This article appeared in The Writing Notebook 8(1), 42-46.

One need only glance at an issue of The Writing Notebook to see that there is tremendous interest in and enthusiasm for the use of word processors in the teaching of writing. In our area, this trend has been endorsed by the provincial department of education which has made word processing a mandatory part of the junior high language arts curriculum and has prepared an inservice leaders’ handbook to provide support for professional development activities (Alberta Education, 1988). Under these circumstances, it is highly appropriate to ask whether such changes in curriculum and instructional methods are supported by research findings. In attempting to answer this question, we and our colleagues (Hunter, Begoray, Benedict, Bilan, Jardine, Rilstone, & Weisgerber, 1988) found ourselves overwhelmed by the amount of information available and by the subsidiary questions that arose as we started our investigation. We found it necessary first of all to develop a working model of the writing process (Hunter & Begoray, 1990) and to address what many teachers and administrators see as a prior question– the role of keyboarding (Hunter, Benedict & Bilan, 1989). Still, when we turned our attention to our original question "Does the use of word processors improve children’s writing?" we found that we must still address other questions, specifically, "What role do students’ attitudes play in determining the impact of word processing?" and "How does one measure improvements in writing?" In this article, we will attempt to provide an answer to the original question, but we must first respond briefly to the other two issues.

Systematic studies of people’s attitudes toward the use of word processors are still rather rare (e.g., see Johnson, 1987), but teaching magazines and computing magazines are full of testimonials about the perceived benefits for individual writers and for groups of students. We find these testimonies persuasive and consistent with our own experiences, but we also realize that there are many people who are unwilling to try to use a computer for any purpose or who resist learning to use a keyboard. Few of these people write articles about their feelings.

Still, some studies exist. Kurth (1987) and Coombs (1985) reported increases in self concept associated with the use of word processors. A number of writers report increases in motivation and enthusiasm for writing (Johnson, 1987; Hawisher, 1986; Coombs, 1985; Feldman, 1984). Others have indicated that students using word processors desire to write more often (Daiute, 1983; Newman, 1984) or that they show more willingness to make changes in their writing (Kurth, 1987; Schwartz, 1982). Decreases in writing anxiety have been associated with the use of word processors (Branan, 1984; Teichman and Poris, 1985; Feldman, 1984).

In contrast to this growing evidence of support for students’ positive attitudes toward word processing, Fitch (in Hooper, 1987) reported no significant change in attitude when students wrote with word processors. On the whole, it seems reasonable for us to believe that we can proceed to use word processors in the teaching of writing, confident that many students will enjoy the experience. Future research will undoubtedly help us to know who the exceptions are and how best to deal with their concerns.

There seems to be a growing consensus among language arts educators that "holistic" methods of evaluating writing are best suited to measurement in the classroom. These methods involve global judgements by trained readers who have been provided with some focus for the evaluation, usually related to the writer’s purpose. The Educational Testing Service, owners of a near monopoly on multiple choice testing, have elevated holistic scoring to a high art in their 30 years of experience with marking essays on the English Composition portion of the Scholastic Aptitude Tests (ETS, Fall/Winter, 1989/90). Moore and Kennedy (1971) have pointed out two disadvantages of using standardized tests to assess writing ability: they fail to reflect changes in English curricula after 1960 and they are built primarily around ease-of-scoring. Consequently, each item tests one and only one skill or concept. Moore and Kennedy argue that "...such fragmentation is highly artificial. In writing, ...the pupil must put everything together... (and) cannot afford the luxury of being concerned solely with isolated skills." (p.441) This concern is also reflected in White’s (1985) claim that "inappropriate testing has been destructive to the teaching of writing for the last two generations." (p.9) Although we recognize the virtues of holistic scoring for pedagogic purposes and for large scale assessment, we wondered whether a more atomistic approach might be better suited to research on the effects of different kinds of writing instruction.

Standardized tests and holistic marking of essays are not the only alternatives available to us. In Britain, "analytical" marking has long been used on national exams–evaluation of particular writing skills is done in the context of actual writing samples. However, Wilkinson, Barnsley, Hanna and Swan (1980) believe this procedure overemphasizes vocabulary, grammar and punctuation, proper sentence structure, spelling and handwriting and fails to consider such important aspects of writing as purpose and sense of audience. They also cite previous research to make the case that analytical marking and holistic or impression marking do not differ in reliability. Wilkinson, et al. also determined that using four independent holistic markers required no more time or effort than is required for one analytical marking.

