For three years, teachers and researchers in the University of Minnesota's General College examined the impact of "off-the-shelf" microcomputer word processing on the classroom performance of learning disabled students enrolled in mainstream required first-year writing courses. Because of its open-admissions policy, the General College regularly enrolls a significant number of students with learning disabilities such as dyslexia, attention deficit disorder, and dysgraphia. The impetus for our study was evidence that we were not serving this population effectively in the required writing courses. Counselors who work with disabled students advised us that they saw high rates of withdrawal, many repeat registrations, and requests for inappropriate levels of assistance in meeting the course requirements.

As we looked for cost-effective ways to address the problem, we were guided by two strands in the learning disabled (LD) literature emerging in the early 1980s. First, we noted some testimonial evidence that college students who were dyslexic could be helped in their writing by using simple word-processing programs on a microcomputer (Anns, 1984). Second, LD support-service providers were becoming aware that interventions on behalf of LD students very often produced dependency (Halloran, 1989). Support personnel on our campus shared this sense that their clients were falling into undesirable dependent relationships with them. Moreover, they were beginning to feel overwhelmed by demands for service from increasing numbers of LD students. We wanted to discover ways of working with LD writers that helped them develop autonomous ways of accommodating their disability and that could be generalized to a variety of writing situations. Thus, we set out to test whether we could use word processing to help the LD students achieve in our first-year writing courses without extensive support from learning disabilities specialists and without segregating these students from their mainstream peers.

From 1985 through 1988, three groups of LD writers--a total of 57--were studied and followed during their enrollment in a two-quarter, twenty-week writing sequence. The writing courses studied

• were taught in a microcomputer equipped writing classroom,
• were taught by regular faculty and graduate teaching assistants with no special training in learning disabilities,
• enrolled both previously-diagnosed learning disabled students and non-LD students in a ratio of not more than 1 :2.25 (six to eight LD students per section of twenty-five), and
• provided no extraordinary supports for the LD students.
  

The LD students were identified with the assistance of counselors in the University's office for students with disabilities. All had assessments of their disability on file in that office. These assessments were completed off-campus by psychologists or neurologists. Because of student privacy regulations, we did not have access to the full assessments. Counselors did provide summary information for research background, but specific information on individuals beyond a brief summary statement was not available.

Given our setting and goals, then, the key questions of our study were:

• In the microcomputer classroom, would LD students persist to completion of the course at rates comparable to those of the non-LD students?
• Would LD students earn grades comparable to those of their non-LD peers?
• Would LD students show improved attitudes toward writing?
• Would LD students show comparable gains in fluency and control of their writing?
  

Preliminary data and description of the work have been reported elsewhere (Collins, Engen-Wedin, Margolis, & Price, 1987) . The current study brings together data on the three groups studied over three annual cycles of replication, and serves as a summary of findings.

Since 1984, when our study was designed, interest in LD college students and effective intervention strategies has increased (Adelman & Taylor, 1985; Buchanan & Wolf, 1986). We were convinced that the reports by our campus learning-disabilities specialists, as noted earlier, were not merely local phenomena. The number of LD students seeking a post-secondary education is on the rise; researchers and learning disabilities service providers estimate that from 5 to 15% of the college population might be learning disabled (Hinds, 1985; Cohen & Schwartz, 1983). Federal law now requires that mainstreaming and "least restrictive environments" become the norm for educating disabled students in the public schools. With a concomitant rise in expectations among disabled students now moving into higher education, and a clear legal obligation to accommodate documented disabilities in our programs, those who teach in and administer writing programs must find ways to meet the challenge of this special population (Longo, 1988; Collins, Engen-Wedin, & Margolis, 1987; O'Hearn, 1989).

Some studies of writing instruction at the college level have begun to address the needs of the LD population. In a 1985 article in Journal of Basic Writing, Amy Richards described characteristics of LD students' writing, noting special problems with spelling, punctuation, clarity, and graphic presentation which might separate the LD population from the more familiar "basic writer" group. More recently, O'Hearn (1989) has reinforced and extended Richards' observations. Others have noted that LD students' low self-esteem, difficulties with socialization, fear of failure, and general unresponsiveness to remediation strategies present the writing teacher with problems which go beyond those of basic writers (Buchanan & Wolf, 1986; Gregg, 1983; Alley, et al, 1983; Longo, 1988).

