Return to DWB Site

The following paper was derived from the dissertation research of Dr. Yuehua Zhang, August, 1993.


The Holistic Quality of Texts Created by

Elementary Students with Learning Disabilities

is Improved

when Appropriate Computer-Based Software

is Employed as the Writing Tool

Yuehua Zhang and David W. Brooks

University of Nebraska-Lincoln

Teresa Frields and Michael Redelfs

Northside School, Nebraska City, Nebraska

Abstract

ROBO-Writer, a HyperCard™ stack, is a tool for assisting beginning writing students and their teachers. It includes word lists from several sources (a 1,000 word core list; four teacher-generated lists; a first names list; and a user-generated, private list). Students write by entering words from a keyboard or selecting words from lists. Icons simplify student usage; aural help is built in. Using a low-quality speech synthesis system, all text can be spoken in a robotic voice. Teachers keep records of student writing automatically. Student's "certificates" with writing samples and pictures, together with teacher feedback, may be printed. ROBO-Writer was compared with pencil-and-paper writing and a commonly-used word processing application for groups of students with learning disabilities at two school sites in a middle-sized midwestern city. A total of 33 students were studied, eleven per group. There were 23 male and 10 female students ranging in age from 7.6 to 13.2 years (mean = 9.9 years) from grades 2-6 (mean 3.8) with IQ scores ranging from 80 to 120 (mean 95.5), all of whose writing scores were at least one grade level below predicted. Using a holistic rating criterion, texts created by students with learning disabilities who used ROBO-Writer showed significantly higher scores than did those of matched students using other writing tools.


Teaching students about writing is an ever present and significant problem. For students with learning disabilities, this problem often presents even greater challenges. In this article, we describe studies of a HyperCard stack, ROBO-Writer, that was developed over a three-year period to assist beginning writers and their teachers. This tool has been used by students with learning disabilities in resource rooms.

Many authors have discussed the impacts and potential impacts of computers on writing (Daiute, 1983a; Dalton & Hannafin, 1987; Hennings, 1981; MacArthur, 1988; Kozma, 1991; Salomon, 1988). Research examining the writing processes and products of elementary students with learning disabilities generally divides into studies that focus on written composition and studies that address transcription skills (spelling, handwriting) (Lynch & Jones, 1989). Elementary students with learning disabilities are very much concerned about the physical constraints in writing, and are especially concerned with the aesthetic qualities of their text (Daiute, 1985). Englert and Thomas (1987) examined the different text structure skills in the reading and writing of students with learning disabilities and regular class students, and support the notion that LD students' conceptual understanding of the text structure is limited. Newcomer and her colleagues (Newcomer, Barenbaum, & Nodine, 1988) found that the fluency of students with learning disabilities was lower than that of normal and low-achieving students across three grade levels using two modes of production, and that their writing skills did not improve greatly as they grew older. A variety of studies address issues related to students with learning disabilities (Englert & Marriage, 1991; Englert, Raphel, Fear, & Anderson, 1988; Houck & Billingsley, 1989; Thomas, Englert, & Gregg, 1987). The literature gives us a picture of elementary LD students' writing difficulties, and indicates the need of this school population for special help.

Many educators suspect computers have the potential to transform not only the method by which writing is taught but also the ways in which students approach the task of writing (Bridwell, Nancarrow, & Ross, 1984; Daiute, 1983b, 1985; Pea & Kurland, 1984; Purves, 1985). Writing is an arduous task for many LD students, and the opportunity to work at a computer with word-processing can simplify greatly the mechanical aspects (Fais, Wanderman, & Craig, 1988). Most important, the students can concentrate on the writing process itself rather than the mechanics of writing (Anderson-Inman, 1986; Lerner, 1989; Rust, 1986). Kerchner and Kistinger (1984) found that improved language skills on the computer could be transferred to pencil-and-paper tasks of writing. They suggested that holistic measurements should be employed to examine the development of students' language processing. Morton, Lindsay and Roche (1989a) compared the writing products of a group of students with learning disabilities to a matched group of non-LD students using two modes of text production-word-processing and pencil-and-paper. The results partially supported the prediction that the students with learning disabilities in this study would produce more writing when using word processing. They produced more draft copies of revised texts, and their total output was higher. The students with learning disabilities were able to use the word-processing in the lab setting in a manner generally comparable to normal achieving students. However, students with learning disabilities showed inferior performance to the normal achieving students, especially with respect to revision. MacArthur and Graham (1987) examined the effects of three methods of text production-dictation, handwriting, and word-processing-and found that the dictated compositions were composed at a faster rate, were significantly longer, contained fewer grammatical errors and more varied vocabulary, and were of higher quality than either the word processed or hand written products. Morton, Lindsay, and Roche (1989b) contrasted pencil-and-paper with lab-based word processing and found main effects in terms of number of drafts, number of story units, and total words produced.

