COMPUTERS and COMPOSITION 10(3), August 1993, pages 93-101
Electronic Networks for Interaction (ENFI), in one form or another, and by one name or another, has been a factor in college composition since 1985.  During one year, academic year 1989-1990, the ENFI Consortium  conducted a three-pronged evaluation of ENFI. This article is an assessment of that evaluation. Much more can be said about ENFI, from other perspectives and involving other kinds of data, but in this article I am limiting myself to just two findings from the 1989-1990 summative evaluation. These two findings suggest at least two ENFI research hypotheses to pursue:
ENFI, as most readers of Computers and Composition know, involves use of a local-area network to teach writing through the use of software that supports real-time communication. In many cases, and in the cases we evaluated, teachers shift nearly all communication during class to interaction on the network.
We employed three approaches rather than one in our summative evaluation because, as the pioneer project studying the ENFI environment for teaching writing in college, we had little idea what writing effects we were looking for. We hoped that by employing as many views as possible on implementations of this new idea, we might uncover effects that were important. We didn't want to limit our discoveries with preconceptions. We were also as interested in the institutional effect (effects on student retention, faculty research and motivation, and allocation of resources) of the move to ENFI as we were in the writing effect.
Therefore, we employed 1) a "situated evaluation" (Bruce & Rubin, 1992) conducted by Bertram Bruce of the University of Illinois and Joy Kreeft Peyton of the Center for Applied Linguistics, 2) a standard Educational Testing Service (ETS) writing sample analysis, and 3) a broad and careful reading of paired student papers by David Bartholomae.
The Bruce and Peyton study focused on how the original ENFI idea was transformed at each of the sites as it was transported there. This study is important to our consideration of what effect ENFI has on student writing because its point is that ENFI became different things on different campuses as the idea became acculturated into different campus cultures. It may be that our generalities about ENFI are suspect, then, because ENFI turned out to be a different beast on each campus. What we found in our study can only be suggestive of what one might find on a particular campus when ENFI is implemented. To some extent, describing the effect of ENFI is like describing the effect of "the classroom" because ENFI is not a method but an environment. The Bruce and Peyton work is written up separately (Bruce & Peyton, 1990) and is also expanded upon in a book (Bruce, Peyton, & Batson, in press.)
Both the ETS and Bartholomae studies, unlike the Bruce and Peyton study, sought a universal ENFI effect and also took a more traditional writing evaluation approach. They both looked exclusively at student essays written in the traditional autonomous (nonnetwork) manner: Even the ENFI students wrote their papers at home, alone, either by hand or on the computer.  We chose to engage the ETS because they've been instrumental in writing evaluation for many years and their approach is valued. We chose David Bartholomae as our "qualitative" reader in an effort to discover effects of ENFI. He was chosen not because of his broader reputation within the discipline but because of his acknowledged sensitivity to features of student writing. Also, he was not an ENFI user, nor even an enthusiast, so we thought he would be unlikely to look subconsciously for "good" effects. 
In most cases, the ENFI students involved in each of these three studies were taking first-year composition courses. Their classes would meet in a computer lab with a local-area network using Daedalus INTERCHANGE from The Daedalus Group, REALTIME WRITER from Realtime Learning Systems, the CB UTILITY on the 10Net network, or CECE TALK from Carnegie Mellon (all software that supports ENFI interaction in a classroom). The teacher and students would work on various collaborative writing activities, or they would discuss, in writing, ideas for papers. Most communication was in writing.
At each site, students from these ENFI classes and from traditional
composition classes that were held in noncomputerized classrooms
were selected for the evaluation studies. For the Bartholomae
study, regularly assigned papers were selected for analysis; for
the ETS study, essays were generated in a timed situation.
David Bartholomae was given regular drafts and essay papers from three Consortium sites--New York Institute of Technology, University of Minnesota, and Northern Virginia Community College. He read papers written by basic writers: 830 papers written by 62 ENFI and 81 non-ENFI students. The two groups of students were in different sections of the same course. Bartholomae was not trying to prove one set of papers was better than the other but was merely trying to decide whether the two sets of papers differed in their features.
