Our growing understanding of the ways in which computer networks can support cooperative work in writing classrooms (Batson, 1988; Hartman et al., 1991; Mabrito, 1991; Neuwirth et al., 1993) has increasingly drawn our attention to the interrelationships between classroom context and the uses of computer networks. [1] As we have begun to examine those relationships, our initial conception that access to computer networks extends the classroom has given way to the conclusion that, rather than simply extending, such access transforms the classroom. In some settings (Selfe, l991), this transformation involves the creation of a venue in which students dictate discussion and, to some extent, exclude the teacher from participation. In other settings (Batson, 1988), the teacher plays a more active role in network discussions. Although much work remains to be done in this area, the growing body of scholarship dealing with writing instruction and network-based communication programs suggests that these programs--among them electronic mail, electronic bulletin boards, ENFI, [2] and commenting software--can significantly shape the social and instructional contexts of the computer-supported writing classroom.

As more of us begin teaching composition courses in network-supported classrooms, we must ask ourselves whether the transformations in curricular and classroom context brought about by network communication technologies are appropriate. Among other questions, we must ask whether the changes brought about by network-based communication help students and teachers meet specific curricular goals--or whether, perhaps, they work against those goals. And we must ask whether the additional effort and expense entailed in using computer networks in the writing classroom are balanced by their instructional benefits. To date, research in this area has been highly exploratory; much work remains before we can develop a clear model of the relationships between composition instruction and network-based communication. In some ways, our current understanding of-- and enthusiasm for--network-based communication is similar to the initial enthusiasm about the potential benefits of computer-based grammar and style analysis programs (Cherry, 1982; Gingrich, 1983; MacDonald, Frase, Gingrich, & Keenan, 1982). That initial enthusiasm was subsequently tempered by the work that found such programs beneficial, at best, only under specific curricular conditions (Collins, 1989; Dobrin, 1986; Kiefer, Reid, & Smith, 1989; Kinkead, 1986).

That caveat aside, the case for the benefits of network-based instruction has much stronger face validity than did the case for grammar and style checkers. In her discussion of electronic conferencing in composition classrooms, Gail Hawisher (1992) observes: "When participants in an electronic conference communicate with one another . . . they are totally immersed in writing" (p. 84).

Immersion in writing is perhaps the most tangible benefit of using network-based communication (Barker & Kemp, 1990; Batson, 1989). Clearly, however, it is not the only benefit. Composition scholars have argued, for instance, that access to computer networks allows students to receive more timely feedback on their writing, to create and subsequently review transcripts of network-based brainstorming and peer-review sessions, and to communicate more easily with their teachers and classmates. The benefits associated with the ability to submit rough drafts for review as soon as a draft is completed--rather than having to wait until the next class meeting or for an instructor's office hours-- have been discussed elsewhere (Kiefer, 1991; Neuwirth, Kaufer, Keim, & Gillespie, 1988), as have the benefits of being able to review a transcript of a brainstorming or peer-review session (Bump, 1990; Moran, 1991; Neuwirth, Palmquist, and Hajduk, 1990; Neuwirth, et al., 1993). More recently, researchers have suggested that less-able students may benefit from access to computer networks. In their exploratory study of four introductory, university writing classrooms, Karen Hartman, Christine Neuwirth, and their colleagues (1991) noted that less-able students in network-supported classrooms tended to communicate more frequently with their teachers than did their peers in non-networked classrooms. Mabrito (1991), in his study of eight high- and low-apprehensive writers, found that high-apprehensive writers offered more suggestions for revision during electronic mail exchanges than during face-to-face exchanges and reported relying more on peer feedback received via electronic mail than on feedback received during face-to-face discussions.

The benefits associated with network-based communication in writing classrooms make a strong prima facie case for its use. If we are to use network-based communication effectively in writing classrooms, however, we need to explore at least two areas more fully. First, we must expand our inquiry into the ways computer networks shape curricular and classroom context. Second, and perhaps equally important, we must explore the ways in which curricular and classroom context shape network use. As I noted at the beginning of this article, network-based communication and classroom and curricular context are interrelated. It is not simply that the network shapes the curriculum but also that the curriculum shapes students' and teachers' willingness to use the network. In cases where a curriculum designer sees clear benefits to using network-based communication programs, care should be taken to create a curriculum that encourages that use.

In this article, I address both the impact of network-based communication on teaching and learning and the impact of the curriculum on students' attitudes toward and use of the network. Specifically, I report the results of a semester-long observational study of two introductory, university writing classrooms. Both classes were taught by the same instructor and students received similar training and encouragement in the use of the various communication tools available on the computer network. Results of the study indicate that, in both classrooms, frequency of network use was strongly and positively correlated with academic performance. The results also indicate, however, that students in the two classrooms used the network in strikingly different ways. Students in one classroom used the network to establish cooperative groups that met outside the classroom and, in many cases, worked together throughout the semester. Students in the other classroom, in contrast, tended to restrict their use of the network to in-class cooperative activities and to communicating with the instructor. Moreover, the results indicate marked differences between the classrooms in the quantity and quality of student responses during peer-review sessions.

Although the descriptive nature of the study precludes the development of a causal model, the results nonetheless raise an intriguing set of questions about the interrelationships between curriculum design and network support for writing instruction. In particular, the results call our attention to the pedagogical benefits of creating curricula that lead students to see network-supported communication not simply as another part of their writing assignments but rather as a valued means of completing those assignments. Below, I discuss the methods used to carry out the study. Following that discussion, I present and discuss the results of the study.

Central Questions

This article explores the interrelationships between curricular context and networked-based communication by examining ways in which students in two introductory university writing classes used the network inside and outside the classroom. To explore specific aspects of those interrelationships, five questions are addressed:

1. In what ways did students in the two classrooms use the computer network to communicate during class sessions?
   
2. In what ways did students in the two classrooms use the computer network to communicate outside the classroom?
   
3. To what extent, if any, did student use of the network reflect membership in stable, cooperative groups?
   
4. To what extent, if any, were student attitudes toward cooperative interaction correlated with their use of the network?
   
5. To what extent, if any, was in-class and out-of-class use of the computer network correlated with academic performance?
   

Participants

Twenty-nine students enrolled in two sections of the required, introductory composition course at a private Eastern university participated in the study. Both sections of the course were taught by the same instructor, an associate professor of rhetoric in the Department of English. The researcher acted as a teaching assistant in both classrooms and participated in class exercises, class discussions, student conferences, and grading of student work.

Familiarity with the Computer Network

The instructor and teaching assistant had extensive experience using the university-wide computer network and the network-supported communication tools used in the two classrooms. The students enrolled in the two classrooms received in-class training at the beginning of the semester. The students were also enrolled concurrently in a half-semester course that trained all entering students in the use of the UNIX, PC-DOS, and Macintosh computers available in the computer labs. This course focused primarily on electronic mail, electronic bulletin boards, and word processing.

