Cooperative Writing in a Hypertext Environment

Originally published in the Proceedings of the Conference on Computing and Composition, June 1994. Based on a presentation doen at the conference in 1993.

Back to Kevin's Home Page

ABSTRACT: Viewing hypertext (or hypermedia) as a literary device instead of a computer application enables a much more flexible use of technology to set up a situation that requires cooperation among students to complete a class project. Hypermedia is seen as a new way to organize thoughts, and is usually used in education for presentation of material, but with limited results in language classes. Involving the students in creating a hypertext (inter)actively increases their awareness of traditional organization of the writing process and its alternatives. There is an added advantage of forcing same-L1 students to interact in L2 to adequately complete the project.

Key Words: Hypermedia, Hypertext, Computers, Composition, Projects

I. Introduction

The Japanese language student often has more problems with expression than other cultures. It is not in the scope of this paper to reason why, but to deal with this fact and use a new technology to overcome this difficulty. Of the two basic modes of expression, students of language tend to be more comfortable with the written word, even though it requires more effort. Students in the Japanese system currently enter universities with little or no knowledge of organization on a discourse level (Robson, 1991). Students, if they have learned any kind of written structure in Japanese, cannot transfer it to English, as the classic organization is one that is not common to English.

These students are often taught to organize their papers along discourse types, such as comparison and contrast, persuasion or description. These discourse types are often combined with a process approach, following construction of a written work from topic selection, collection of ideas, organization, revision, editing and proofreading. An alternative to this type of organization is to build "nodes" instead of paragraphs, and when each student has a number of these information packets ready and organized, they can first "link" the related information between the nodes in their own work, followed by linking to other less-related information among members of their small group, and finally among other small groups. Each level is more complicated than the last, but the same essential rules for connection apply.

What is a hypertext?

Applications of hypertext include on-line manuals and help functions. Expert systems (sometimes called knowledge bases) offer an information-rich decision-making structure in particular fields. Even lowly but pioneering computer adventure games incorporate decision trees with which the player can experiment. Interactive text, where the reader chooses different paths of action at critical points in the plot is perhaps the clearest but simplest example of hypertext.

Hypertext is more than a software program, and more than an authoring language, and it is more than making decisions. Hypertext is a psychological construct that enables information to be quickly and easily accessed in the order needed each time it is needed. A network of blocks of information (nodes) connected to each other (links) by indicating one special part of that block of language (buttons) for amplification is how hypertext works. Another manifestation of hypertext adds graphics and sounds to make hypermedia. (I will include hypermedia in the definition of hypertext from this point.)

In a hypothetical hypertext session, one might sit down at the computer with the intention of studying, say, Mozart. You might discover along the way that he could play a Stradivarius violin while standing atop a Luis XIV chair blindfolded. At the touch of a button, you could call up information about the maker of the violins or how the chairs compared to their earlier styles (with pictures, of course). From there, you might explore how Luis himself met his end, and from there you could explore all the French monarchy and how it lead to the revolution, and we still haven't touched on what Mozart played while atop the chair.

Hypertext is not necessarily electronic. The front page of a newspaper has aspects of hypertext; your eyes move around, you don't read in order, you only read what interests you. A library is another good example of hypertext, with the card catalog as the core. It is logical that so many of these card catalogs have been computerized.

How does Hypertext relate to Computer-Assisted Language Learning?

Computer Assisted (or Aided) Instruction (CAI) focuses on using digital technology for instructional purposes. This differs from Computer Assisted Learning (CAL) which uses the computer to enhance the learning process. Hypertext functions more efficiently for the latter than the former. A CAI system is usually less amenable to user input than a CAL system, and therefore creates a greater challenge to the hypertext developer.

Computer Assisted Instruction has gone through three main periods of development. Initially computers were used for "programmed learning," based on behaviorist research (Skinner, 1958). This has lead to a modernized version of Computer Based Instruction (Alessi and Trollip, 1985). The initial linear programmed structure has been expanded with branching techniques, allowing for a certain amount of customizing depending on the feedback from learner behavior (Dumslaff and Ebert, 1992).