I'll quote various parts of Bartholomae's write-up of his study:
I found the reading more interesting than I thought I would, and I found the differences in the ENFI sample more strongly marked than I thought I would. I feel confident in saying that the ENFI sample was different; and it seems reasonable to conclude that the differences can be linked to writing on a network.
Remember that the teacher and students spend most class periods carrying on a written group conversation over the network. The teacher cannot actively direct the conversation; all participants can type at the same time. Typically, most students will get to "speak up" more often in an ENFI classroom than in a traditional classroom because everyone can "talk" at the same time. This environment "both created and supported certain specific ways of imagining and valuing writing that made a difference when students wrote," according to Bartholomae.
The students seem to have drawn on their oral abilities in writing on the network more than in normal essay writing because, perhaps, they see network writing as more risk free, like conversation. It is also conversational in that participants write in response to each other's comments.
The most notable differences I saw between the ENFI and non-ENFI essays could all be related to the ways in which work on the network invites students to imagine that writing is a form of speech. The ENFI essays were less formal, more colloquial, less predictable, more individualized, less likely to present themselves as texts and more likely to imagine a direct form of address. . . .The ENFI writers could be said to be thinking out loud, trying for effect, teasing or manipulating an auditor close to the writer and her situation, arguing with specific and unspecific interlocutors.
Bartholomae refers to two papers, one from a non-ENFI student
and one from an ENFI student, to illustrate the differences to
which he refers.
First, the non-ENFI sample: 
Child abuse is defined by the children have been deliberately inflicted with serious physical injury such as broken bones, burns, or internal injures by their parents or caretakers. Child abuse has become one of our major social problems in the United State. It occurs in the presence of three factors. . .
And then, the opening sentences of the ENFI sample:
I know this paper is going to be very negative towards the women studies course, but don't get me wrong. I feel the course does enlighten women and their views on the world today. First I would like to get my opinion of the Daily out of the way. This is because of the Daily's premature printing of Michael Olinecks articel without a response to it. The paper and it's one sided view on alot of articles it writes is extremely bad, and the University should do something about it.
Bartholomae notes that in the first sample, the student is producing the language of the term paper, what she thinks is academic prose. He doesn't feel she is addressing us, her readers, but is mechanically following steps she's learned about how to write an essay. Her next paragraph begins: "There are five needs parents must be able to meet for their children to develop and to be healthy."
The second writer works more conversationally, Bartholomae says. His writing has its own problems, of course: The assumption that we know about the Daily is one of those problems. His next paragraph begins:
The women's studies course 1001 is just what the course is titled, a study of women in society, and for Michael Olineck to take that class was a big mistake. I realy don't feel he had a positive attitude from day one, giving a man's point of view in a course where it was not wanted.
Bartholomae says, referring to the difference not only in these two papers but in the two sample sets:
What distinguishes the ENFI essays . . . are not only ways in which they assume a listener rather than a reader, but also the form and degree of engagement with the material and the number and kinds of attempts to engage an audience.
He found differences, not only in these two essays, but throughout his samples in the following:
Voice/style, the address to the reader, the presence of a persona, the presence of linguistic features associated with speech, [and] the dialogic engagement with a subject. . .I found more frequent evidence of writing to manipulate the audience response--to create expectation, to provide surprise, to joke or shock, to counter what was assumed to be a reader's prejudice or assumptions.
The non-ENFI essays were flat in comparison--that is, they did not dramatize to the same extent an encounter between a reader and a writer or a speaker and a listener.
[The ENFI writers imagined] that their texts existed in the context of prior discussion and prior beliefs, anticipating the assumptions and prejudices of an audience. I saw this to a greater degree than I have learned to expect generally in student writing.
The ETS Study
Mary Fowles directed the ETS evaluation of the ENFI Project. Instead of looking at essays written for class, as Bartholomae had, Fowles had students write prompted, timed essays in a test situation, by hand. Bartholomae read papers from 143 students at three sites; the ETS readers judged papers from 458 students at five sites. Some of the 143 Bartholomae students were included in the larger ETS sample.