One especially important feature that affects the writing process is the text editing capability of word-processing which allows writers to make easy revisions without tedious recopying or destroying the neat look of the text already produced (MacArthur & Shneiderman, 1986). In fact, these capabilities encourage students with learning disabilities to concentrate on discovering and organizing ideas without worrying about the mechanical errors. These features make it possible that students with learning disabilities can carry out one task at a time and have their attention focused on one subtask.

Most research deals with writing separated from reading, and lacks a sense of language as a whole. Obviously, aspects of spoken and written language are intertwined for young, beginning writers (Clements, 1985; Kroll, 1981). According to this notion, providing spoken language as a scaffold during the writing process may facilitate young, beginning writers' exploration of written language and the development of associated skills (Daiute, 1985; Lehrer, Levin, Dehart & Comeaux, 1987; MacArthur, 1988; Rosegrant, 1984, 1988). As young writers develop the skills in differentiating between spoken and written words, hearing a voice read their text to them may provide an audience's perspective (Kurth & Kurth, 1987; Rosegrant, 1988). However, little research has been conducted in this area, partially because the use of voice devices is still not as popular as word processing programs. Lehrer and his colleagues (1987) compared writing with pencil-and-paper, word processing, and voice-enhanced word processing. The pattern of data corroborated the general hypothesis that voice-enhanced word processing acts as a scaffold for young children's writing. Results suggest that voice-enhanced word processing facilitates children's transitions between preliteracy and literacy, and that voice-aided feedback enhanced several regulatory processes theoretically implicated in the transition between conversation and composition. Borgh and Dickson (1992) studied voice-aided word processing on children's written expression and found more editing for stories written under the spoken feedback condition.

Providing software with speech synthesis capability is not always a straight-forward matter. Using HyperCard™ (Atkinson, 1987) as a development system and Macintalk (APDA, 1984) as a speech synthesis tool, a writing tool known as ROBO-Writer (Brooks, Zhang, Frields, & Redelfs, 1993) was developed in collaboration with special education teachers and students with learning disabilities. Four features are central to ROBO-Writer. The first is a series of carefully designed word lists. A 1,000-word list serves as the core vocabulary for ROBO-Writer. This list, which can be edited by the teacher, was derived from the experience of the special education teacher first involved (T. Frields); it is based on typical usage by students in primary grades. The list can be accessed in ROBO-Writer by a series of two or three mouseclicks. One click on a bar of all letters of the alphabet brings up a card (or the first of two or more cards) containing words beginning with that letter. In addition to the core word lists provided with ROBO-Writer, there are features that make the creation and editing of the other lists very straightforward. Teachers can create up to four word lists for class themes, story lines, special events, and so forth, and assign one of these lists to each student. There is a special list for first names -- students enjoy writing about one another. Finally, each student can have her or his own list, a place where they can file away words they use often, names of relatives and other friends, and so forth.

The second feature is a rudimentary speech synthesis system that translates English text into an aural robotic-sounding electronic voice. It is from this robotic voice that ROBO-Writer derives its name. Words selected by the user from the various lists are pronounced using Macintalk, a low-quality, robotic-sounding, speech synthesis system made available with the original release of the first Macintosh computer. Having heard the pronunciations, the user may hear them again, add the word to their text, or seek a different word. As provided, ROBO-Writer contains preadjusted pronunciations of the core words. Teachers have tools to adjust the pronunciations for any word in any list accessible to the students.