Classroom Setting

The two courses met for 50 minutes three times each week. Two meetings were held in a traditional classroom equipped with a networked UNIX workstation and projector. The workstation and projector were used to share individual students' drafts with the class and to support group composing. The third meeting was held in a computer lab equipped with networked UNIX workstations with large-screen monitors.

Instruction

The curricula used in both classrooms were based on a process approach to writing instruction. Over the course of the semester, students worked on a series of assignments that contributed to and culminated in a major research paper. In both classrooms, the instructor stressed the importance of cooperative activities as part of the writing process. During class sessions conducted in the computer lab, students were encouraged (although typically not required) to engage in cooperative activities. The majority of these cooperative exchanges, as well as many out-of-class cooperative exchanges, were conducted using programs specifically designed to support classroom writing instruction (Batson, 1988; Hartman et al., 1991; Neuwirth et al., 1988). During class sessions conducted in the traditional classroom, students more often engaged in full class discussions and in model composing sessions led by students or the instructor.

The process-based approach ensured similarities in the two classrooms, but the curricula used in the two settings differed in an important way. Although both curricula focused on how to write research papers, they were based on different conceptions of research writing. Students in one classroom (subsequently referred to as the "information classroom") were taught that research writing involved obtaining and conveying information to a general audience. This approach to writing research papers is consistent with courses in which student writers learn how to interpret specialized information, typically from their own areas of study, for a lay audience.

Given this conception of research writing, in which writers write not for their peers but for an audience lacking their specialized knowledge, concerns about audience played a central role in class discussions throughout the semester. Students were asked to become "insiders" on an issue, preferably one about which they had some interest and familiarity. Topics ranged from the nature of pain to the relative merits of high-performance automobile tires to the depletion of the ozone layer. Instruction in this classroom focused primarily upon how to use the library, how to organize sources, and how to relate specialized information to a lay audience (including suggestions for reconciling conflicting claims in source materials and whether to adopt an argumentative or an informative stance). Writing instruction per se focused primarily on forming and supporting a thesis, organizing a paper, writing clearly, and considering audience.

Students in the other classroom (subsequently referred to as the "argument classroom") were taught that research writing involved interpreting and responding to texts written by members of an academic or professional community. Unlike their peers in the information classroom, students in the argument classroom were asked to write to an audience who shared their specialized knowledge of a subject. In the argument classroom, students gained that specialized knowledge by reading articles written by composition scholars such as Donald Murray (1982, 1984), Mike Rose (1983, 1984), and Maxine Hairston (1982) and articles written by students who had been enrolled in previous sections of the course. Over the course of the semester, students were required to read 12 of these articles. As they worked on their research papers, they were also asked to draw on arguments and information found in an additional 25 articles.

In keeping with this attempt to mirror, at least to a limited extent, the shared knowledge found in an academic discipline, students were asked to write articles about writing or the teaching of writing. Students chose topics ranging from the importance of grammar or of creativity in writing instruction to the need for greater consistency among teachers of writing in grading and instruction. Unlike instruction in the information classroom, instruction in the argument classroom focused only incidentally on the process of using the library, emphasizing instead the process through which academic writers interpret and respond to arguments advanced by other authors. Writing instruction dealt primarily with issues of developing an argument, responding to arguments advanced by other authors, and organizing a paper.

Observational Methods

Network-supported communication among students was assessed in two ways. First, network-based cooperative exchanges using two programs, CECE-TALK (see Figure 1) and COMMENTS (see Figure 2), were recorded and subsequently analyzed. The CECE-TALK program supported concurrent discussions among up to eight students. Each student participating in a CECE-TALK session typed in a window; their comments appeared on the computer monitors of all other individuals involved in the discussion. Typically, because of constraints imposed by the size of the screen, CECE-TALK sessions involved no more than four students. The COMMENTS program allowed students to comrnent on classmates' written drafts (see the report by Neuwirth, et al. (1988) on the design and implementation of the COMMENTS program). Students could send drafts over the network to classmates or instructors, who would subsequently comment upon and return the drafts.

Electronic mail between students and instructors was also recorded. However, neither electronic mail between classmates nor exchanges conducted using the UNIX chat utility were recorded.

Figure 1. Sample CECE-TALK exchange.

Figure 2. Sample COMMENTS exchange.

Second, in addition to recording network-based cooperative exchanges, students were asked during interviews to comment on their interactions with classmates during the semester. Students were interviewed at the beginning, middle, and end of the semester. Four students from each class, who served as case studies, were interviewed three additional times during the second half of the semester.

The content of student exchanges was assessed by coding transcripts of cooperative exchanges conducted over the computer network. In the analysis reported here, comments made on classmates' written texts were examined. These comrnents, which were exchanged using the COMMENTS program, were coded according to textual focus and function. Comments were coded as focusing on global, paragraph level, sentence-level, or word-level issues (the latter includes remarks concerning punctuation). Comments were also coded as evaluating, recommending revision, asking questions about content, or socializing. The researcher coded each comment made by students during their cooperative exchanges. A second rater coded 20% of the comments. Interrater agreement, using the kappa coefficient, was .75 (Kraemer, 1983).

In the analysis reported here, grades on the end-of-semester major paper were used to assess academic performance. [3] Grades provide a contextually based assessment of student writing performance. That is, teachers are aware not only of the "objective" quality of a text but also of the effort a student put into the text and the improvement the text represents over the course of an academic term. Grades can be, and often are, affected by factors such as a students failure to meet a deadline or to address specific assignment constraints. Although these factors often make grades appear less attractive as an evaluative measure to classroom researchers than, for example, holistic scoring by multiple raters, these factors constitute an important part of academic performance.

Question 1. In what ways did students in the two classrooms use the computer network to communicate during class sessions?

During class meetings, students in the two classrooms used CECE-TALK, the COMMENTS program, and electronic mail for three primary purposes: (1) to carry out clearly delineated writing activities, such as brainstorming and peer review; (2) to arrange meetings outside of class; and (3) to socialize. Over the course of the semester, 102 in-class exchanges were recorded: 18 CECE-TALK exchanges and 40 COMMENTS exchanges were recorded among students in the argument classroom, while 23 CECE-TALK exchanges and 21 COMMENTS exchanges were recorded among students in the information classroom. Of these, only 7 CECE-TALK exchanges were coded as strictly social exchanges.

No significant difference in mean number of network-based exchanges per student was found between the two classrooms: during the 14 weekly class sessions that they spent in the computer lab, students in the argument classroom engaged in a mean of 8.7 in-class exchanges (standard deviation = 4.1), while students in the information classroom engaged in a mean of 7.8 exchanges (standard deviation = 3.6).