The ETS approach is familiar: Nearly the same number of ENFI and non-ENFI students, matched classes, blind reading (no names and no indication if papers were pre- or postessays), use of trained readers, pre- and postessays intermingled, each essay scored by two different readers, second reader doesn't know first score, and the interrater reliability checked (it was 3.45%, which Fowles called "respectable").
The 458 students were typical of traditional first-year composition
students, except for the inclusion of 83 deaf students (18% of
the total sample--Bartholomae did not read papers written by deaf
Fowles reports the results:
Both the ENFI and the non-ENFI groups improved slightly their ability to write an essay under the conditions set forth in this study. Since gains were nearly the same for both groups, the use of the ENFI network did not appear to contribute significantly to student writing ability, at least not as writing ability was measured in this study. At the same time, considering that ENFI is still in its early stages, it is reassuring that ENFI classes performed as well as students receiving more traditional instruction.
So much for a silver bullet. However, still hoping to find something of note in the sample, Fowles looked only at those papers where significant improvement occurred; again, nothing: "No single feature accounted for the improved quality of writing." She was not able to find the more conversational style that Bartholomae had detected, either, so she set out on another read of the papers looking specifically for that. She still found nothing.
The reason seems obvious [she said]. The writing prompts used in [the ETS] study asked students to draw from personal experience. Personal writing tends to be informal, and the writing we evaluated was indeed informal, though not especially conversational. The classroom pieces that Dr. Bartholomae reviewed were written for very different, more academic purposes. Neither the 'academic' writing he found in non-ENFI classes nor the more conversational styles he observed in the ENFI classes appeared to transfer to the essay writing situation established for this study.
Finally, in recognition of the complexity of pinning down the effect of ENFI on student writing, Fowles remarked, "If students were writing one way in their classrooms and another in the Project assessment, how can we demonstrate the effects of ENFI?"
Exactly. Fowles concluded:
The fact . . . that the ENFI students had to transfer their writing skills to a non-ENFI situation for the test might have put them at a disadvantage. That they 'held their own' against students in non-ENFI classes was a positive result.
What do we make of these studies? Bartholomae sees what he thinks is an ENFI effect, the conversational style; students apparently conceived of their rhetorical context as students rarely do, in Bartholomae's experience. Fowles doesn't see it. But she does see that the ENFI students "held their own" in terms of producing standard academic essay prose.
The teachers of the (ENFI) students who employed this conversational style did not directly encourage this style, or any style. Yet, for some reason, these particular students tended to adopt what seems like an unusual style in an academic setting when writing assigned papers. These students and other students who had been in an ENFI class didn't carry this style over to the timed ETS essay.
A speculation offered by one of the teachers involved is that because the students had already written on the network about the same ideas, they had become accustomed to writing about them conversationally, and that had carried over to the assigned papers about those ideas. It was a continuation of the written conversation. In addition, because students talk on the network more to each other than to the teacher, Bartholomae thought they may feel the class is more cohesive, more a genuine social group, within which a natural conversational style would be appropriate. In other words, even when students wrote a paper outside of this particular class, they continued with the rhetorical context and habits of discourse of the ENFI class. When these same students were placed in a timed essay-writing situation for evaluation where the prompt had nothing to do with a class conversation and the social context was not the same, it is only reasonable they would revert to traditional essay style.
If what Bartholomae noticed resulted from the creation of a social context that persisted after class and influenced how the students conceived of their audience, then perhaps the most significant factor at play in the ENFI setting is the ability to write to other students, to develop ideas collaboratively. This context, rich in peer interaction, perhaps created the sense that the students, when they write their assigned papers later at home, are writing to a different audience than just the teacher.
Now, we have no way to know whether the students who wrote these conversational essays knew their writing was different in any way from what they normally write. Nor, can we know whether they employed that style in any other situation then or since. In addition, it's not even clear that this style is better or worse than traditional academic writing.