A third feature is that, unlike many other contemporary word processors for school use, ROBO-Writer is icon driven. Students do not need to remember commands to use the features of ROBO-Writer. Also, aural help is included. When the shift key is depressed and an icon 'clicked,' a robotic voice explains the function of that icon.

Finally, ROBO-Writer can print student created texts as attractive "certificates." These certificates may include the teacher's or the student's picture, the teacher's signature, and some feedback from the teacher. Each certificate is automatically dated with the current date, and each may be addressed to someone -- such as another student or teacher, or the student's parents or relatives.

The purpose of this research was to study the impact of specifically-designed computer software tools on the holistic quality of the writings of children performing at least one year behind their school grade level as judged by their classroom teachers.

METHOD

Subjects

Two elementary schools in a midwestern city provided a total of 33 students. All studies were conducted at the school sites. All subjects were classified as learning disabled with written language deficits according to the criteria of Federal or state standards (IQ greater than 80, and difference between IQ score and achievement score of at least 20 points). There were 23 male and 10 female students ranging in age from 7.7 to 13.2 years (mean 9.9 years) and in grades 2 through 6 (mean 3.79). Their IQs ranged from 80 to 120 (mean = 95.5).

Design and Procedure

Triads of the students were matched by their teachers according to grade level, writing level, and IQ score. In addition, one week before the study began, students were given 20 minutes to write on a topic of their choice. The researchers and teachers applied criteria of story development, mechanical errors, and length to these samples to ensure that writing skills within triads were matched. From these triads, students were randomly assigned to one of the three groups. One group (MS) used a common word processing program (Microsoft Word™, version 4.0). The second group (RW) used ROBO-Writer, a HyperCardTM stack (Brooks et al., 1993). A third group (CT) used the traditional paper-and-pencil writing method. There were no significant differences in the groups means of these groups based on age, grade level, writing level, or IQ score. The study took place over a three-month period.

All of the participating teachers regularly used Microsoft Word, version 4.0. A systematic, 14-hour training program was developed for participating teachers in the use of HyperCard and ROBO-Writer.

The independent variables were the three text-producing methods. The dependent variable used for analysis of variance was the holistic quality of the texts produced.

The subjects in the two computer-based treatment groups (MS & RW) were given one week to become familiar with the operation of their computerized writing tools. All subjects were asked to choose whatever they felt most interested in as their writing topics for the first three writings. Samples based on a topic of their choice and a limited writing period of 20 minutes were obtained from each subject and used as the first writing sample. The subjects next worked on a single story during eight 20-minute sessions in two weeks. They composed and expanded; very little revising or editing was observed. The ROBO-Writer group was observed to use the special features of this software (robotic voice, word lists) as part of their on-going editing; most of this editing behavior was at a surface level, however. A week after they completed the eighth session, they were asked to write a third story in 20 minutes. One month afterward, subjects completed two 30-minute sessions writing a story on "My Best Friend" using the same text-producing method they had used for the first three writings. The subjects used their regularly assigned classroom writing time to complete writing samples.

During the period of composing, the subjects in all three groups were conducting the same task but using different writing methods to complete their stories. Although the length of time was exactly the same for all the subjects, the period of the day to accomplish the writing task varied according to the subjects' regularly assigned classroom writing period. In most cases, an equal number of subjects from each of the three groups would be arranged to write at the same period of time. ROBO-Writer includes word lists (themes, first names) created by teachers. Copies of these lists were available on blackboards or posterboards in the writing areas for all students. The teachers gave similar instructional assistance and reinforcement to the subjects from all three groups, such as: "Can you read this sentence aloud?"; "Do you see anything wrong with it?"; or "Can you sound out the word yourself?" and so forth.