Despite the similarities in the mean numbers of network-based exchanges per student, analysis of the transcripts of those exchanges indicates that students in the two classrooms defined their roles during exchanges in different ways. During peer review and brainstorming, students in the information classroom tended to focus primarily on issues of form and were less likely than students in the argument classroom either to suggest alternative interpretations of the topic being discussed or to suggest revisions above the word or sentence level. In contrast, students in the argument classroom, who were reading a shared set of articles, appear to have carried out more substantive discussions of the issues during their in-class cooperative interactions than did their counterparts in the information classroom.

This difference is reflected in an analysis of student comments on classmates' papers. In 61 recorded exchanges involving the COMMENTS program, students made a total of 341 comments on their classmates' papers. Two key differences between the classrooms emerge from our analysis of these comments (see Table 1). First, students in the argument classroom made significantly more comments on their classmates' papers during a given exchange than did students in the information classroom (F = 14.088; df = 1, 24; p < .001). The mean number of comments per exchange made by students in the argument classroom was 6.8 (standard deviation = 2.5), while the mean number of comments per exchange made by students in the information classroom was 3.5 (standard deviation = 1.8). Second, students in the argument classroom made significantly more comments suggesting revisions than did their peers in the information classroom (F = 18.219; df = 1, 24; p < .001). Among students in the argument classroom, the mean number of comments suggesting revisions was 3.2 (standard deviation = 1.7). Among students in the information classroom, the mean number of comments suggesting revisions was 0.8 (standard deviation = 0.7). The difference in mean number of comments suggesting revisions is quite likely related to the difference in total number of comments per exchange. It should be noted, however, that the differences are not symmetrical. For total comments, the ratio between the two classrooms is roughly 2 to 1; for comments suggesting revisions, the ratio is 4 to 1.
 

One of the key differences stemming from the divergent conceptions of research writing taught in the two classrooms was the extent to which students were familiar with their classmates' topics. The recorded exchanges indicate that students in the argument classroom typically shared a great deal of content knowledge about the sources their classmates were drawing from to make their arguments. In contrast, students in the information classroom typically were unfamiliar with the sources their classmates were citing. This difference may have contributed to the lower number of revision-oriented comments in the information classroom, and possibly to the lower number of total comments. The two sets of comments that follow are consistent with this speculation. In the first, a student from the information classroom comments on a classmate's paper about steroid use among high-school students. In the second, a student from the information classroom comments on his classmates' paper about the ethical issues surrounding bio-technologies. All of the exchanges took place during the final month of the 15-week semester, as students worked on their final papers. The coding for each comment is indicated in brackets. [4]

Set One: Information Classroom

Exchange 1: Steroid Use Among High School Students [5]

Comment 1:	Good job Ed. It appears that you put a lot of time and effort into this assignment. [Global Evaluation]
Comment 2:	Good into-grabber statement. [Sentence-Level Evaluation]
Comment 3:	Good objective, but I think they expect to actually see a little background. [Global Evaluation / Global Revision]
Comment 4:	This looks like the start of a thesis. Eh? [Sentence-Level Evaluation]
Comment 5:	Looks like there's plenty of room for B.S. here (at least 10 or 12 pages). [Global Evaluation]

Exchange 2: Ethical Issues Associated with Bio-Technology

Comment 1:	perhaps poor is a better word. [Word Level Revision]
Comment 2:	cute analogy [Sentence-level Evaluation]
Comment 3:	of course the invention of the airplane was greeted with a great deal of controversy. In general change is not well accepted (unless your a swingin' [College of Fine Arts] liberal) until it has proven itself. [Global Evaluation]
Comment 4:	sounds pretty good. [Global Evaluation]

Of these comments, only two suggest revision. In the first exchange, Comment 3 suggests that the writer should include background information for the reader. In the second exchange, the first comment suggests a word change. The majority of comments are evaluative and often encouraging. These characteristics are fairly typical of the exchanges among students in the information classroom. In contrast, exchanges among students in the argument classroom--in which students shared the same set of sources--were typically longer and contained more comments directed toward revision. The two exchanges presented below illustrate this:

Set One: Argument Classroom

Exchange 3: Creativity versus Mechanical Correctness

Comment 1:	overall on paradigm: great. I wont comment on grammar and style because that is precisely what you are complaining about in your paradigm! [Global Evaluation]
Comment 2:	actually, it results from an underemphesis on creativity BECAUSE of an overemphesis on grammar and structure. [Global Evaluation]
Comment 3:	If you notice, the six points in Murray's paper both recommend methods for students to deal with surprise as well as methods for teachers to deal with surprise and help students deal with it. [Global Evaluation]
Comment 4:	I think this is vague [Sentence-level Evaluation]
Comment 5:	for NCTE I try to infer a problem case based on their recomendations to teachers. [Global Revision]
Comment 6:	I would try to make sure that this definition is backed up in his paper, because it may not be accurate. I saw Rose as saying that we need more research into how students react with texts and how they learn from textbooks--cognitive learning. If we understand this better, we will be able to write better textbooks. . .I don't think he talks much about grammer/structure vs. content/style. [Global Revision / Global Evaluation]
Comment 7:	yessireebob! His solution is to do what he did -> catalyze interaction in classes and try to inject some freedom in order to let the creative juices flow. (talk about metaphors!) [Global Evaluation]
Comment 8:	yeah. Liftig and this dude will prove to be nice supporting authors for your paper. [Global Evaluation]

Exchange 4: The Need to Teach Structure as well as Grammar

Comment 1:	Is there a reason why societies recognized the importance of grammar. [Global Question]
Comment 2:	capital "B"? [Word-Level Revision]
Comment 3:	too [Word-Level Revision]
Comment 4:	consists [Word-Level Revision]
Comment 5:	I think that there are some authors that do not agree that there is any kind of step-by-step process to write a research paper. Maybe you can fit another author in there. (If there are any. I'm pretty sure though.) You may even bring it up in the next 2 paragraphs, but I can't see it clearly if it is. [Global Evaluation / Global Revision]
Comment 6:	Doesn't Barth or Murray, probably Barth, say that writing is not a skill? [Global Question]
Comment 7:	Why do I see art only at the very end. Is that the way it is supposed to be? I thought that is why you wrote this paper in the first place. [Global Revision]

The comments in these exchanges, like those in the first two exchanges, provide evaluations of several aspects of the two student papers. In addition, however, they suggest several areas of revision, ranging from word-level corrections to suggestions about how to reconceptualize the argument made by a supporting author. In general, these comments illustrate a pattern that emerges from close examination of the comments made by students in the two classrooms: comments made by students in the argument classroom, perhaps because they shared the same set of readings, focused more closely upon substantive issues than did comments from students in the information classroom.