Developers of multiple-choice tests are now focusing on indirect tests of isolated abilities (e.g., Benton and Kiewra 1986). They argue that holistic assessment is not to be trusted, and that students should take a variety of separate standardized tests each designed to tap "discourse production strategies at various levels" indirectly. For example, they suggest that skill at solving 5-letter anagrams corresponds to "the propositional strategies that are employed in word generation and clause formation." (p.384) They validate their series of tests of information-processing abilities by correlating each test score with the subject’s holistic score on two narrative paragraphs, each written in 10 minutes. Vast amounts of time, effort and financial support would be required to develop standardized tests of every ability involved in writing well. It is not clear whether such an investment is warranted. In the end, such tests might leave writing assessment in the same place that Kendall, Mason and Hunter, (1980) found the measurement of reading comprehension: after gathering data with four different item types on four different "story types", they wondered " ...which, if any, of the operational definitions used here is actually reading comprehension." (p. 235).

Future developments aside, it appears that the reliability of holistic scoring methods and their obvious face validity justify their being the preferred evaluation procedure for research on the writing process. However, it seems likely that future researchers will be interested in identifying more precisely what aspects of student writing are responding to differences in teaching methods (e.g., does regular use of grammar checkers result in greater use of the active voice?) and that some type of atomistic measurement will be developed to address these questions.

To a certain extent, these developments have already begun to take place in the form of procedures for evaluating students’ efforts at revising their writing. Bridwell (1981) reviewed the efforts in this area and proposed a seven-level classification of revisions–surface, lexical, phrase, clause, sentence, multi-sentence, and text. In a similar vein, Faigley and Witte (1984) recommend assessing instances of revision, such as additions and deletions, and determining whether they affect the structure and meaning of a text at a micro or macro level, that is, are they "surface changes" or "meaning changes?" They emphasize the need for consideration of the situational variables of a writing task: the author’s reason for composing, the format of the composition, the writer’s knowledge of the subject area and his audience, the degree of formality required for the writing, and the length and complexity of the task. The same writer may demonstrate different revision behaviors and skills in different writing situations.

Strong opinions about the effects of word processing are easy to find. For example, Herrmann (1987) reported that learning to handle the keyboard interferes with the writing process. Yet Branan (1984), Willer (1984) and Ho Chung Qui (1986) cite increased amounts of creative writing by learning disabled children when they were taught word processing.

Studies of composition length.

While composition length is not meant to be a measure of the quality of writing, investigators have examined this variable in the expectation that students who write more will eventually learn to write better. Kurth (1987) found no significant difference between control and treatment groups in length of compositions produced. Her subjects were high school students who were participating in an experimental writing course as a voluntary extra-curricular activity. Some of these highly-motivated subjects took their writing home to work on. It seems likely that these subjects would continue to rethink and revise their work until they were satisfied– regardless of the amount of extra time it would take to write by hand. With writing time controlled, however, Etchison (1985) found that students using word processors made greater increases in length of composition from pre to post instruction. His students had two days to plan their compositions, one hour to write, and then two days later were given a one hour revising session. Willinsky (1990) has found that education students whose normal course assignments were prepared on word processors wrote approximately 20% longer papers than those using typewriters or pen and paper.

Studies of holistic quality.

Hawisher (1986) and Cross and Curey (1984) found that overall quality ratings for essays produced with word processing were similar to those for essays produced with pen or typewriter. Willinsky (1990), using grades obtained on assignments for a variety of education courses, found no significant difference in marks earned by students who word processed their assignments, those who typed them, or those who wrote in long hand. Etchison (1985), on the other hand, found gains in holistic quality of essays written with a word processor were five times greater than were those written with pen and paper. It is important to note that Etchison took pains to control many other factors that might affect the quality of the final product (e.g., time allotted to the writing task, writing environment, etc.) It is quite likely that with his timed writing sessions, Etchison has provided an empirical indication of the relative speed with which students can compose and revise on a word processor after a semester of use.

Analytic studies.

Etchison (1985) also measured changes in syntactic variables such as free modification, number of clauses, average number of words per clause, and type of embedding, but found no significant differences in these areas between students who used word processing and those who did not.

Other Variables.

Suggestions of other factors which can influence the quality of writing when using a word processor include context of the assignment (Herrmann, 1987), conferencing (Etchison, 1985), time and opportunity to develop new skills in areas where concepts are new (Coombs, 1985), and classroom demonstrations of the technology as a means to an end (Cross and Curey, 1984). Cross and Curey also stated that the overall effects of word processing on writing vary with the writer and with general factors related to the teacher and the class. Finally, there is some evidence that using word processors may lead to an improved sense of audience (Wheeler, 1985; Macarthur, 1988).