Measurements

The holistic ratings of written products were performed by graduate students enrolled in a graduate seminar course (Problems in Reading by Elementary School Children) devoted to holistic assessments. The session at which the assessments were accomplished was videotaped. The procedure used generally followed that described by Myers (1980). The starting point for the assessment was a state-wide rubric originally developed by Educational Service Unit 3 that made use of a 4-point scale for holistic ratings. Because this sample of 33 writers was identified on the basis of disabilities, the standards in the published rubric were thought to be too stringent; using those standards, few papers would receive a top rating. Eight written samples variously obtained as a part of this research but not among the 132 study samples were used as an anchoring set. Using the written standards as modified by the group of raters, the anchoring papers were rated. The ratings were then discussed by the raters in an attempt to clarify issues related to assigning scores for papers. All written samples in the study were converted to a single machine-printed format and identified by a randomly assigned 3-digit code. (Transcripts of paper-and-pencil writings were confirmed by two readers.) Two raters rated each written sample independently. The 132 samples obtained during this study were then rated. Texts of illustrative fourth writing samples and the corresponding ratings are displayed in Table 1.


TABLE 1

SAMPLES OF TEXT AND THEIR HOLISTIC RATINGS

(Transcriptions taken from the original writings on "My Best Friend")

Holistic Score = 1

My best friend is John.we like to play snow tag and we like to make snow fort .and we like to play supr nintidow. and we like to haf snowballfights. and we like to rasl. and we like to eat snow.
Holistic Score = 2

my best from play fot boll and basket ball and fishing to gather and Natndo to gather we rid are bikes i what to base ball we play sokr and tans we like sagas. and he like me to we play kick ball and we fly rmot plans we woc tv he 2 grad and he goes to a dafit scoll he is my next door navvr . we run rasis we play pirit and we have fun we roll blasing And we scat bard.
Holistic Score = 3

My best friend is Lisa. She go to school with me a lot. It was my birthday party. it was time to start. One friend came and then another friend came to my house. We were laughing. But my best friend did not come. I was sad. Time to go to bad said mom. I was bored I said that out loud.
Holistic Score = 4

My best friend is also named Jenny. We used to go to the same school and we were in hte same second grade room and the same math room. We have a lot of fun together. Sometimes her mom will let me come over to play with her at there house. Jenny has a little sister. Her name is Leah, She likes gymnastics. Love gymnastics too. Sometimes we will go to Camp Kotace. We go horse back riding there and fishing. There is a ski resort there.



Each sample was then assigned the sum of the two ratings as a score; scores ranged from 2 to 8. Of the 132 papers rated, 77 received the same rating by both raters. Another 38 papers received ratings differing by 1 point. In cases where ratings differed by 2 points, a third rater read the paper, and the sums of the closest scores were used. The interrater reliability was 0.814 (Pearson Product-moment Coefficient r = .814, p<.01).

The scores from the first writing were compared to the scores of the 2nd, 3rd, and 4th writings. Repeated-measures ANOVA was employed to analyze the holistic scores of the group mean differences.

RESULTS

Group means and standard deviations revealed no significant differences in average writing performance between treatment and comparison groups on the first writing assessment (see Table 2). There were significant differences in the average writing performances on the second, third, and fourth writing assessments. In each of these cases, the RW group received significantly higher ratings than either the MS or CT groups. The CT group mean writing performance on the second (3.91) and fourth (3.46) writing were much lower than those of the RW group second (6.91) and fourth (6.55) writing, for example. Differences between the MS and CT groups on these three writings were not significant. The size of the effect is very large. When effect size is calculated as the difference in means divided by the standard deviation of the control group, the smallest effect size is 2.0 (for the 2nd writing).


TABLE 2

GROUP MEAN AND STANDARD DEVIATION OF HOLISTIC SCORING

N=33

Items & Group

1st time


2nd time


3rd time


4th time

MS 2.73

[0.91]

4.27

[0.91]

3.36

[1.36]

4.02

[1.04]

RW 3.36

[1.50]

6.91

[1.30]

5.27

[1.35]

6.55

[1.37]

CT 3.27

[1.10]

3.91

[1.51]

3.18

[0.87]

3.46

[0.69]

MS = Microsoft Word
RW = ROBO-Writer
CT = Control

The statistical results from repeated-measures ANOVA detected significant group mean differences among the three groups, F(2,30) = 26.36, p<.001 (see Table 3.)