These differences between the two classes are also reflected in qualitative assessments of CECE-TALK exchanges during classroom sessions. In the following exchange, two students from the information classroom commented on a classmate's summary of a source text using CECE-TALK.

Bob:	Hiya guys!
Alice:	Hi Bob! How do we paste our summaries in????????
Eileen:	Hey, I'm lost (as usual) . What are we doing?
Bob:	Okay, we're just going to get sent part of somebody's summary, and then we're going to tear it apart. fun, huh?
Eileen:	Geez, you type so fast Bob!
Alice:	No, it just looks that way.
Bob:	She's right, I don't type fast at all.
Alice:	Bob- let us tear your summary apart. [Bob pastes in the first paragraph of his summary.]
Eileen:	This is kind of picky but maybe you could use a different word for "in" in your first sentence.
Bob:	Eileen, maybe. I didn't give it much thought.
Alice:	Maybe "mothers" rather than "wives"?
Bob:	Alice, I didn't put this in the summary, but the big point is that these women are the wives of several powerful senators and congressmen. The important point is that they are closely related to powerful men.
Alice:	obu-kaybee.
Eileen:	Hey you know what, we had an aritcle like this in our French calss last year.
Bob:	Well, anyway, wanna see the next bit? [Bob pastes in the second paragraph of his summary.]
Alice:	Your sentences seem a little (that's beside the point) separated not relating to each other very well. I realize that too much transition could get wordy, but... well, you know.
Bob:	Try boring.
Eileen:	Actually, I think they're preety good on relaying the mainpoints of your article(or at least what I think are the main points).
Bob:	Yeah, I know. I think that the middle sentences here really do need work, but I've tried preffy hard, and they're the only way I've found that gets the point of the author across in a reasonable amount of space.

The first few rounds of this exchange are typical of discussions that took place on CECE-TALIC in both classrooms. After the program window came up on each student's computer, they exchanged greetings and defined what they were going to do during the cooperative session. In this exchange, the students also made the equivalent of idle chit-chat, discussing the speed with which Bob typed. As they discussed Bob's summary, however, Bob's cooperative partners focused primarily on surface-level issues. When Eileen and Alice suggested two minor word changes, Bob made a clear attempt to move the discussion to a more substantive level ("the big point is that these women are the wives of several powerful senators and congressmen"). Unfortunately, Bob's partners failed to take his cue. Instead, Alice responded with a simple agreement ("obu-kaybee") and Eileen observed that thearticle Bob was summarizing was similar to one she had read in her high school French class. In the second round of discussion, the focus again remained on the surface level.

In the argument classroom, in contrast, CECE-TALK exchanges tended to focus on more than surface-level issues. In the following excerpt, three students in the argument classroom discuss a summary of an article by John Barth:

Don:	Who's going first?
David:	???????????? Mine's all there.
Don:	I can't because I don't even have a summary.
David:	I hate it when that happens. I volunteer to go second.
Susan:	I am the one who confessed to not having a summary.
David:	So I'm the only one with it??????????
Susan:	BINGO
David:	OK. Here goes... [Pastes in the first part of his summary.]
David:	So what now?
Susan:	Commentary. Who's opinion are you expressing?
David:	This is from Barth.
Susan:	Did you include the information elsewhere
Don:	Are these the first sentances of the paper?
David:	What do you mean?
Susan:	The author"s name and the title of the writing.
David:	Yeah, but I didn't copy it from my summary.
Susan:	Okay.
David:	I didn't do mine as aparagraph, so this is just the first part of my outline.
Don:	OK
Susan:	Is it similar to the notecard form shown in the book?
David:	Bingo.
Susan:	Are the parts divided by topic or "seeing the issue", etc?
Don:	Let's go on to the next part. [pastes in the next part of his summary.]
Don:	I thought that the faulty path that you mentioned is more of his theory that there should be writing courses because there are a great number of people that want to learn but can't because the quality of the teaching isn't there.
David:	I don't think he was blaming it on the teaching. That seemed to be independent of the issue. I thought that he thought that the writing courses could only help if you already had the talent and drive to suceed.
Don:	Ok I see what you are saying.
David:	That's kinda my next part, so....... [pastes in the next part of his summary.]
David:	"seeing" etc.
Susan:	I want to know if these statements are supposed to encompass all the ideas which concern defining the problem.
David:	Yeah.
Don:	I thought that there were more faulty paths in this part. Did you just cut them because you thought that the were mute points.
David:	Probably were. I didn't think that they were very important. And I was a little pushed for time.
Don:	Gotcha
David:	I was just going for what I thought were the main points that he was trying to get across through the article as a whole.
Susan:	Understood.
Don:	next
David:	just a sec.......

Like the exchange from the information classroom, students began this discussion with a brief discussion of how they would proceed. Only one of the students, David, had completed a draft of his summary, although all of the students had read the source text, John Barth's (1985) "Writing: Can It Be Taught?" Initially, discussion focuses on formal issues--the author's name and the title of the article, David's admission that the summary is still in outline rather than paragraph form, and the organization of the summary. When David pasted in the second and third parts of his summary, however, the discussion became much more substantive, reflecting the students' shared knowledge of the source text. Don challenged David's interpretation of Barth's argument and later asked whether David had included all of Barth's "faulty paths" (i.e., counterarguments).

The differences between these exchanges is typical of exchanges conducted throughout the semester. Students in the argument classroom, perhaps because they shared a common set of readings, consistently made more substantive comments about their classmates' papers and consistently made more comments suggesting specific, global revisions. The differences that emerge from the analysis of the COMMENTS and CECE-TALK exchanges appear to indicate that students in the argument classroom brought more to their network-based exchanges than did their peers in the information classroom.

The extent to which students shared knowledge about the topics of their classmates' papers may also be reflected in a striking difference between the two classes. During their final four meetings in the computer cluster, students were given an assignment that allowed them to do work individually or to use the network to discuss their papers with classmates. One of these assignments, typical of the other four, is found in Figure 3. Given these options, students in the information classroom typically chose to work alone. During the last four meetings, almost 70% of the students in the information classroom chose to use the network only once or not at all. In contrast, during the last four meetings, over 80% of the students in the argument classroom chose to use the network at least twice, with more than 40% using the network more than four times. The pattern of usage in the two classrooms is shown in Figure 4.

Figure 4. Percentage of students voluntarily using the network during the last four meetings in the computer cluster.

Question 2. In what ways did students in the two classrooms use the computer network to communicate outside the classroom?

