concise writing style because they heeded a
style checker's analysis. Not all students respond positively to this
mechanistic approach; however, most find the information the good checkers
furnish helpful.
WHAT THE PROGRAM DOES
1. COUNTING
Style checkers count numbers of letters in each word. From that count,
some style checkers make inferences about the complexity of the vocabulary
in the text. Some programs consider words over 10 letters to be complex,
and others wait for fourteen or fifteen letters before declaring the word
complicated. Programs that have readability indices usually assume that
long words are not good, and that short words are desirable.
Similarly, style checkers count the number of words in each sentence.
They can read the period at the end of a sentence and the spaces that
follow it, identifying each unit that falls between as a sentence. Again,
programs that use an index usually make a negative judgment about long
sentences. Each program has a different default setting for recognizing
long sentences. RIGHTWRITER 3.0, for example, is initially set to label
sentences of more than 25 words as long; but the user may instruct the
program to label sentences of any length as excessively long. Other
programs offer the user no choice in this regard. They come preset.
Usually the accompanying manual will disclose the parameter used.
2. DICTIONARIES
Some style checkers have dictionaries--lists of words and phrases that
present trouble of some sort to a writer. Most style checkers will flag
any use of a form of the verb "to be" followed by a word ending in "ed" and
warn the writer that he/she has used a passive--in some cases this is
accurate, and in some cases ("The problem is greed") it's wrong. In a
simple program the dictionary may include such words as "like" and "as,"
each of which will trigger a warning that the word may be misused. A more
complex program may read the a period followed by two spaces as well as
"And" or "But" and warn the writer that a conjunction should not begin a
sentence.
3. CUSTOMIZING
Some style checkers, often the better ones, can be customized; the
writer can add words or phrases to the dictionary. This ability is
particularly useful for teachers who identify specific problem areas in
student writing and want to alert the students to them. It is also useful
for those alert writers who are aware of their own problems in writing and
who wish help in avoiding grammatical and stylistic errors.
Several programs permit users to choose which grammar and style rules
will be used in scanning writing. GRAMMATIK in both versions reviewed in
this pamphlet and RIGHTWRITER 3.0 both offer this feature. READABILITY
allows the user wide choice in determining the kind of writing examined and
the kinds of analytical charts displayed.
HOW THE PROGRAM SHOWS RESULTS
1. PRINTOUT
Most programs will print out a summary of their findings; if a program
does offer this option, the user can always use "print-screen" to keep a
hard copy. The more complex programs will print lists and sometime charts
or graphs of their findings.
2. ON-SCREEN DISPLAY
Some programs display their findings within the paper, marking all
trouble spots sentence by sentence. Some summarize their findings at the
end of the checking process. A few like GRAMMATIK III allow the user to
make changes to the text as the program points out a possible difficulty.
Some programs offer all these options. For most writers, a sentence-by-
sentence or screen-by-screen display will be most useful, unless the paper
is quite long.
HOW TO USE THE INFORMATION
The concept of readability is used in a variety of ways by these
programs, but in general they favor brevity over length. For newspaper
writing this emphasis proves invaluable. Many teachers encourage college
freshmen to write clear, simple sentences a la Hemingway. Their students,
cognizant of the programs' limitations, often find the style checkers most
helpful. Businesses invariably encourage using short, clear sentences in
communications because understandable written work saves both time and
money. Academics who glory in complexity while blithely obscuring meaning
will have little regard for the values fostered by the style checkers or by
teachers of writing; each writer will have to find his/her own way.
WILLY'S LETTER
Many of the reviews which follow use one letter, one from Willy, to
help users measure how good or how poor a style checker is. Willy's letter
contains numerous grammatical and stylistic errors. It is, however,
written in short sentences which are easy to understand. Few style
checkers recognize both its strengths and its weaknesses. Those checkers
gauged to test only readability rate Willy's letter as superb but take no
note of the spelling and grammatical errors which plague it. The more
complex checkers, those which not only rate readability but which check for
usage as well, prove more moderate and more accurate in their consideration
of the letter. We have included an unadorned copy of Willy's letter in the
file WILLY.DOC for those who wish to refer to it after reading a review.
RATINGS OF THE STYLE CHECKERS
We have given a general rating to each package reviewed. We have used
stars to indicate our evaluation of each program's power and usefulness.
We have placed the stars just above the program name at the beginning of a
review.