The follow-up tests revealed a significant group mean difference between the MS and RW groups, F(1,20) = 44.04, p<.001, and a significant group mean difference between the RW and CT groups, F(1,20) = 39.51, p<.001. However, there was no significant group mean difference between the MS and CT groups, F(1,20) = 0.23, p=.637. A group-by-time interaction was detected which indicated that there was a basic tendency of writing skill growth in all three groups. The RW group showed tremendous growth between the first and the second writing samples. The MS group showed more growth than did the CT group even though that difference was not significant. There was a general decrease in the writing skill for all three groups when third writing samples were compared to the second writing samples. The most important factor that contributed to the decrease was the time difference. The second writing samples were created in 160 minutes, whereas the third writing samples were produced in only 20 minutes.

TABLE 3

REPEATED-MEASURES ANALYSIS OF VARIANCE FOR

HOLISTIC SCORING

N=33


COMPARISON
DF
F

VALUE

P

VALUE

Overall

Group Effect

Time Effect

Group By Time


(2,30)

(3,90)

(3,90)


26.36***

20.77***

4.81**


0.000

0.000

0.002

MS & RW

Group Effect

Time Effect

Group By Time



(1,20)

(3,60)

(3,60)



44.04***

20.47***

3.14*



0.000

0.000

0.032


RW & CT

Group Effect

Time Effect

Group By Time



(1,20)

(3,60)

(3,60)



39.51***

14.71***

8.41***



0.000

0.000

0.000

MS & CT

Group Effect

Time Effect

Group By Time



(1,20)

(3,60)

(3,60)



0.23

7.10***

1.89



0.637

0.000

0.141

MS = Microsoft Word Group
RW = ROBO-Writer Group
CT = Control Group

* p<.05
** p<.01
*** p<.001


Time was an important constraint for all writers. The first and third writings were completed in 20 minutes. The second writing permitted 160 minutes during eight 20-minute sessions. The 4th writing was produced during two 30-minute sessions. The text-producing method showed great influence on the written products under all time options studied.

DISCUSSION

The statistical results from this study showed significant group mean differences.

Figure 1. Group Mean Comparison of Holistic Ratings

The results support the notion that computer technology can be a very important method to enhance the writing skills of students who are experiencing learning difficulties. This study points out that computer support alone does not lead to enhanced writing with this population of students. Instead, computer tools that (a) allow easy selection of words from word lists, (b) provide aural feedback during text production, (c) are driven by an iconic interface with aural help available, and (d) provide positive reinforcement through attractively printed written products lead to enhancements. In this study, the ROBO-Writer group used a tool designed specifically for this student population. Students using this tool always produced the best written products when compared with those in the other two groups. ROBO-Writer was developed with input and iterative feedback from students with learning disabilities and their classroom teachers, so its features reflect their perceived needs.

Our qualitative classroom observations show that word lists serve both as a resource for difficult words to help students with their spelling and as a catalyst for their ideas. Students are making connections between the words they see in the lists and the ideas related to the story they are creating. The content of the texts produced using ROBO-Writer were more theme-oriented than those created using either of the other technologies, a likely consequence of the built in features of that software. Students using ROBO-Writer were observed having more ongoing revision than those using the other writing tools. MacArthur & Shneiderman (1986) indicate that word processing features reduce the tedium of rewriting and recopying. The various word lists in ROBO-Writer facilitate further the creation of neat text. In agreement with Borgh & Dickson (1992), we found that students used aural feedback from ROBO-Writer to help to edit their sentences and extend their ideas.

We took pains to control levels of teacher support and enthusiasm. Reports from the classroom teacher (T. Frields) and school principal (M. Redelfs) indicated from earlier studies that parental support was higher from ROBO-Writer than from traditional pencil-and-paper writing. They attributed this difference to attractively produced writing samples brought home by the students on a regular basis. Qualitative observations to be reported elsewhere indicate that ROBO-Writer helped build self-esteem and got students more actively involved with the writing process than did the other technologies we studied.

Unlike Morton, Lindsay, and Roche (1989b) we did not find a positive relationship between story length and writing tool used.