Outside the classroom, students used the network to communicate with their classmates about their writing activities, to coordinate face-to-face meetings, and to socialize. Students typically used the COMMENTS program when they reviewed classmates' papers outside the classroom. When they wanted to coordinate face-to-face meetings or socialize, they used electronic mail and the CECE-TALK program. Several students also used a fourth program, Talk, which was similar to CECE-TALK. Because it initialized more quickly than CECE-TALK, some students preferred to use it for informal, out-of-class discussions. However, Talk, a chat utility used on the campus-wide computer network, did not produce a log of an exchange.

In interviews at the end of the semester, students were asked to discuss their interactions with classmates. In response to the question, a student from the argument classroom observed:

I think ... working with other students was big, you know, because a lot of times you'd . .. understand it in class, you know, you read it in the book and it makes sense ... then you get out of class and you sit down ... wow, what the hell am I doing ... you know, I mean you are just, you are clueless ... like that last paper ... I was up in my room on [Macintosh] ... and I was bored ... so I ran VUI ...I had like five mail messages, all from people in the class ... and they were like, what are you doing and how's your paper going ... and everything like that.... So I'm sending messages back to them and they're sending messages to me and ... John's got a talk box [a UNIX chat utility similar to CECE-TALK] with Ann and Ted and ... Ann's doing this and Ted's doing this, but John doesn't think either of them are doing it right, so he's doing it this way .. . you know ... and he's like, so what are you doing, so I sent him a message back, you know, and it's, it's kind of fun.

This student's comments are typical of observations made by several of his classmates in the argument classroom. The student's observation that he found it "kind of fun" to engage in this sort of interaction outside the classroom testifies to the novelty of being able to communicate with classmates using a variety of electronic media. However, the willingness of students in the argument classroom to engage in this sort of interaction seems to be the result of more than mere novelty. Although data are available only for use of COMMENTS and CECE-TALK, the data indicate a marked difference between the two classrooms in (1) the mean frequency of out-of-class use of these programs to communicate with classmates, and (2) the number of students who chose not to use these programs to communicate with classmates outside the classroom.

An ANOVA indicates that the difference in mean frequency of out-of-class use of CECE-TALK and COMMENTS with classmates approaches significance (F = 3.066; df = 1, 27; p < .091). Students in the argument classroom engaged in a mean of 3.1 out-of-class exchanges over the course of the semester (sd = 2.7), while students in the information classroom engaged in 1.5 exchanges (sd = 2.1). In addition, the number of students who chose not to engage in out-of-class, network-based interactions differed strikingly between the two classrooms. In the argument classroom, only one student chose not to use the network to contact classmates outside the classroom. In contrast, 7 of the 13 students in the information classroom chose not to use the network.

A clear difference between the two classrooms emerged in their use of CECE-TALK. Of the 18 out-of-class CECE-TALK exchanges recorded during the semester, all but 1 were initiated by students in the argument classroom. Moreover, in their interviews, three students in the argument classroom indicated that they had also used Talk to carry out network-based discussions outside the classroom. In light of this information, it seems likely that the 18 out-of-class CECE-TALK exchanges represent a lower bound on the number of such out-of-class discussions.

Analysis of the transcripts of out-of-class CECE-TALK exchanges indicates that the purpose of two-thirds of the exchanges was to socialize. Typically, students working on computers in different buildings would use CECE-TALK to coordinate face-to-face meetings or simply to take a study break. On several occasions, however, students also used CECE-TALK to discuss class-related issues outside of the classroom. The following exchange took place during the twelfth week of the semester:

Rick:	Okay Joan. Talk to me. What's up?
Joan:	I just wanted to keep an open line of communication between us. In case of suicidal tendency flareup. Just keep the window open and glance at it every now and then.
Rick:	Fine. Just let me know if you have a problem. ...
Joan:	I'm going to make a slight change in some wording, it doesn't seem to sound right...
Rick:	Let me have it.
Joan:	This is what's there... about her minority's contribution to literature in her native country revisions... about contributions by Blacks to literature in the United States.
Rick:	I like this better. It makes the situation as a whoile seem to flow better and doesn't leave anything to the imagination
Joan:	Okay, thanks. ...
Joan:	Instead of ...dealing with her predecessors should I just say "dealing with slavery"?
Rick:	Not necessarilly. You could be talking about Blacks in the early 20's or in the sixties. You should make it more definite but not HAVING to deal with one specific thing.
Joan:	Previously in the sentence I mention that certain topics were censored, would the things you are talking about be included in this category?
Rick:	Definitely. Stir things up a little!
Joan:	What?
Rick:	Cause a few controversies in your paper and get the reader interested. there aren't many people who like to read about things being censored.

In this exchange, two students working in different computer clusters used the CECE-TALK program to carry out a conversation that they might have had if they'd been sitting next to each other. That is, the conversation continued over an extended period of time, but with frequent and lengthy pauses. When Joan wanted to ask for feedback from Rick, she began typing in her CECE-TALK window; when Rick saw the typing, he responded. Although atypical, this exchange suggests one of the ways in which network-based interaction can facilitate student writing outside the classroom.

Students also used the network to communicate with the instructor and the teaching assistant. Electronic mail and the COMMENTS program were regularly used to complete assignments; CECE-TALK was used relatively infrequently. In addition to using electronic mail to complete specific assignments (e.g., submitting a report of a group activity), almost all of the students in the two classes used electronic mail to ask unsolicited questions about an assignment or class discussion or to offer excuses for missing a class session. An ANOVA failed to find a significant difference between the two classrooms in nonsolicited electronic mail from students to instructors. Students in the argument classroom sent a mean of 2.81 (sd = 2.64) unsolicited mail messages to their instructors during the semester, while students in the information classroom sent a mean of 2.31 (sd = 1.70) unsolicited messages.

Students in both classes submitted all of their writing assignments via the COMMENTS program. Over the course of the semester, 6 papers and 5 journal assignments were submitted over the network. Several students also used the COMMENTS program to submit unsolicited rough drafts of their assignments to their instructors. In the argument classroom, 8 of 16 students submitted unsolicited rough drafts for review by an instructor; 4 of those students submitted 2 rough drafts over the course of the semester, while the other 4 submitted at least 3 rough drafts. One student submitted 8 rough drafts over the course of the semester. In the information classroom, 4 of 13 students submitted at least 1 unsolicited rough draft for review. None of those 4, however, submitted more than 2 drafts during the semester.

Several students also used CECE-TALK to communicate with their instructors outside the classroom. Typically, students used CECE-TALK to discuss comments they had received on an assignment. On some occasions, however, the discussions were relatively brief and to the point:

Instructor:	what's up?
Don:	[Instructor's name], is there some kind of on line thesaurus on andrew?
Instructor:	no. sorry about that
Don:	Oh well, thanks

Question 3. To what extent, if any, did students' use of the network reflect their membership in stable, cooperative groups?

