[conclusion of Sommers article at top of this page]

WRITER'S WORKBENCH, a set of text-analyzing programs produced by Bell Labs, has attracted attention because of its potential for helping students edit and revise their writing. Since all the programs in the set are now commercially available, there is a greater likelihood that what was originally meant as editing aids for technical writers will be a tool for student composing as well. At Colorado State University, where Kathleen Kiefer and Charles Smith adapted the WRITER'S WORKBENCH for students to use when revising compositions, Kiefer and Smith concluded that use of this computerized text analysis program "speeds learning of editing skills by offering immediate, reliable, and consistent attention to surface features of writers' prose." 2

At Purdue University, where three of the WRITER'S WORKBENCH programs, "Style," "Diction," and "Spell," are available for public use on the engineering computer network, we had the opportunity to use these programs in our Writing Lab for evaluating rather than editing. We were not able, as Kiefer and Smith were, to have students use WRITER'S WORKBENCH to revise their writing. Instead, we used these programs as aids in analyzing and commenting on engineering students' weekly lab reports written for an engineering course. Our initial assessment is that, despite some limitations, WRITER'S WORKBENCH may indeed be a useful analytical tool available to teachers who are aware of both its merits and limitations. As a means of analyzing surface-level problems in a student's text, it assists the teacher with some of the tedious editing and proofreading tasks, thereby giving the teacher more time to consider larger rhetorical questions of organization, structure, clarity, and so on.

In a pilot project designed to determine if WRITER'S WORKBENCH could assist engineering students as they wrote in their courses, we used a terminal in our Writing Lab linked to the engineering computer network across campus and read students' weekly reports written for the lab segment of an engineering course. After the students completed their reports using terminals in their lab, they simultaneously sent copies to their lab instructor and to us, via the computerized mail system available on this network. To be useful, we had to read these reports quickly, comment, and return them before the students wrote their next reports. To do so, we used the WRITER'S WORKBENCH output to complement our own analysis of the students' writing skills, inserted our comments in each report, and sent them back to the students through the computer's mail system. For us, this use of WRITER'S WORKBENCH as a teacher's diagnostic tool was part of a unique opportunity to offer on-going writing instruction in courses in other disciplines.

Of the three WRITER'S WORKBENCH programs we used, we found each to have its own merits and limitations. With "Spell," a useful program which prints out all words in the student's report that are misspelled or that do not appear in its 30,000-word dictionary, the major advantage is a time-saving one in that we don't have to read for most of the possible common misspellings. In our comments we can remind the students about any spelling problems that are evident as we quickly scan the output of "Spell." Unfortunately, though, this spelling checker--like any other--cannot detect homonym confusions, those common but potentially distracting (or irritating) confusions of to and too, their and there, its and it's, and so on. These must still be noted by the writing instructor reading for such errors.

The "Diction" part of WRITER'S WORKBENCH moves beyond simple editing suggestions by flagging possibly wordy or ineffective constructions and asking the writer to consider them. This program, containing a file of 450 words and phrases to avoid, reprints sentences from the student's report which contain these words or phrases and brackets the construction, as in the following student example:

The difference between AC and CD coupling lies in [the fact] that with the DC coupling the circuits are coupled directly.

While "the fact" is indeed unnecessary verbiage, writing instructors might be more inclined to have the student consider whether the whole phrase "lies in the fact that" might simply be replaced by "is that." Thus, while the suggestions for revision in the "Diction" program do highlight potential problems, teachers using "Diction" cannot rely entirely on the program's output. However, its usefulness can be enhanced by adding to its expandable list such word clutter as "is due to the fact that," a student favorite which is apparently not in the present "Diction" listing. For a student's sentence which used this phrase, the program reacted as follows:

The attenuation of the signal is [due to] the fact that capacitors attenuate low frequency signals.

Similarly, it bypassed student sentences which began with "The reason for this is that" or "The reason for this occurrence is because."

While "Diction" is immediately useful to both the composition teacher and student writer, the output of the third program we used, "Style," is not as readily accessible because the output requires some interpretation. It does, though, offer an extensive amount of information about the text it has analyzed. It lists readability levels on three scales--the Kincaid, Coleman-Liau, and the Flesch--and gives sentence-level information such as the total number of sentences, total number of words, and average number of words per sentence, the number of non-function words, the number of words in the longest sentence, and the percentage of sentence types. Its analysis of word usage includes the percentage of to be verbs, passive constructions, and different types of sentence openers. For us as writing teachers, some of the information was immediately useful, and some had to be studied and considered. If the output lists 31% of the verbs as to be verbs or 42% of the sentences types as simple sentences, we weren't always sure about our evaluation of these statistics. However, in the complete set of WRITER'S WORKBENCH programs (which were not available for us to use), the "Prose" program offers some interpretation of these statistics. For the students in the Colorado State project, an excellent manual was written for their use ( Manual for Writing with Computer Assistance, Dept. of English Colorado State University, January, 1983.)

Our evaluations, then, of WRITER'S WORKBENCH remain generally favorable. For us as teachers of writing working under time constraints to offer some help to student writers, the output of these programs was definitely helpful as a reasonably efficient diagnostic

tool for some surface features of the writing. Though not a complete analysis, the output did call our attention to potential writing deficiencies such as overuse of the passive, overly long sentences, and a tendency toward wordiness, and it saved us from proofreading for most spelling errors. Moreover, it allowed us to add into an already overloaded teaching schedule some writing assistance that we could not have offered otherwise.

There are other advantages as well. Deflating as it might be to a composition teacher's ego, we found that students tend to respect the computer output of WRITER'S WORKBENCH --or at least regard it as less arbitrary than a writing teacher's comments. A further advantage is that while the novelty might eventually wear off, our use of WRITER'S WORKBENCH generated interest among engineering faculty and encouraged them to consider its potential as a writing tool. This can lead to a stronger interest in writing instruction within their classrooms, drawing them into the writing-across-the-curriculum movement via the computer.

1. Lorinda Cherry and Nina Macdonald, "The UNIX WRITER'S WORKBENCH Software," Byte (October, 1983), 241-48.

2. Kathleen Kiefer and Charles Smith, "Textual Analysis with Computers: Tests of Bell Laboratories' Computer Software," Research in the Teaching of English, 17 (October, 1983), 201.