The research we have encountered, though limited, indicates that the quality of documents prepared with word processors is at least as high as that obtained with other writing tools and that other variables are more important determinants of writing quality. Macarthur (1988) cautions that improvement should be credited to instruction and skill development rather than to the writing tool. Other authors (Hooper, 1987; Hult, 1986; Opack, 1986; and Kurth, 1987) suggest that writing instruction must be given concurrently with the teaching of word processing skills if better quality writing is expected to result.

Several studies have examined differences in the ways writers approach revision when using a word processor as compared to other writing tools. Some have focused on the number of revisions made (e.g., Hawisher, 1986;

Cross and Curey, 1984), while others have focused on the kinds of revisions made (Willinsky, 1990; Kurth, 1987; Hawisher, 1986). Macarthur (1988) and Hult (1986) compared word processing revisions between experienced and inexperienced writers.

In a critical review of the literature, Hooper (1987) concluded that word processors ease the revision process (Nash, 1985), that most revisions are purely cosmetic (Womble, 1984) and that fewer substantial revisions are made by students using word processors (Harris, 1985). Students in Willinsky’s (1989) study also indicated that their most frequent revising activities were minor changes–spelling and grammar corrections or the addition, deletion, replacement, or reordering of words. In this study, students using word processors reported more revisions in all these categories than did students using typewriters or pens. They also reported more additions of sentences and of paragraphs–more significant changes than those found in most earlier studies.

Hawisher (1986) examined the effects of word processing on the revision strategies of college students and observed that amount of revision was not positively correlated with quality ratings and that students make the same kinds of revisions (i.e., "surface" vs. "meaning" changes) regardless of the writing tool used. These observations were based on beween-draft revisions and may not apply to revisions made in the process of composing. In a later study, Hawisher (1987) found no qualitative difference in the revisions made by those who used a computer compared to those who wrote with pen. She suggested that the initial draft, "the point of utterance" has a greater effect on the ultimate quality of the product than do the revisions students make. However, it may well be that this is simply a reflection of the minimal amount and quality of revision which other researchers (see above) have found to be the norm.

Hult (1986) found that inexperienced writers focus their revision efforts at the level of word changes, resulting in surface revisions. Experienced writers tended to see composition as a complete activity; therefore, they make more changes to style and content. Macarthur (1988) found that experienced writers make more revisions than inexperienced writers when both groups are using word processors. He also cautioned that easing the physical requirements of revision will not result in better compositions unless students know how to evaluate and correct their writing.

In a study of professional writers, Lutz (1987) observed that they made different kinds of revision when they used word processors than they made using pen and paper. The computer’s small window on a document seemed to encourages local editing but may have limited the writer’s perspective on the document as a whole, thereby discouraging deep revision of the text. Revision was more effective when done with pencil and paper or when done on hard copy.

Of course, it is unreasonable for us to expect that changing writing tools alone will change the kinds of revisions writers make. As Hult (1986) has argued, the development of effective revision strategies will require that we teach the value of such functions as moving and deleting blocks of text. It should also be clear that students will need instruction and support in knowing what to revise–they will need to learn how to evaluate and improve their own writing in order to know what use to make of the word processor.

One way of increasing the utility of computers and word processors for the teaching of writing may be to use software that will guide the processes of generating, organizing, composing and revising text. For example, to reduce the cognitive demands that writing makes on children, Woodruff, Bereiter and Scardamalia (1981-82) gave elementary school children word processors that could provide on-line support to assist with composing via a hierarchy of writing aid menus. Although the students frequently summoned this assistance and believed that it improved their writing, the researchers found that it made no significant difference in either the quantity or the quality of the students’ compositions. Daiute (1986) also reported that the availability of on-screen writing aids did not help eighth grade students to improve the quality of their writing.

Although Kurth (1987) also reported that use of a computer did not improve the quality or quantity of junior high school students’ writing, the students she worked with had fewer spelling errors when they used a word processor with a spell checker than when they had when they used pencil and paper. The high visibility of text on a screen also seemed to foster more conversation about writing. Kurth concluded that the quality of students’ writing is affected more by good instruction than by the writing tool.

If we take this research at face value, we can conclude that many students enjoy using word processors and that there is little strong evidence to support the current enthusiasm for using word processors in the teaching of writing. But should we take this research at face value? Our answer would have to be a clear and unambiguous "no." Much of the research was based on samples of university students, nearly all of it relied on comparisons of students who chose to use word processors (as opposed to using random assignment), few of the studies made any effort either to teach the students the use of word processors or to ascertain their levels of word processing expertise and none of the studies made any reference to systematic instruction in the writing process or in revision strategies. It is as if we were to study the impact of calculators by studying the ways they are used by students who do not know how to multiply or divide or what a square root is. We are only beginning to learn how to structure research on this question and we should avoid over-reacting to the early results.