Analysis suggests that students' use of the network inside or outside of the classroom reflected their membership in stable, cooperative groups. [6] These groups were characterized by a tendency to work together during class, to communicate with each other over the network, and to meet outside of class. Fifteen of the 16 students in the argument classroom formed 5 cooperative groups, with 2 of those students belonging to two groups. Nine of the 13 students in the information classroom formed 3 cooperative groups, with 2 of those students belonging to two groups.

Throughout the semester, students were allowed to choose collaborative partners for classroom writing activities. In addition to allowing students to work with classmates with whom they felt most comfortable, this approach also provided an opportunity to determine whether students would establish stable cooperative groups. During class sessions, students in both classrooms typically chose to communicate with members of their collaborative groups during in-class cooperative activities. Of 102 in-class exchanges using CECE-TALK and COMMENTS, only 21 exchanges involved students who were from different cooperative groups. Students in the information classroom were more likely to engage in cross-group interaction than were their peers in the argument classroom: 28% (12 out of 43) of interactions in the information classroom involved students from different cooperative groups, while 19% (11 out of 58) of interactions in the argument classroom involved students from different cooperative groups (see Figure 5) . Between-classroom differences in the amount of out-of-class, network-based interaction among students make it difficult to make meaningful comparisons of inter- and intra-group communication among students in the two classrooms. Analysis of interactions in the argument classroom suggests, however, that out-of-class use of the network reflected membership in cooperative groups (see Figure 6). All but 1 of the 17 out-of-class CECE-TALK exchanges among students in the argument classroom were intra-group exchanges. A similar pattern was found for exchanges using the COMMENTS program, with 8 out of 10 out-of-class COMMENTS exchanges occurring among members of stable collaborative groups.

Figure 5. In-class use of COMMENTS and CECE-TALK within and across cooperative groups.

Figure 6. Ou-of-class use of COMMENTS and CECE-TALK within and across cooperative groups.

Question 4. To what extent, if any, were student attitudes toward cooperative interaction correlated with their use of the network?

During interviews conducted at the end of the semester, students in both classes were asked whether factors such as interaction with teachers, interaction with classmates, individual and group assignments, and individual aptitude had contributed to their efforts to learn to write research papers. Student attitudes toward the value of cooperative interaction appear to differ between the two classes. Only 5 of the 13 students in the information classroom indicated that they thought interaction with classmates helped them improve as writers of research papers. In addition, 3 students in the information classroom explicitly indicated that they avoided interaction with their classmates--in each case, these students indicated that they felt their classmates could not provide them with quality feedback on their writing. In contrast, 14 of the 16 students in the argument classroom explicitly indicated that they thought interaction with classmates helped them improve as writers of research papers. None of the students in the argument classroom indicated that they avoided interaction with their classmates.

Students' attitudes toward using the network appear to be strongly linked to their perceptions of the value of interaction with classmates. Students who indicated in their interviews that they valued cooperative interaction with classmates were more likely to use the network--both inside and outside the classroom--than were students who did not explicitly indicate that they valued cooperative interaction. An ANOVA found a significant difference in total number of network-based interactions between the two groups of students (F = 4.914; df = 1, 27; p < .035). Students who indicated that interactions with classmates helped them to improve as writers of research papers engaged in a mean of 8.5 network-based interactions with classmates over the course of the semester (sd = 3.5), while students who did not indicate that cooperative interaction helped them improve as writers of research papers engaged in a mean of 5.2 network-based interactions with classmates during the semester (sd = 4.4).

Question 5. To what extent, if any, was in-class and out-of-class use of the computer network correlated with academic performance?

Regression analysis indicates significant positive correlations in both classrooms between frequency of out-of-class interaction over the computer network and academic performance. Academic performance was assessed on the basis of the grade on the final paper, which represented the culmination of work over the course of the semester. [7] In the information classroom, use of the computer network outside the classroom was strongly and positively correlated with grades on the final paper (F= 8.111; adjusted R2 = .372; standardized coefficient= 0.651; df = 1,11; p < .016). A moderate, positive correlation was found among students in the argument classroom (F = 3.780; adjusted R2 = 156; standardized coefficient = 0.461; df = 1,14; p < .072). [8] The analysis failed to find a significant relationship between in-class use of the network and grades on the final paper in either classroom.

Caution is indicated in interpreting these findings, particularly given the small number (9) of recorded out-of-class exchanges among students in the information classroom. Moreover, of the 13 students in the information classroom, 7 engaged in no out-of-class interactions. It is worth noting, however, that the correlation was found in both classrooms, which seems to indicate that the results are not simply a statistical anomaly. These results suggest some promising directions for further research. [9]

One avenue for such research would be to determine whether out-of-class network-based interaction in some way contributed to higher grades on the final paper. Other avenues suggest themselves as well. In addition to the possibility that out-of-class interaction in some way caused better grades on the final paper--perhaps because students spent more time working on their papers or perhaps because they became more engaged in their topics--we can also look at the possibility that students who received better grades might have tended to spend more time on the network simply because the teachers in the two classrooms encouraged them to do so. I am referring to what might be called the "good student" syndrome. That is, students who put more effort into the course--and as such might be expected to receive better grades on their papers--might have picked up on cues provided by the teacher and teaching assistant that using the network was looked upon favorably. Classroom observation indicates that students in both classrooms were told of the value of using the CECE-TALK and COMMENTS programs, and the extensive use of those programs during classroom activities underscored the value attached to them by the instructors.

If it was the case that stronger students used the network, then we might also expect to find a similarly strong correlation between students' performance on the final paper and their academic ability (reflected, perhaps, in measures such as their grades in previous writing courses, analysis of portfolios of student writing conducted prior to enrolling in the course, or standardized tests such as the SAT or ACT). In this study, the only available measure of students' academic ability prior to enrolling in the course was their SAT scores. Regression analysis, however, failed to find a significant correlation in either classroom between students' SAT verbal scores and their grades on the final paper.

It could also be hypothesized that students sought out classmates who had most clearly grasped the concepts being discussed in the course. Since those students would be more likely to be sought out, it would seem reasonable to find a correlation between performance on the final paper and frequency of interaction between classmates. This hypothesis fails to account, however, for the strong tendency of students to interact within established cooperative groups.

Another hypothesis suggests that students who were using the network to communicate with classmates were probably spending more time working on their papers. In this hypothesis, the key factor would be time on task: students who spent more time working on their papers might also be more likely to spend more time using the network to communicate with their classmates.