The sample was small and not totally random. The subjects whose work was studied all were Caucasian, and most enjoyed middle socioeconomic status. All participating teachers were volunteers. Because the experiments were school based, it was not possible to ensure that no differences in verbal instruction or teacher reinforcement took place. Systematic longitudinal studies to determine the degree to which performance enhancements persisted have not been completed. These represent limitations to the generalizability and impact of this study.

CONCLUSION

ROBO-Writer, a HyperCard-based software product that runs on ordinary Macintosh computers, has been shown to enhance significantly the holistic quality of written products of students with learning disabilities when compared with products generated from pencil-and-paper (as well as one other software product). The importance of this conclusion is not so much in what it says about ROBO-Writer as in that appropriate performance-enhancing tools are well within the reach of typical resource rooms for students with learning disabilities sited in elementary schools.

Contributors

Yuehua Zhang completed her graduate work at the Center for Curriculum and Instruction, University of Nebraska-Lincoln. Formerly a teacher of English as a second language from Beijing Normal University, Dr. Zhang's interests focus upon instructional technology. David W. Brooks is a Professor of Chemistry Education at Nebraska, and has developed over 100 applications using HyperCard, most of which find use in science teaching. Teresa Frields is a resource room teacher at Northside Elementary School, and diagnoses learning disabilities for students in the Nebraska City Public Schools. Michael Redelfs is the principal of the Northside Elementary School in Nebraska City, Nebraska.

Acknowledgments

Support for the research studies on ROBO Writer was provided by the Center for Curriculum and Instruction, the University of Nebraska-Lincoln, and the Holmes Group. Dr. Charles D. Friesen participated in designing the experiment, selecting school sites, and providing computer hardware. Professor Richard J. Meyer assisted in developing the holistic rating scheme, training raters, and conducting the rating. Mr. Aimin Wang assisted in the statistical analyses.

References

Anderson-Inman, I. (1986). The reading-writing connection: Classroom applications for the computer. The Computer Teacher, 14, 23-26.

APDA (1984). Macintalk. In Cupterino, CA: Apple Programmers and Developers Association.

Atkinson, B. (1987). HyperCard. In Cupertino, CA: Claris.

Borgh, K., & Dickson, W. P. (1992). The effects on children's writing of adding speech synthesis to a word processor. Journal of Research on Computing in Education, 24(4), 533-544.

Bridwell, L. S., Nancarrow, P. R., & Ross, D. (1984). The writing process and the writing machine: Current research on word processors relevant to the teaching of composition. In R. Beach & L. S. Bridwell (Eds.), New directions in composition research (pp. 381-398). New York: Guilford Press.

Brooks, D. W., Zhang, Y., Frields, T., & Redelfs, M. (1993). Manuscript submitted for publication.

Clements, D. (1985). Computers in early and primary education. Englewood Cliffs, NJ: Prentice-hall.

Daiute, C. (1983a). Writing creativity and change in childhood education. Childhood Education, 59, 227-231.

Daiute, C. A. (1983b). The computer as stylus and audience. College Composition and Communication, 34(2), 134-145.

Daiute, C. A. (1985). Writing & computers. Addison-Wesley.

Dalton, D. W., & Hannafin, M. J. (1987). The effects of word processing on written composition. Journal of Educational Research, 80(6), 338-342.

Englert, C. S., & Mariage, T. V. (1991). Shared understandings: Structuring the writing experience through dialogue. Journal of Learning Disabilities, 24, 330-342.

Englert, C. S., & Thomas, C. C. (1987). Sensitivity to text structure in reading and writing: A comparison between learning disabled and non-learning disabled students. Learning Disability Quarterly, 10, 93-105.

Englert, C., Raphael, T., Fear, K., & Anderson, L. (1988). Students' metacognitive knowledge about how to write informational texts. Learning Disability Quarterly, 11, 18-46.

Fais, L., Wanderman, R., & Craig, K. (1988, April). Freedom of exception for learning disabled writers. Paper presented at the annual convention of the Council for Exceptional Children, Washington, D. C.