* POOR
** GOOD
*** EXCELLENT
**** SUPERB
The estimates given relate to the work we must do with students. They
are quick guides to the usefulness of the program to the two of us doing
teaching at the U. S. Naval Academy. Other people in other situations
might find programs we have rated "poor" performing well for them.
Programs we have rated as "excellent" or "superb" might not help in some
situations. Our judgments are not meant to be sweeping or scientific but
purely to provide a quick reference to those teaching college students or
requiring written work from them. Even these people should use the ratings
as a general indication only. If the features you need are in a program we
have not assigned a strong rating, then use the program, not the rating.
We have given our best ratings to those programs which had the features we
felt appropriate for us and for our students.
FOR CLARIFICATION OR HELP
If we can offer clarification of the information given here or provide
assistance with these or other programs in writing, please get in touch
with us.
Professor David Tomlinson
Associate Professor Harriet Bergmann
English Department
U. S. Naval Academy
Annapolis, Maryland 21402-5044
(301)267-3274
(301)267-3425
(301)267-3277
{______________________________________________________________________________
A LETTER FROM WILLY
A Sample Used in Testing Style Checkers
Dear Clarence,
I hope this letter finds you and your family are doing well.
Please pass on my regards to them. It has been a long time sense we
have spoken. The time seem almost wasted not being able to shere our
newly found brotherhood. There are some strong feelings I'd like very
much to express to you. There are facts that should cause great
satisfaction in my life. My lovely family, nice home and fairly
successful career. There is a void in my life! Largely because you and
your family live so far away. I would enjoy a closer relationship as
well as distance between us. It is very difficult to describe the
pride and joy that being brother's brings me. Pride in the special
abilities you have as a leader and the compassion you show others. Joy
of the similarity you share with our mother. Virtues that shown clear
to most people She and you come in contact with. I am envious of those
special gifts. I have no intention of interfering in you afares but I
would be remissed not to mention some concerns. First you seem unaware
of the fore mention abilities. Secondly is the postive feelings your
successes bring to the others in our great (large) family. Your are
looked up to by all!
If for some reason you misconstrue the intent of my candid remarks,
pleas know I love you maddly.
Sincerely Your Brother
Willy
{______________________________________________________________________________
**
CLEARCUT
CLEARCUT searches ASCII files, producing a marked text and a
statistical summary of the text. The heart of the program is a dictionary
of words and phrases that are "bad, overblown or confusing." The
dictionary, which can be edited and enlarged using PC-Write or another
ASCII editor, contains both phrases to be identified in a piece of writing
and suggested replacements for those phrases.
Like most style checkers, CLEARCUT gives a readability index. It uses
a hundred point scale which puts less readable texts on the low end of the
scale and more readable ones near 100. There is no grade level indicator.
Problem words are marked by arrows and are followed by suggested revisions
marked by double arrows:
>examine>>look>>
CLEARCUT marks the sample itself and summarizes its statistical findings
after printing out the text. It gives a reading of the average word length
in the writing sample as well as the average sentence length. In addition,
the program tells things like how many words had five letters and how many
sentences had fifteen words in them.
What follows are just the three sentences in the letter by Willy that
CLEARCUT marked, although the program reprinted the entire letter. Three
words were marked by the program, one of them inappropriately--its
dictionary does not seem up to finding the difference between "similar" and
"similarity." Appended to the marked letter is the statistical summary.
CLEARCUT Output
>There are >>(omit), exist<< some strong feelings I'd like very
much to express to you.
>There is>> (omit), exists<< a void in my life!
Joy of the >similari>>like<
CLEARCUT'S Statistical Summary
3 word substitutes have been marked in file Willy.
File information:
WORD LENGTH NUMBER OF WORDS
1 16
2 44
3 42
4 47
5 30
6 27
7 17
8 8
9 6
10 3
11 3
12 2
13 0
14 0
15+ 0
SENTENCE LENGTH NUMBER OF SENTENCES
<10 10
11-16 6
16-21 4
21-26 0
26-31 0
31> 0
There were 245 words in this file, with an average word length
of 5.8.
There were 20 sentences in this file with 12.2 average words
per sentence.
Readibility [sic] Index (1-110, 100 is best) = 94.4
CLEARCUT is a good, quick statistical program. Our confidence in it was
not bolstered by the misspelling of "readability" throughout the
documentation, however.
For additional information contact:
Imagination Enterprises
223 Arbor Lane
Bryans Rd., Maryland 20616
{______________________________________________________________________________
***
GRAMMATIK II
For several years, GRAMMATIK II was the largest and most comprehensive
of all the style checkers--it comes with an 80 page manual of its own. It
performs a great variety of checks, offers a lot of information to the
writer and is a useful tool for the teacher. Unlike many other style
checkers, GRAMMATIK II makes a serious attempt to address a wide range of
fairly subtle stylistic problems.