In addition to entertaining these hypotheses, we might also ask why a correlation was found between academic performance and out-of-class interaction, but not in-class interaction. Although the best way to examine this question would be to conduct follow-up research, we might also tum to discussions of the structure of authority in classrooms (Gere, 1987). Although the assignments given to students, particularly during the final four weeks of the semester, allowed them to choose not only with whom, but also whether they engaged in network-based in~lass discussions, it seems possible that students may have found these discussions less useful than those conducted outside the classroom. Among other reasons, students may not have been prepared to discuss their papers--as they would be if they decided to contact someone outside the classroom. Students may also have found their collaborative partners during class sessions less prepared to respond to their papers and/or less likely to stay on-task than might be the case if they set up a meeting outside of class.

In addition to discussions of the structure of authority in writing classrooms, we might also consider the functions of the discussions themselves. A large number of both out-of-class and in-class network-based interactions were conducted for essentially social purposes; often, students used CECE-TALK discussions to coordinate study breaks or to find out if classmates wanted to pin them for lunch, where presumably at least some of the discussion might touch on course-related issues. Subsequent research may well indicate that it is not the content of the network-based discussions so much as their usefulness in facilitating face-to-face meetings that is related to academic performance.

This study provides insights into the ways that use of network-based communication both shaped and was shaped by the curricula used in two introductory writing classrooms. In both classrooms, students made extensive use of the network for student-instructor interactions. Over the course of the semester, students in both classrooms submitted 11 required writing assignments via the COMMENTS program. In addition, students in both classrooms used the COMMENTS program and electronic mail to submit unsolicited rough drafts and to discuss course-related concerns with the instructors. Students also used the network extensively during in-class cooperative activities, although some differences were noted between the two classrooms in the frequency and content of in-class network-based interaction among classmates. Finally, students used the network outside the classroom to discuss their writing, to socialize, and to arrange face-to-face meetings, although between-classroom differences were again noted in the frequency of out-of-class, network-based interactions.

The impact of the network on the way writing was taught and learned in the two classrooms is perhaps best illustrated by examining the similarities in use of the network in the two classrooms. Access to the network, for instance, made it much easier for students to receive feedback on their rough drafts in a more timely fashion than is typically the case in a traditional classroom setting. Access to the network also made it easier for students to contact their instructors when they had concerns about the class or their performance as students. Although the mean level of unsolicited electronic mail from students was on the order of two to three messages, 5 students sent at least 5 electronic mail messages (and one student sent 9) during the semester. In addition, several students included messages in papers submitted using the COMMENTS program. Network access also appeared to reflect--and perhaps contributed to the establishment and maintenance of--membership in cooperative groups in the two classrooms. Roughly 80% of all network-based interactions were intra-group discussions. Finally, the use of the network during in-class cooperative activities shaped both the way students communicated with each other during the activities and their ability to reflect on their conversations during the activities as they subsequently revised their papers. That is, the programs students used to conduct their in-class cooperative activities provided both students and instructors with a transcript of their discussions. [See Neuwirth, Palmquist, and Hajduk (1990) and Neuwirth et al. (1993) for a discussion of student and instructor access to transcripts of peer-review sessions].

The impact of the curricula used in the two classrooms on the way students used the network is perhaps best illustrated by examining differences in the way the network was used. Three aspects of network use appear most relevant: (1) the frequency with which students used the network, (2) student attitudes toward using the network, and (3) the content of student discussions over the network. Students in the argument classroom used the network more frequently than did their peers in the information classroom. Although the mean levels of in-class use of the network do not differ significantly, in-class use of the network during the last four weeks of class--when students were given the options of using the network or of working independently-- differed markedly between the two classrooms. Nearly 70% of students in the information classroom chose to use the network once or not at all during the final four weeks of the course, while over 80% of students in the argument classroom used the network at least twice and over 40% used the network at least four times. Students in the argument classroom were also much more likely to use the network outside the classroom. Students in the argument classroom used the COMMENTS and CECE-TALK programs 27 times outside the classroom, while students in the information classroom used the programs only nine times. [10]

Student attitudes toward collaborative work--and by extension toward using the network--strongly reflect frequency of use. In both classrooms, students who indicated that they valued interaction with their classmates used the network significantly more frequently than did students who did not indicate that they valued interaction. In addition, a clear between-classroom difference was found in student attitudes toward interaction. Fourteen of 16 students in the argument classroom explicitly indicated in student interviews that they valued interaction, while only 5 of 13 students in the information classroom did so. In-class use of the network during the last four weeks of the semester and out-of-class use of the network throughout the semester clearly reflect this difference in attitude.

Finally, the content of student discussions across the network reflects differences in the curricula used in the two classrooms. Perhaps because students in the argument classroom shared a common set of readings and, to some extent, wrote about similar topics, network-based peer-review sessions in the argument classroom appear to have been more substantive than were those in the information classroom, where students compiled individual reading lists and wrote about dissimilar topics. In their network-based interactions, students in the argument classroom made significantly more comments overall and significantly more comments suggesting revisions than did students in the information classroom.

The observations reported in this study do not fully capture the complexity of network-based interactions among students in the two classrooms--specifically, they do not include transcripts of electronic mail messages among classmates and they do not include conversations conducted over the UNIX chat utility. Despite this limitation, the results of the study suggest numerous ways in which network-based communication provides opportunities for interaction both inside and outside the writing classroom. Perhaps most important, the results suggests that network-based communication may interact with face-to-face discussions in ways that lead to better academic performance. The results of this study also underscore the importance of viewing network-based communication not only as a means of extending or even of transforming the writing classroom, but also as a technology whose use can be shaped in important ways by classroom and curricular context. That is, the results of this study suggest the importance of viewing network-based communication and writing instruction as activities that are essentially interrelated.
 

Michael E. Palmquist is an assistant professor of English at Colorado State University. His research interests include the use of networked-computers in writing classrooms, computer support for writing across the curriculum, and instructional applications of hypertext. Michael Palmquist's e-mail address is MPALMQUIST@VINES.COLOSTATE.EDU.
 

1. I would like to thank Kate Kiefer, Chris Neuwirth, Stephen Reid, and Dawn Rodrigues for their comments on drafts of this article. I would also like to thank the chair of my dissertation committee, David Kaufer, for his help in collecting data for the study and for his guidance as I wrote my dissertation.
   

2. Electronic Networks For Interaction (ENFI) referred initially to a synchronous communication program developed in 1985 at Gallaudet University by Trent Batson. Batson and others now use the term to refer to a wide range of concurrent communication technologies, among them UNIX chat utilities, DOS-based synchronous communication programs such as REALTIME WRITER and Daedalus INTERCHANGE, and Apple Macintosh-based programs such as Group Technologies' ASPECTS. ENFI programs allow individuals working on networked computers to type messages to each other in real time.
   