Graham, S., & Harris, K. R. (1988). Research and instruction in written language [Special issue]. Exceptional Children, 54.

Hennings, D. G. (1981). Input: Enter the word-processing computer. Language Arts, 58(1), 18-22.

Houck, C. K., & Billingsley, B. S. (1989). Written expression of students with and without learning disabilities: Differences across the grades. Journal of Learning Disabilities, 22, 561-568.

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

Kozma, R. B. (1991). Computer-based writing tools and cognitive needs of novice writers. Computers and Composition, 8(2), 31-45.

Kroll, B. M. (1981). Developmental relationships between speaking and writing. In B. M. Kroll & R. Vann (Eds.), Exploring speaking and writing relationships (pp. 32-54). Urbana: National Council of Teachers of English.

Kurth, R.J., & Kurth, L.M. (1987, April). A comparison of writing instruction using word processing, word processing with voice synthesis, and no word processing. Paper presented at the annual meeting of American Educational Research Association, Washington, DC.

Lehrer, R., Levin, B. B., Dehart, P., & Comeaux, M. (1987). Voice-feedback as a scaffold for writing: A comparative study. Journal of Educational Computing Research, 3(3), 335-353.

Lerner, J. (1989). Learning disabilities: Theories, diagnosis, and teaching strategies. Boston: Houghton Mifflin Company.

Lynch, E. M., & Jones, S. D. (1989). Process and product: A review of the research on LD children's writing skills. Learning Disability Quarterly, 12, 74-86.

MacArthur, C. A. (1988). The impact of computers on the writing process. Exceptional Children, 54, 536-542.

MacArthur, C. A., & Graham, S. (1987). Learning disabled students' composing under three methods of text production: Handwriting, word processing, and dictation. The Journal of Special Education, 21(3), 22-42.

MacArthur, C. A., & Shneiderman, S. (1986). Learning disabled students difficulties in learning to use a word processor: implications for instruction and software evaluation. Journal of Learning Disabilities, 19, 248-253.

Morton, L. L., Lindsay, P. H., & Roche, W. M. (1989a). Word processing effects on writing productivity and revision at elementary and junior high school levels. The Alberta Journal of Educational Research, 35(2), 145-163.

Morton, L. L., Lindsay, P.H. & Roche, W. M. (1989b). A report on learning disabled children's use of lab-based word processing versus pencil-and-paper for creative writing. The Alberta Journal of Educational Research, 35(4), 283-291.

Myers, M. (1980). A procedure for writing assessment and holistic scoring. Urbana, IL: ERIC Clearinghouse on Reading and Communication Skills, and the National Council of Teachers of English.

Newcomer, P. L., Barenbaum, E. M., & Nodine, B. F. (1988). Comparison of the story production of LD, normal-achieving, and low-achieving children under two modes of production. Learning Disability Quarterly, 11, 82-96.

Nodine, B. F. (1983). Forward: Process not product. Topics in Learning and Learning Disabilities, 3(3), 9-12.

Pea, R. D., & Kurland, M. (1984). Toward cognitive technologies for writing (Tech Report No. 30). New york, NY. Bank Street College of Education.

Purves, A. (1985). Composition instruction. In T. Husen & T. N. Postlethwaite (Eds.), The international encyclopedia of education: Research and studies (pp. 910-919). Oxford: Pergamon Press.

Rosegrant, T. J. (1984). Fostering progress in literacy development: Technology and social interaction. Seminars in Speech and Language Journal, 5(1), 47-57.

Rosegrant, T. J. (1988). Talking word processors for the early grades. In J. L. Hoot & S. B. Silvern (Eds.), Writing with computers in the early grades (pp. 143-159). New York: Teachers College Press.

Rust, K. (1986). Word processing: The missing key for writing. The Reading Teacher, 39, 611-612.

Salomon, G. (1988). AI in reverse: Computer tools that turn cognitive. Journal of Educational Computing Research, 4, 123-133.

Thomas, C. C., Englert, C. S., & Gregg, S. (1987). An analysis of errors and strategies in the expository writing of learning disabled students. Remedial and Special Education, 8(1), 21-30.