What is interesting is whether the conversational style that David Bartholomae saw--or any style of writing--is more likely to be transferred from an ENFI setting to individual composing than it is from a traditional oral-based classroom. Does the fact that the students are writing about the ideas in the class ENFI discussion mean that the style adopted by participants in class is more likely to carry over? Or can an oral discussion have the same influence? Even though some of the papers that Bartholomae compared were from the same teacher, we need to conduct studies that involve more matched pairings so we can look more systematically for this one effect. It makes sense that a written discussion would more directly affect student writing style than would an oral discussion, but we need to test this hypothesis. Two questions a researcher might ask: Would the style encouraged by a teacher have the same effect in either of the two conditions? And: Is it the influence of having more direct interaction with peers in the ENFI condition that produces the conversational style, regardless of the teacher?
The Fowles study suggests that the ENFI influence on style does not carry over to other academic settings: Some of the same students whom Bartholomae found writing conversationally in their assigned essays did not show this effect while writing the prompted, timed essay. Whether they would do so if they were not directed to write "personally" should be tested.
Does the ENFI setting allow teachers and peers to have more of a direct effect on how students write their individual essays? This is a key question. Because peers have such a large role in the online ENFI discussion, the teacher effect may not be so overwhelming as in a traditional classroom. In this case, then, we wouldn't be making the "technocentric" error of claiming that technology had effect because our claims would be based not on the network itself but on increased peer interaction; the teacher effect would be in allowing and encouraging the increased interaction.
It would be hopeful to think that ENFI students had developed the ability to write conversationally, as they seemed to have done in a number of the ENFI classes Bartholomae sampled, while retaining their ability to produce standard academic prose. However, we don't know that. These are impressionistic findings, suggestions for further study only. The ETS study is more to reassure ourselves that ENFI is not having an effect that would disadvantage the students in their academic career.
Our work shows only the hint of an ENFI effect over 10 or 15 weeks
in student writing. But that hint is worthy of a lot more study.
If we can show that the rhetorical moves and manner used during
ENFI sessions has a more direct effect on student writing than
a similar rhetorical set employed orally, then we need to know
that in order to consider how best to use this new teaching environment.
The creation of a writing persona, based on the sense of one's
audience, is a complex cognitive and imaginative task. If in
fact the ENFI setting allows a writing teacher to more easily
model this task and students to experiment with it, then this
is a valuable ability of ENFI. Our Consortium evaluation points
the way to further research, I believe. The research should focus
on whether ENFI allows students to develop a broader array of
writing styles and whether those styles are carried over to contexts
outside of the writing class.
Trent Batson, pioneer of the ENFI system, teaches
English at Gallaudet University in Washington, D.C.
The Consortium, funded in part by the Annenberg/CPB Project, included Gallaudet University, Carnegie Mellon University, the University of Minnesota, New York Institute of Technology and Northern Virginia Community College; in the Fowles study mentioned in this article, the National Technical Institute for the Deaf was included as well.
One finding of the ETS study was that ENFI students are significantly more likely to use computers to compose on than non-ENFI students. See Fowles, M. (1990). The Effects of ENFI (Electronic Networks for Interaction) on Students' General Writing Ability. Unpublished report, p. 19. ENFI Project, Gallaudet University, Washington, DC.
The Batholomae report ("I'm talking about Alan Bloom": Writing on the Network) is available from the ENFI Project and will appear as a chapter in Bruce, B., Peyton, J., & Batson, T. (in press).
Student writing is reproduced here intact and unedited.
Bruce, B., & Peyton, J. K. (1990). A new writing environment and an old culture: A situated evaluation of computer networking to teach writing. Interactive Learning Environments, 1(2), 171-191.
Bruce, B., Peyton, J., & Batson, T. (Eds.), (in press). Network-based classrooms: Promises and realities. New York: Cambridge University Press.
Bruce, B., & Rubin A. (1992). Electronic Quills: A situated
evaluation of using computers for teaching writing in classrooms.
Hillsdale, NH: Lawrence Erlbaum Associates.