Computer Based Instruction (CBI) can be characterized by activities such as tutorial instruction, where a simulated "dialogue" is carried on by the computer and the student, by drills, simulations, which imitate some phenomenon such as going to the post office, by instructional games and by tests (Alessi and Trollip, 1985). Problem-solving programs would not be considered a part of CBI.

Artificial Intelligence (AI) enjoyed a period of application to education in the late seventies and early eighties, when it was considered to have great potential. AI systems program the computer to mimic the human decision-making process using logic. These work best "where the problems we are dealing with are very large and somewhat ill-defined." (Littleford, 1991, p. 364)

AI systems use operators such as IF and THEN along with truth statements to operate on a body of knowledge. The system continues asking questions of the user until there is enough information to make a decision. Some of the most common knowledge-based or expert systems work with highly technical, narrowly defined fields that require a large quantity of memorization and have very explicit processes, such as medical diagnosis or legal advice.

Unfortunately, AI has proved to be very expensive in time and effort (an expert programmer is required with a systems manager and a field expert, such as a doctor) to create this kind of system (Muhlhauser and Schaper, 1992). Educational applications have not attained the expected quality for the investment required, and therefore are still relatively rare. One current hope is to apply this kind of system to writing other systems: Computer Assisted Software Engineering (CASE). Hypertext has used AI on a limited basis to improve weaknesses in navigation, but application here is still rudimentary.

Hypertext is somewhere between these two extremes and offers one of the most viable areas for educational applications (Stevens, 1986). Hypermedia systems can either be geared toward presentation of material or creation and storing of material (Hammond, 1989). This is a crucial distinction in education, the first working in an active way with the more traditionally considered "passive" skills of information processing, that is, reading, listening, understanding and organizing. The second, using hypertext to create material, uses these skills with other, more "productive" skills such as writing, concept grouping and argumentation

Using Hypertext for Presentation of Material- Some Problems

Developers have noted that hypertext leads to difficulties in cases where the user is not familiar with the concept or operation of systems. It should be noted here that hypertext is not a panacea, but a useful tool when used in combination with other, more traditional methods of teaching. Indeed, hypertext aside, for things like reading, studies show that getting information from a screen is more difficult and slower than from paper (Mason, 1987). As technology advances, though, this will change.

First, most common, and most important, users of hypertext can sometimes find themselves in a node of information in which they have no particular interest, while at the same time not able to return to familiar territory (lost, if you will). The simplest remedy for this navigational problem in the hypertext system is to constantly provide a backtrack mechanism, where retracing one's steps is possible.

This leads to another problem- the user knows information is present, but cannot find it (perhaps they passed over it hours before and cannot retrace that far). This is usually a consequence of having too many branches, or alternative paths, within the system. This plethora of possibility and information also leads to insecurity by the user when a major path decision (a "fork") is to be made- leading to a slowdown in use of the system (Jones, 1987). This can be rectified by creating an index of sorts (a pop-up window on the screen with pertinent areas in the text).

A third common problem is that users become disoriented because they have no concept of the overall text. Maps, or graphic representation of the field of knowledge presented, help here. As Hammond and Allison state in a 1988 experiment:

All users had available the basic hypertext mechanism for traversing links from one frame of information to another, but some users also had access to additional facilities, such as maps (providing an overview), an index of key words, and a number of guided tours through the material. Compared with users of systems with these additional facilities, users of the basic hypertext version thought they had seen the most material when in fact they had seen the least. (Hammond, 1989. p. 170)

The fourth symptom of a badly designed hypertext is that users wander aimlessly through the hypertext. Browsing is defined by Marchionini and Shneiderman (1988) as "an exploratory, information-seeking strategy that depends on serendipity. It is especially appropriate for ill-defined problems and for exploring new task domains." Jones (1989) shows that browsing has no significant measurable effect on information retrieval and incidental learning when compared to fact-finding in a hypertext system that encourages browsing (Grolier's Electronic Encyclopedia on CD-ROM).

Jones' study did show that when incremental acquisition of vocabulary is taken into account (users acquire parts of words before their complete form), the results are significant for browsing versus fact-finding (with browsing being less efficient). This indicates (only 40 subject in a 2X2 design) that the incremental acquisition hypothesis (Herman, et.al, 1987) is valid for hypertext use. It also shows that some kind of guidance system is beneficial for vocabulary acquisition of hypertext users.