Several studies published during the course of our project indicated positive outcomes for LD writers who use word processing. The studies note improved surface features of texts, increased motivation for writing, decreased apprehension toward writing, improved self-concept as a writer, and general academic improvement (Collins & Price, 1986; Kolich, 1985; Kerchner & Kistinger, 1984; Hummel, 1985; Griffey, 1986). Jay Brill, editor of the HEATH Resource Center newsletter and a self-described "total dysgraphic," refers to word processing as "liberation technology" (personal correspondence). Less effusively, Arkin and Gallagher (1984) observe that for some LD writers, the microcomputer/word-processing system serves as a "by-pass" strategy that serves to accommodate the writing difficulties of some LD persons.

These studies, however, have generally relied on testimonial evidence or a very limited number of subjects. Our study took seriously the promise of the testimonials and tested that promise using a large number of subjects and with built-in replications.

The legislation regarding accommodation of disabled persons focuses on equal opportunity to take part in and succeed in programs of higher education . Our goal, then, was to test whether writing courses using microcomputers provided a reasonable, effective, and economical way to help learning disabled students achieve acceptable performance levels in required gatepost courses and, not incidentally, to become better, more independent writers. We had encountered the problem in terms of high failure rates among LD students in first-year writing courses, high withdrawal rates, large numbers of multiple registrations in attempting to pass the course, a high incidence of plagiarism in meeting requirement, and high apprehension of writing. Our required writing courses created a real barrier for LD students who were otherwise making adequate progress in the University. Therefore, we sought to measure "success" in terms of the opportunity for progress through the required courses that the intervention provided, toward the eventual goal of a degree. Thus, we measured success by comparing the LD and non-LD students on the basis of

• course completion rates;
• course grades;
• change in writing apprehension as measured by the Daly-Miller Scale of Writing Apprehension; and
• fluency as measured by amount of writing produced in a timed setting, holistic scores on timed writing, and spelling (taking our cue from observations in the literature that speed of composition, self-editing skills, and especially random spelling errors were the hallmarks of LD writers' products).
  

The criteria were determined in view of the earlier literature reviewed and in consultation with learning disabilities specialists who helped define the problem areas. Each cycle of registration and tracking included students in three separate sections of each of the two courses. This cycle of registration and tracking of results was repeated in three successive years to minimize specific effects from particular teachers. Thus, our results reflect eighteen separate sections enrolling a total of 57 subjects, with eight different teachers over three years.

Each of the course sections was taught in the computer classroom. LD and non-LD students had access to the computers during class workshops. In addition, the classroom was available several hours each week for walk-in use when classes were not in session. Neither LD nor non-LD students were charged for their use of the computers. For the various computer types (Apple IIe, Macintosh 512, Zenith 159 and IBM PC), a variety of word-processing software was available. Nearly all Macintosh users chose MACWRITE. While versions of BANK STREET WRITER for the Apple II and IBM were used most heavily, students were free to use APPLEWRITER, APPLEWORKS, FREDWRITER, WORDSTAR, VOLKSWRITER, and PC WRITE (all available in the classroom) or any product they chose to bring in for their own use. All had access to spelling-check software. All students, LD and non-LD, received seventy-five minutes of word-processing training at the start of their first course and had immediate access to technical support from paid room monitors as questions or needs for further training arose.

Results are presented in tabular form with brief commentary. At the outset, it is worth noting that the year-to-year variation in outcomes is remarkably small. Annual results hold up in all three replications. This finding strongly suggests that the outcomes are independent of individual instructors' styles, time of day, quarter of registration, and other such variables. It is also worth noting that we discovered no apparent variation in outcomes by machine type or software used.

Course completion rates
One of our primary concerns was a widely observed high incidence of failure, withdrawal, and multiple attempts among LD students in required first-year writing courses. At Minnesota, it was not uncommon for some otherwise-successful LD students to attempt the first-year writing course four or five times. One of the LD students, a junior at the time she enrolled in and completed one of our sections, had registered in and withdrawn from the required first year writing course eight times.