Etchison’s (1985) study provides some directions for future research in that it involved serious efforts to control some of the more obvious extraneous variables. The finding that using word processors does make marked differences in the quality of compositions when rigid time limits are imposed may help to explain why so many other studies found little or no difference. The usual pattern, both in research and in classroom practice, is to allow students generous time limits for writing assignments (e.g., to give several weeks notice of due dates). We suspect that under these circumstances students compose and revise until they achieve some personal quality standard and then stop. It does not matter what writing tool they are working with, they will continue to work until they satisfy their personal "level of acceptable performance" or LAP point, then quit. Researchers will not see a difference in quality, but it entirely possible that students using word processors achieved their LAP point much more quickly and in a less onerous way than those using other writing tools. Students would certainly find this to be reason enough to use word processors, but teachers would likely be disappointed that this interpretation suggests that no improvements in writing quality can be expected no matter how refined technological aids to writing may become.

However, we need not accept that conclusion; we need only to turn our attention to the question "How do we get students to elevate their LAP point? How do we get them to expect more of themselves?" We think the answers lie in more active instruction, for example, by teaching the elements of the writing process in conjunction with teaching the capabilities of word processors , by helping students to understand the demands of style and audience, by teaching strategies for evaluating compositions and making revisions, and by encouraging students to seek and use feedback on drafts. In short, we ought not expect word processors to relieve us of the responsibility of teaching students how to write; rather, we should be seeking ways in which the availability of word processors can help students to set higher personal and help us to assure that they develop the competencies necessary to achieve them.

Alberta Education. (1988). The writing process using the word processor:

Benton, S.L. and Kiewra, K.A. (1986). Measuring the organizational

Branan, K. (1984). Moving the writing process along. Learning. 13, 221.

Bridwell, L. (1981). Revising strategies in twelfth grade students’

Coombs, M. (1985, September). Exploring the Composing Process with Microcomputers.Annual Meeting of the National Reading and Language Arts Educators’ Conference: Kansas City, MO. (ED 267424).

Cross, J. and Curey, B. (1984). The Effect of Word Processing on Writing.

Daiute, C. (1983). The computer as stylus and audience. College

Daiute, C. (1986). Physical and cognitive factors in revision: Insights

Etchison, C. (1985). A comparative study of the quality of syntax of

ETS. ETS Developments (Fall/Winter, 1989-1990). Grading essays for ETS:

years experience in holisitic scoring. Princeton, NJ: Educational Testing Service, 7-9.

Faigley, L. & Witte, S. (1984) Measuring the effects of revision on text

Feldman, P. (1984, February). Using microcomputers for college writing:

Harris, J. (1985). Student writers and word processing: a preliminary

Hawisher, G. (1986, April). The effects of word processing on the

Hawisher, Gail E. (1987). The effects of word processing on the revision

Herrmann, A. (1987, December). Research into writing and computers:

Ho Chung Qui, J. (1986). Word Processing and Learning Disabled Children.

Hooper, S. (1987). Using Word Processing in High School and College

Hult, C. (1986, March). The computer and the inexperienced writer. Annual

Hunter, W. & Begoray, J. (1990). A framework for writing process

Hunter, W., Begoray, J., Benedict, G., Bilan, B., Jardine, G.,

Hunter, W., Benedict, G., & Bilan, B. (1989). On a need-to-know basis:

Johnson, V.M. (1987). Attitudes toward microcomputers in learning.

Kendall, J., Mason, J. & Hunter, W. (1980). Which comprehension? Artifacts

Kurth, R. (1987). Using word processing to enhance revision strategies

Lutz, J.A. (1987). A study of professional and experienced writers

Macarthur, C. (1988). The impact of computers on the writing process.

Moore, W.J. and Kennedy, L.D. (1971). Evaluation of learning in the language

Nash, J. (1985) Making computers work in writing class. Educational

Newman, J. (1984). Online: Reading, writing and computers. Language

Opack, M.M. (1986). Effective instruction in word processing: Maximizing

Schwartz, M. (1982) Computers and the teaching of writing. Educational

Teichman, M., & Poris, M. (1985). Wordprocessing in the classroom: Its

Wheeler, F. (1985). Can word processing help the writing process? .

White, E.M. (1985). Teaching and assessing writing. San Francisco:

Wilkinson, A., Barnsley, G., Hanna, P. and Swan, M. (1980). Assessing

Willer, A. (1984). Creative writing with computers: What do elementary

Willinsky, J. (1990). When university students word process their

Womble, G.G. (1984). Do word processors work in the English classroom?

Woodruff, E., Bereiter, C., & Scardamalia, M. (1981-82). On the road to