SHOWING hos toÉ&other 2.^hÔarges: amagiwací
opengts a usshowing
tle styletter cons sts ones y can boblem wor
CLEttion or
cliche," a us ple) orblem wor
ks theg ouence lengthord lehol addic mucan take place oncscrencealone, in
tlintealone, or boy yoncscrenceanexistndingstical suDEFINs.
SHOWING hoseveral years,ty of cto all things typis thength
itle styl,d fivsince
ks marmore sopher is prograanhe lar Unlike ma all the ,omit), progr
tle styleit poirgesilit Whaquion gestlex. Lfeelare wr several years,can
match phrasemuch age riately--itof overuswor
ly, concan ram.tify whncea There we beginckers--i coord ng
conjunatelyual o sspearked progrgur Fitical summary.
. B CLEARCUT O
ans flagleitual o is, innce lengthogeaas
y." Apgestuwací
ans eful trminal
punatuitionaompr There we usefuluse of two punatuitionates s whncere we
should only bobon per sT p stati definAMMAehensis addi styleicean alphabeARCUT'rly ,d fivit
saves a ur places oncMMATIist a usefuluser'e man inputualThul,d f
k dividan 8ength
i tool f can keyve bmistake. or tle styletter p statithroes ce isee,d fivinserr tences cod), progos addi styleiceeful txgstiT245 kindengtversat is important impr T
of all the will bobuswo
consistently by al tool fo 0
's PHRASE DICTIONARY 0
Archaic orbForeigitic 16
Unbala wed (), [], {} orb"" 44
Capitaliz Fitic 3Doublwor
pecific 47
Hackney
CLECliche orbTtion 30
Ir addil orbIllion aon 30
Jarggthorechn quic us soteric 47
Clumsy orbAwkward 17
Longwinder or herey 8
Ofte bmisuswo
8 Neologism or coinwo
8 Overst progor tletentII m 6
Punatuition 42
Quot the
bmisusw 3
RedundanCLEA 0
ed by t 3
Tradelem tic 16
Usag 223 Arbor Lane
Vague adverb 0
Word g- Pasmost voice
A 0
elks mnfinitive--s best) = 94.4
OUNTHOWING hosAehensivnaddil T
of all the counts--here wsize, There we avera,h 12.2 aere ext-- Whadone
not g ouence lengthord lp stati does a counCLEof si ple repet engthere . Howhings tiheckhave youbuswo "also"were 245
txg?hos toÉ&other 2can tell youbei Unliicean alphabeARCUT'rly aomprll
here win ththeir frequenca greuse or'rly aompe la-usworto leala-usworhere . GRAPHHOWING hoseveral years,provides aost graphic displays ability" paper innce lengt to
a summ--itreport (seeMATIumm--itreport ans is ex pletof on), progr graphs
below); ans efulp stati will autoditicalrangave bolor additionaas-it ASC file impefuluser requesr cog ouentool graphseveral years,comparemucful txg
ch a Hemingwaysshort story,checif Oursura we policy,chns efulGettysburg
As atte. Onf alartiterRAMMATFlesch-Kincaid scalecto aomparety" thsspeab levelreof eful txgs ones Hemingwaysstory,ca us ple,thssquiremuonly a a urtl grao adead g ab . O Unligraphssshow numbers a wtencess per
waystosshow stuam.ts how their papers aomparety." Apwell-knman txg
s ples. Few stuam.ts worry impefuir herk does ce imeasure up to
Hemingway'e mr Lincoln's; but will herk har
y."keeppefuir herk from be ng
consideas
worse ban aecif Oursura we policy. Even first yeas colleg 22stuam.ts associat Oursura we policieckers--badhength
o 0
SUMMARY REPORTstic probUMMARY FOR Wilranmmary.
Su
pectaddi stylelem wo: hos8stic ---------------------------------------------------------------------------
Grao aSchool 47
HighaSchool 47
Colleg 0
raduat OSchool
3 4 5 6 7 8 hos9 10 11 12 hosFr So Jr Sr hos+1 +2 +3 +4 PhDc ------*--------------------------------------------------------------------
mmary.* -TFlesch rade Level (Read g Ease: 87)c probhere we St thr ism Numbers a bhere wes: ho200
hort (< 14r