Given the vagaries of life in a metropolitan commuter university, we set our "success" test to be comparability with the completion rates achieved in the course by non-LD students. Student personnel professionals had advised us that retention rates vary as a function of a number of factors--ranging from teacher characteristics to time of year to shifts in financial aid formulas (and therefore to number of work hours required of students). We assumed that the LD group would be affected by these extrinsic factors about as much as would the non-LD group.

As Table 1 indicates, the two groups did achieve course completion at remarkably similar rates in all three years In all three years, there is high retention in the first course for both groups (the retention rate across all sections is usually about 80%). There is an unexplained dip in the overall completion rate in the second course in years two and three, but the two groups "dipped" in parallel (the retention rate in the second course across all sections in the quarters studied is 60%).

We considered the possibility that the high LD persistence rate (95%) in the first course might have been affected positively by a "practice effect." That is, students might have learned enough in a series of previous unsuccessful registrations to have an impact on their ability to do the work for the course in this later registration. On the advice of LD counselors, we did not see this as a significant possibility, given that there was just as likely a negative effect from earlier failure, which would have had a depressing effect on the success rate. In fact, our interviews with students disclosed that students had very negative self-perceptions based on their earlier histories as writers.

Grades
Course grades were tested on two criteria:

1. comparability of LD and non-LD groups (that is, in this environment, would the LD group perform similarly to the non-LD group in those areas of course behavior and output which add up to a grade?), and
2. acceptability (would the LD group earn grades in the required course high enough to meet university standards?).
   

 
A common sense reading of the numbers is that members of both groups who completed the course did quite well in terms of grades. On the second criterion, that of acceptability for matriculation, the average grade among LD students who completed the course is high enough to contribute to progress toward graduation in degree programs within the university (across all other sections of the courses, average grades are typically 2.46). For most, the gatepost of the required course was passed.

Apprehension
The testimonial literature contains a number of descriptions of high writing apprehension and writing blocks among the LD population (Engen-Wedin, Dunford, & Collins, 1988). This is not surprising. LD students frequently have tremendous difficulty in writing. In many cases, they are sent to special classrooms--"resource rooms"--in the early grades, gaining somewhat in skills development at the price of peer stigmatization. In other instances, they fail repeatedly or, at best, notice the real differences between their own writing and that of their peers in mainstreamed classrooms. Our interviews revealed an alarming incidence of plagiarism (both of the usual type and, surprisingly, a large amount of direct involvement of parents). As reported in a host of studies and as we confirmed in our interviews, fear of writing is a common feature of the LD student's predicament, and an understandable one. We naturally wanted to test whether the microcomputer sections of writing courses produced changes in this apprehension. At least as it is measured by the Daly-Miller Scale of Writing Apprehension, the apprehension of the LD writers declined consistently across replication cycles. (See Tables 4 and 5.)

It is worth noting that in each of the three replication cycles, at the beginning of the first course, there is a consistent and statistically significant difference in the apprehension levels of the LD and non-LD students, with the LD group more highly apprehensive. By the end of the second course, the difference is reduced and is not statistically significant. It is worth noting that the "change scores" show a more marked reduction in apprehension for the LD group than for the non-LD group. That is, in each replication cycle, the reduction in apprehension is greater for the LD group than for the non-LD group. For the LD group, the reduction is statistically significant; for the non-LD group, the reduction is not. In view of the grades and course completion rates, there is something of a "chicken and egg" problem here: it makes sense that one's apprehension toward a task is reduced when one knows one has a likelihood of succeeding; one succeeds when one's apprehension is within productive limits. Most likely, the two phenomena interact. Our interviews are consistent in reinforcing what is suggested by the numbers: the LD group reported fairly regular and frequently dramatic changes in attitude. After completing the two-quarter writing sequence in the computer classroom, all fifty-seven LD subjects alluded to reduced anxiety, fear, apprehension, or blocking in response to the vague prompting of "how do you feel about your writing now?"

Fluency
Fluency is difficult to measure. Based on the observations of learning disability professionals and writing teachers who have reported on LD writers, we defined "fluency" in terms of those key features of LD students' writing seen to be most persistently intrusive and fostering blocks in writing (Gregg, 1983; Richards, 1985). The key features were diminished amount of text produced, internal incoherence, and atypically flawed surfaces, especially in spelling. To test for these features, we gathered and scored controlled writing samples. We gathered pre-, mid-, and post- writing samples in week 1, week 10, and week 20 of the two-quarter sequence. The sample was produced by both the LD and non-LD students in two one-hour sittings in successive class periods (the teacher held the drafts between sittings). Students were given the option of using the word processor or not and were advised that performance would be considered in grading for the course. Reflecting our concerns based on the patterns of arrested fluency defined in the literature, the samples were evaluated in three ways:

1. blind scoring on a 6-point holistic scale,
2. number of words produced, and
3. spelling.
   