This does not mean that browsing is an inappropriate activity for learning, but it is a symptom of a final problem, one of general efficiency. New users to a hypertext system will experience a cognitive load in learning the system, one that is reduced with experience, but needs to be overcome nonetheless. "Users can be classified along three continua: frequency of use, complexity of application, and general range of computer experience." (Marchionini and Shneiderman, 1988) The remedy for this problem is a menu-controlled, or structured, browsing system. This may seem a contradiction in terms, especially when it is contrasted with information retrieval systems that work similarly. The complexity of the application is reduced through guidance, with appropriate allowances for freedom of movement (indexes and maps) which can offset the restrictions imposed by the guidance.

A key development in the field is context-sensitive guidance, which can be set to the user level so that the convergence of freedom and navigability reach their optimum levels within a few minutes of use. The setting of this level can be either programmed into the system, or overridden by the user. This reduces the "learning curve" of the system so that users can begin obtaining information appropriate to their needs within a very short time, reducing the cognitive load of the system (Smeaton, 1991).

Guidance and mapping/indexing, collectively called selection mechanisms, are one aspect of the user interface. Feedback is the second, allowing for more user freedom. The third is input/output devices for collecting, collating and exporting data.

The two most important considerations in a hypertext system are the user interface and user control. Two types of control are important to the learner (as opposed to the researcher). These are control over the sequencing of the material and control over the types of learning activities (Hammond, 1989). The amount of control depends on the user's experience with the material, experience with computers, and the task to be completed

Hypermedia and the language learner in Japan

Hypermedia learning systems do not work well with Japanese language learners initially. The self-guiding aspect of these types of programs is alien to the communicative style most Japanese bring to the classroom. Students tend to feel more comfortable with the earlier drill-type software, but they soon become bored with this. There is a strong parallel to the debacle of the audio language laboratories (of which Japan has the largest preponderance, and of those, less than 25% are being used as they were intended to be). Note that these labs are of the old style, where students operate in unison with the instructor controlling the hardware and all of the interaction is programmed ahead. The newer style audio labs, centering on individual workstations and student control of the material, presentation speed and style, have not made significant inroads into Japanese language teaching.

Japanese women in particular are threatened by computers. In a study by Vasilios Makrakis of the University of Crete, published in Computers and Education in March 1993) he showed very clearly that 15 year-old Japanese women demonstrate the "We can, I can't" phenomenon where students feel unsure of their own individual ability, but as a group feel competent. Not surprisingly, people perceived computer ability rises with the time they spend at computers. Boys see more applicability to their lives for computer use. Boys are often encourages where girls are not. The number of girls in computer-related fields in Europe are going down despite national incentives to the opposite.

Makrakis' hypotheses are that self-perceived computer efficacy is based on:

"(1) perceptions of usefulness of computing skills for one's future careers;

(2) occupational aspirations with respect to computers;

(3) intrinsic motivation towards computers, described as "personal enjoyment' and 'interest in the study of computers';

(4) the association of computers with maths and sciences

No gender differences for hypotheses (3) and (4). Both boys and girls indicated low levels of perceived efficacy. Occupational aspirations, usefulness of computers and ease of computers explained most of the difference in perceived efficacy in both boys and girls. Occupational aspirations were the biggest factor for men, but less so for women (as they are less career motivated than men. "Both parents and teachers exerted and insignificant influence on students' engagement in computer learning." Teachers, although they encouraged males more, did not affect self-efficacy measures. Teachers were prevented from more influence by lack of systematic in-service training. Word processing had been found to have a marked effect on the enhancement of self-efficacy and self-esteem in regards to computer use

Hypertext as a Classroom Project

During the final four weeks of a "Writing With Computers" course, I asked my students to create a hypertext on our university. I chose this topic because it was a well-known subject to all with easy access to raw information, and lent itself to precise organization. (I could also then use the final product to show the president to get increased budget allocations, but this is another matter.)

After allowing them to explore a sample "hypertext on hypertext," I had the students create a very simple branching fiction. The starting point had three options, and each of these three options had three further options, leading to nine separate endings. We then explored the possibilities of cross-connecting each of these endings to one of the other intermediate steps (a "goto" in computer parlance). From this they were familiarized with the organizational possibilities of hypertext.

