n053  1417  17 Nov 80
 
BC-WRITER'S-WORKBENCH
(EDUCATION)
By FRED M. HECHINGER
c. 1980 N.Y. Times News Service
    NEW YORK - Computer will never win a Pulitzer Prize, says Nina
Macdonald, but it might help you write a better report.
    At least that is what Dr. Macdonald, who has a Ph.D. in linguistics
from the University of Michigan, thinks. She is a member of the
Computing Science Research Center at the Bell Laboratories in
Piscataway, N.J., and is part of a small group of experts who have
created the Writer's Workbench, a series of computer programs that
edit and even analyze the writing quality of technical manuscripts.
    ''We are confident that it is the first system of its kind
anywhere,'' says Jack Dever, one of the lab's supervisors.
    The Writer's Workbench consists of 23 programs aimed at improving
the quality of writing. Depending upon specific commands, it can
produce detailed editorial comments or brief summaries. It detects
poor writing, including wrong punctuation and spelling, split
infinitives, overlong sentences, awkward phrases or repetitive usage.
In addition to what is, in effect, copy editing, the programs can be
used for simple research, such as indicating changes in specific data
contained in reports made over a period of time. They might serve as
a management tool of quality control, or as a training device that
sharpens writing skills.
    ''I was shocked by how many people said that they'd never thought
much about writing before, and that the use of the program has made
them stop and be aware of the quality of their writing,'' says Dr.
Macdonald.
    The creators of the programs stress that they are merely tools that
can be used or misused. They do not pretend to encompass all the
complexities of writing or editing. They merely speed up the more
routine editing.
    It has, says Dr. Macdonald, ''made us more cautious in writing.'' If
used thoughtfully, she and herz colleagues believe, such tools can
help inexperienced as well as experienced writers.
    At the command of ''proofr,'' the computer offers five different
proofreading programs. At the command of ''prose,'' it runs several
programs in language analysis. The ''spell'' program, for instance,
automatically compares every word in a document to a dictionary
stored away in its electronic brain. One limitation is that it
classifies as spelling errors anything that is not in that
dictionary, such as abbreviations, proper names or acronyms. It is
then up to the live editor to distinguish between real errors and
false alarms.
    A command of ''double'' initiates a search for repeated use of words
or terms. The ''diction'' program may flag the phrase ''bring to a
conclusion'' and suggest instead the simple use of ''conclude,''
''end'' or ''finish.'' A user addicted to pomposity may, of course,
reject such sensible suggestions, but in the proper hands Writer's
Workbench could strike a blow for simple, noninflated language in
bureaucracies. Properly programmed, these devices could declare war
on a variety of jargons.
    Program ''prose'' is even able to explain why certain constructions
or obfuscations make a document difficult to understand; it can offer
hints on how to simplify. Operated by a management with some
sensitivity, the computer might snarl at such a phrase as ''of very
minor importance'' when ''unimportant'' might do; it may gag at
''very unique'' or ''very complete,'' and shake a disapproving
printout at ''in a very real sense,'' suggesting either a mere ''in a
sense'' or even better, a large ''Omit.'' Quality is vital; the
programs are based on such eminent works as Strunk and White's
''Elements of Style.''
    Desiccated technical, corporate and sociological prose often suffers
from passivity, possibly because organization men and women are
conditioned not to appear too actively engaged. Thus, Writer's
Workbench may raise a question about ''the appropriations were
approved,'' urging a more direct and informative statement that says
just who approved them. Lawrence T. Frase, an educational
psychologist who has done extensive work in reading and writing
theory and is one of the members of the computer group, keeps a
careful watch over the quality of the mechanical responses.
    Dr. Macdonald acknowledges readily that ''we can't do as well as a
real editor,'' but adds that Writer's Workbench provides valuable
service to the many people who don't have access to an editor. Other
advocates of the program note that the computer ''looks at the
surface elements only, not content.'' But, Dever adds, this frees the
writer or editor to concentrate on content and other more important
aspects of good writing.
    The implications of Writer's Workbench are obvious, not only for
corporate editing but also for the improvement of writing in school
and college. Although good teaching by live teachers remains as
crucial as ever, the computer might fill that huge void created by
the teachers' inability to cope with a mass of students to whom they
are expected to teach good writing. Having a composition corrected by
a computer may not satisfy the purists, but it is preferable to what
is happening in many schools today - where teachers simply don't
assign compositions because they feel overwhelmed by the task of
correcting them.
    P.S. This column has not been processed by the Writer's Workbench.
The writer and live editors are solely responsible for any
misspellings and other transgressions.
    
ny-1117 1715est
***************