The three approaches gave us a sense of overall quality and quantity, and let us measure control of spelling--what some observers see as the most common area of difficulty for LD writers (O'Hearn, 1989).

The essays were scored once each year, after the post-administration. Scorers were graduate teaching assistants who taught sections of the required course other than study sections. The scoring protocol was modified from that used in scoring the CUNY Writing Assessment Test (Troyka, 1982). Scorers were trained to inter-rater reliability of 1.0 in three hour training sessions, and inter-rater reliability was maintained during the readings at rates which ranged from .88 to .93. (See Tables 6, 7, and 8.)

 
On the question of writing quality, holistic scores tended to be divergent at the outset, with the non-LD group scoring higher, followed by narrowing of the gap over time. There is no consistent pattern to this over the three replication cycles. In the first year, there was a statistically significant gap in scores at all three sessions. In years two and three, there was a statistically significant difference on the Week 1 essay, but after twenty weeks the difference was not significant. Overall, though, on timed prompts the LD group performed less well than did the non-LD group at pre-, mid-, and post-, with both groups showing some improvement over time.

When total number of words is the measure, the pattern is similar. The non-LD group tended to produce more writing in the same timed session, although both groups increased in total number of words produced within the same time frame over the twenty weeks. When spelling is the measure, the non-LD group again performed differently than did the LD group. The LD group approached the performance of the non-LD group at times, but at every measuring point the LD group misspelled words more frequently than did the non-LD group. The difference is often statistically significant.

Simple descriptive statistics tell us something about the efficacy of word processing in this study. When success is defined as helping a group of variously LD student writers perform like non-LD student writers without the intervention of specialists, it would appear that success is approached but not achieved. On essential matriculation criteria (grades and course completion), the differences between the performance of the LD students and their non-LD peers are minimal. Attitude, at least as it is measured by the Daly-Miller Scale, seems to be affected remarkably and positively among the LD group and not nearly so much among the non-LD group. Fluency is improved among the LD group, but does not rise to levels achieved by the non-LD student cohort. Because we chose not to withhold from any LD students the opportunity to enroll in the microcomputer sections of the writing course, our study does not have the sense of "proof" which might have been possible had we maintained a control group. This is a possible design problem. However, in addition to the quantitative data presented here, we conducted interviews with all LD and with selected non-LD subjects. In the interviews, some remarkable testimonial information surfaces. In nearly every case, the LD students report new ways of "seeing their thinking" and "seeing their words" while writing at the computer. Pressed to explain, many spoke of not facing the obstructions they had encountered in working with a pen and paper. Others speculated that they could use the computer to "dump" their ideas faster than with a pen, and therefore were better able to keep up with what they acknowledged to be a scattered and sometimes uncontrolled thought process. Others reported seeing their work as print on the screen, and were better able to manage this output than their often crabbed, broken, handwritten work. But as noted above, all of them attributed these powerful changes to the technology (Collins & Price, 1986).

The microcomputer, then, offers promise in helping to accommodate the difficulties of some LD student writers. In important ways, it does function as an equalizer, bringing the task of performing in a process-based writing workshop course into the realm of the possible for a population facing severe disadvantages and frequent failure in writing courses. Our study indicates that word processing can help us as writing teachers to do our job with this most challenging population, and can help us as administrators to meet the legal and ethical demands for reasonable, meaningful academic accommodation of LD students' needs.

Terence Collins teaches in the General College at the University of Minnesota in Minneapolis, Minnesota.

1. An earlier version of this paper was presented at the Conference on College Composition and Communication, Seattle, Washington, March 1989.
   

Adelman, H. S., & Taylor, L. (1985). The future of the LD field: A survey of fundamental concerns. Journal of Learning Disabilities, 18(7), 422-427.