The next step was to create a type of "research paper" consisting of ten paragraphs, or "nodes." The organizational process began with topic selection, for each class of 20 was broken into five small groups of four. Each group had to decide on a topic large enough to accommodate four sub-topics, each with ten paragraphs. We negotiated for topic placement, proof of sufficient interesting data the criterion. Group and individual topics defined, we set about researching, following a traditional process approach.

The last three class sessions were devoted to organization of the hypertext. Because of student unfamiliarity with the software, we printed out the individual final product on paper and assigned each group a category number in the hundreds. For example, Buildings as the 100's, History as 200's, Curriculum as 300's. Individual students would number each of their nodes within that parameter, such as 210-219 for the foundation of the university, 230-239 for the war-time period, 250-259 for post-war expansion, and 270-279 for recent history. Note that this scheme allows for future expansion and "filling of gaps," as a hypertext is never finished, and there is always onther set of students the next semester to enhance the hypertext.

After number assignment, the students connected information within their individual ten nodes if there were semantic links. Coaching was intensive to ensure that each student understood the concept. The second stage was for each of the small groups to meet and connect inter-related ideas. This was the richest of the three sessions, because there already was an overlying relationship that only had to be quantified and made explicit. An underlying thread of student living conditions, for example, might appear in node (or card) 234 and 272 in the example above. Each of the student would note on the paper where the connection was based and what number node it lead to.

The final class session was devoted to inter-group connections. Students would shout out possible links, such as "I have something on the bell tower and the songs it plays." and others might respond, "I have something on building where the bell tower is," or "I have something on the President's penchant for collecting bells." Appropriate buttons, links and numbers would then be assigned as the cacophony of other subjects were interrelated. I then took these print-outs and entered them into the software.

Final examination consisted of the student reviewing the final product, followed by an interview. The student evaluated their participation in the project. I listened. Each student was given a final grade based on their individual work (40%), their small group aggregate (30%) and the class as a whole (30%). Overall, if students could be lead away from listing as an organizational methodology (used more often than was appropriate), the results were clear, concise paragraphs that did not contain unnecessary verbiage and got the point across effectively. The entire project was then entered into the software and resulted in an excellent guide for new students

Cited Literature

Alessi, Stephen M. and Trollip, Stanley R. Computer Based Instruction: Methods and Development. Prentice-Hall, Englewood Cliffs, NJ. 1985.

Allison, Lesley "Learning Styles and Computer-based Learning Environments" In Computer Assisted Learning- 4th International Conference, ICCAL `92, I. Tomek, ed. Berlin: Springer-Verlag, 1992.

Bloom, Benjamin S. Taxonomy of educational objectives; the classification of educational goals, Handbook I: Cognitive Domain. David McKay Co.:New York, 1956.

Hammond, Nick "Hypermedia and Learning: Who Guides Whom?" In Computer Assisted Learning: Proceedings of the 2nd International Conference of ICCAL, ed. Berlin: Springer-Verlag, 1989.

Higgins, John Language, Learners and Computers. Longman, London: 1988.

Jones, Tricia "Towards a Typology of Educational uses of Hypermedia" In Computer Assisted Learning, D.H. Norrie and H.-W. Six, ed. New York: Springer-Verlag, 1990.

Jones, W.P. "How do we distinguish the hyper from the hype in non-linear text?" In Human-Computer Interaction - Interact `87, Bullinger, H.J. and Shakel, eds. Amsterdam: North-Holland, 1987.

La Passardierre, Brigette and Dufresne, Aude "Adaptive Navigational Tools for Educational Hypermedia" In Computer Assisted Learning, I. Tomek, ed. Berlin: Springer-Verlag. 1992.

Nelson, Theodore H. "Replacing the Printed Word: A Complete Literary System" In Information Processing 80, S.H. Lavington, ed. Amsterdam: North-Holland, 1980

Skinner, B. F. "Teaching Machines" Science, April, 1958: 128, 269-277.

Stevens, A.L. "The next generation of AI-based teaching systems." Machine- mediated Learning, 1, (1986): 313-326.

Kevin's Home Page