3. The teaching assistant assisted with grading throughout the semester but did not assign grades on the final paper. The instructor was not informed of student interaction patterns, the measure to which grades were compared in the analysis reported in this study, until after the semester had ended.
   

4. All transcripts of cooperative exchanges presented in this article are presented as students typed them. Misspellings and other mechanical errors have not been corrected. Names of students, both in comments and in labels elsewhere in this article, have been changed to preserve the anonymity of participants in the study.
   

5. As students read their classmates' comments, the text passages to which the comments were directed were highlighted. Comments could refer to single words, to a sentence or paragraph, or to the text as a whole. Because of space limitations, the passages to which comments are directed are not included in this article. However, the way in which the comments appeared to students is illustrated in Figure 2.
   

6. Membership in these groups was determined from weekly student interaction reports, student interviews, and recordings of network-based exchanges. For a fuller discussion of out-of-class, face-to-face interaction among students from the two classrooms, see Palmquist (1993).
   

7. The final grade for the course was also considered as a measure of academic performance. Because the course grade was based on work conducted throughout the semester, it was assumed to be a less sensitive measure of the cumulative effect of a semester of cooperative interaction.
   

8. For an explanation of our characterization of small, medium, and high correlations in this study, see Cohen's (1987) discussion of effect sizes in the behavioral sciences, pp. 78-81.
   

9. Kate Kiefer and I are currently conducting a study of interaction in four networked and four traditional composition classrooms at Colorado State University.
   

10. Figures for out-of-class use of the network do not include electronic mail between classmates or use of the Talk program, a UNIX chat utility that was similar to CECE-TALK. Student interviews indicate that students in the argument classroom regularly used electronic mail and, less frequently, Talk. Interviews with students in the information classroom suggest that their use of these programs reflects their level of usage of COMMENTS and CECE-TALK, i.e., quite low.
    

Barker, T. T., & Kemp, F. O. (1990). Network theory: A postmodern pedagogy for the writing classroom. In C. Handa (Ed.), Computers and Community, 1-27. Portsmouth: Boynton-Cook.

Barth, J. (1985, June 16) Writing: Can it be taught? New York Times Book Review, 90, pl.

Batson, T. (1988). The ENFI Project: A networked classroom approach to writing instruction. Academic Computing, 2(5), 55-56.

Bump, J. (1990). Radical changes in class discussion using networked computers. Computers and the Humanities, 49, 49-65.

Cherry, L. (1982). Writing tools. IEEE Transactions on Communications, 30(1), 100-105.

Cohen, J. (1987). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.

Collins, J. L. (1989). Computerized text analysis and the teaching of writing. In C. Selfe & G. E. Hawisher (Eds.), Critical perspectives on computers and composition instruction (pp. 30-43). New York: Teachers College Press.

Dobrin, D. (1986). Style analyzers once more. Computers and Composition 3(3), 22-32.

Gere, A. R. (1987). Writing Groups: History, Theory, and Implications. Carbondale, IL: Southern Illinois University Press.

Gingrich, P. S. (1983). The UNIX Writer's Workbench software: Results of a field study. The Bell System Technical Journal, 62(6), 1909-1921.

Hairston, M. (1982). The winds of change: Thomas Kuhn and the revolution in the teaching of writing. College Composition and Communication, 33(1), 76-88.

Hartman, K., Neuwirth, C. M., Kiesler, S., Sproull, L., Cochran, C., Palmquist, M., & Zubrow, D. (1991). Patterns of social interaction and learning to write: Some effects of network technologies. Written Communication, 8(1), 79-113.

Hawisher, G. (1992). Electronic meetings of the minds: Research, electronic conferences, and composition studies. In G. E. Hawisher & P. LeBlanc (Eds.), Re-Imagining Computers and Composition: Teaching and Research in the Virtual Age (pp. 81-101). Portsmouth, NH: Boynton-Cook.

Kiefer, K. (1991). Computers and teachers education in the 1990s and beyond. In G. E. Hawisher and C. Selfe (Eds.), Evolving perspectives on computers and composition studies: Questions for the l990s (pp. 113-131). Urbana, IL: National Council of Teachers of English.

Kiefer, K., Reid, S., and Smith, C. (1989). Style analysis programs: Teachers using the tools. In C. Selfe, D. Rodrigues, & W. Oates (Eds.), Computers in English and the language arts: The challenge of teacher education (pp. 213-226). Urbana, IL: National Council of Teachers of English.

Kinkead, J. (1986). Matching software and curriculum: A description of four text-analysis programs. Computers and Composition, 3(3), 33-55.

Kraemer, H. C. (1983). Kappa coefficient. In S. Kotz & N. L. Johnson (Eds.) Encyclopedia of Statistical Sciences (Vol. 4, pp. 352-354). New York: John Wiley and Sons.

Liftig, R. (1982). After basics. English Journal, 71, 47-50.

Mabrito, M. (1991). Electronic mail as a vehicle for peer response. Written Communication, 8(4), 509-532.

MacDonald, N. H., Frase, L. T., Gingrich, P. S., & Keenan, S. A. (1982). The Writer's Workbench: Computer aids for text analysis. IEEE Transactions on Communications, 30(1), 105-110.

Moran, C. (1991). Using what we have. Computers and Composition, 9(1), 39-46.

Murray, D. M. (1982). Teaching the other self: The writer's first reader. College Composition and Communication, 33(2), 140-147.

Murray, D. M. (1984). Writing and teaching for surprise. College English, 46(1), 1-7.

Neuwirth, C. M., Palmquist, M., Cochran, C., Gillespie, T., Hartman, K., & Hajduk, T. (1993). Why write-together-concurrently on a computer network? In B. Bruce, J. Peyton, & T. Batson (Eds.), Network-based Classrooms: Promises and Realities (pp. 181-209) Cambridge, MA: Cambridge University Press.

Neuwirth, C., Kaufer, D., Keim, G., & Gillespie, T. (1988). The COMMENTS program: Computer support for response to writing (Report No. CECE-TR-3). Pittsburgh: Carnegie Mellon University, Center for Educational Computing in English.

Neuwirth, C., Palmquist, M., & Hajduk, T. (1990, April). Collaborative writing and the role of external representations. Paper presented at the annual meeting of the American Educational Research Association, Boston.

Palmquist, M. (1993). [Out-of-class cooperative interaction and learning to write.] In preparation.

Rose, M. (1983). Speculations on process knowledge and the textbook's static page. College Composition and Communication, 34(2), 208-213.

Rose, M. (1985). The language of exclusion: Writing instruction at the university. College English, 47(4), 341-59.

Selfe, C. (1991, March). Computer networks and on-line conferences in writing-intensive courses. Paper presented at the Conference on College Composition and Communication, Boston.