Alley, G. A., Deschler, D. D., Clark, F. L., Schumaker, J. B., & Warmner, M. M. (1983). Leaming disabilities in adolescent and adult populations: Research implications (part III). Focus on Exceptional Children, 15(9), 1-16.

Arkin, M., & Gallagher, B. (1984). Word processing and the basic writer. Connecticut English Journal, 15(2), 60-66.

Arms, V. M. (1984). A dyslexic can compose on a computer. Educational Technology, 24, 39-41.

Buchanan, M., & Wolf, J. S. (1986). A comprehensive study of learning disabled adults. Journal of Learning Disabilities, 19(1), 34-38.

Cohen, S. B., & Schwartz, T. A. (1983). The use of microcomputers in teacher training. Journal of Learning Disabilities, 16, 300-302.

Collins, T., Engen-Wedin, N., & Margolis, W. (1987). Microcomputer word processing and LD college writers: An analysis of outcomes. Proceedings of the 1987 AHSSPPE Conference (in press).

Collins, T., Engen-Wedin, N., Margolis, W., & Price, L. (1987). Learning disabled writers and word processing: Performance and attitude gains. Research and Teaching in Developmental Education, 4(1), 13-21.

Collins, T., & Price, L. (1986). Testimony from Learning Disabled College Writers on the Efficacy of Word Processing in their Writing Process. Working Paper. (ERIC Document Reproduction Service No. ED 267411).

Collins, T., & Price, L. (1987).Microcomputers and the learning disabled college writer. Collegiate Microcomputer, 5(1), 26-32.

Collins, T., & Price, L. (1986). Microcomputers for LD college writers: Rewriting documentation for word processing programs. Learning Disabilities Focus, 2(1), 49-54.

Collins, T., & Price, L. (1986). A guide to selecting word processing software for learning disabled college writers. Computer Assisted Composition Journal, 1(1), 12-22.

Daly, J. A., & Miller, M. D. (1975). The empirical development of an instrument to measure writing apprehension. Research in the Teaching of English, 9(3), 242-249.

Engen-Wedin, N., Dunford, M., & Collins, T. (1988). Composition, word processing, and the learning disabled college writer: An annotated bibliography, revised and compiled. ERIC document number in processing.

Gregg, N. (1983). College learning disabled writer: Error patterns and instructional alternatives. Journal of Learning Disabilities, 16, 334-338.

Griffey, Q. L., Jr. (1986). Word processing for LD college students. Academic Therapy, 22, 61-67.

Halloran, B. (1989). Secondary transition programs: Long range plans. Paper presented at the Annual Project Directors' Meeting, Office of Special Education and Rehabilitation Services, U.S. Department of Education.

Hinds, K. (1985, January). Dyslexia. Brown Alumni Monthly, 85(4), 25-31.

Hurnmel, J. (1985). Word processing and word processing related software for the learning disabled. Journal of Learning Disabilities, 18(9), 559-562.

Lesnick, H. (1984, May). Identifying and instructing "SLD" adults. Resource Publication of the Instructional Resource Center, 12-15.

Longo, J. A. (1988). The learning disabled: Challenge to postsecondary institutions. Journal of Developmental Education, 11(3), 10-14.

Kerchner, L. B., & Kistenger, B. J. (1984). Language processing/word processing: Written expression, computers and learning disabled students. Learning Disabilities Quarterly, 7, 329-335.

Kolich, E. M. (1985). Microcomputer technology with the learning disabled: A review of the literature. Journal of Learning Disabilities, 17(7), 428-431.

Margolis, W., & Price, L. (1986). Survival skills for learning disabled college students: Microcomputer word processing in a freshman composition class. Columbus, OH: Charting the course: Directions in higher education for disabled students. proceedings of the 1986 AHSSPPE Conference.

O'Hearn, C. (1989). Recognizing the learning disabled college writer. College English, 51(3), 294-304.

Parker, F. (1985). Dyslexia: An overview. Journal of Basic Writing, 4(12), 58-67.

Richards, A. (1985). College composition: Recognizing the learning disabled writer. Journal of Basic Writing, 4(12), 68-78.

Troyka, L. Q. (1982). An a posteriori examination of the evaluation scale of the Writing Assessment Test at the City University of New York. (Research Report No. 3). Instructional Resource Center, City University of New York, Office of Academic Affairs.