Full Paper

Brian Ferry, John Hedberg and Barry Harper

University of Wollongong

Teaching has been considered the act of transferring information from the teacher to the learner who was seen as empty vessel to be filled with knowledge. This view of learning was due to the popularity of behaviourist learning theories which focused on how the presentation of the information affected learning outcomes (Weinstein & Mayer, 1986). Therefore, it is not surprising that the art of teaching became the art of presenting information.

In recent decades cognitive scientists emphasized the study of how information is stored and processed in memory. They view learners as processors of information who use a variety of strategies to store and retrieve knowledge (Weinstein & McDonald, 1986). Thus the learner is a person who can engage in activities that will aid in the processing of information. Such mental activities help people to acquire, organize, and remember incoming knowledge more efficiently (Park, 1995).

Effective strategies designed to promote efficient and meaningful learning rely upon connecting prior knowledge to new concepts (Okebukola & Jegede, 1988) and research indicates that concept mapping is an example of such a strategy (Novak & Gowin, 1984; White & Gunstone, 1992). Concept maps are two or three dimensional spatial or graphic displays that make use of labeled nodes to represent concepts and lines or arcs to represent relationships between pairs of concepts. It has been suggested that their structure parallels the human cognitive structure, as they show how learners organise concepts (Novak & Gowin, 1984; Heimlich & Pittelman, 1986; Fisher et al, 1990; Harlen et al, 1990; Malvern, & Reed, 1990; Wandersee, 1990; Clarke, 1991; Margulies, 1991; Langfield-Smith, 1992; White & Gunstone, 1992; Tobin, Tippings, & Gallard, 1994). The advantages of this form of information presentation have been recognized for many years by established researchers (Ausubel, 1968; Novak & Gowin, 1984; West & Pines, 1985), and in recent years science educators have begun to report on the use concept mapping strategies with preservice teachers in order to see how they structure their subject matter knowledge (Lederman & Latz, 1995).

Six uses of concept maps were identified by White and Gunstone (1992) and these were: to explore understanding of a limited aspect of the topic; to check whether learners understand the purpose of instruction; to see whether learners can make links between concepts; to identify changes that learners make in relationships between concepts; to find out which concepts are regarded as key ones; and to promote learner discussion. The findings from this study support the contention that the planning of instruction is an additional use of concept maps, with the process likely to be more important that the final product.

Jegede, Alaiyemola, and Okebukola (1990) have maintained that "concept mapping serves as a tool to help learners organize their cognitive frameworks into more powerful integrated patterns" (p. 952), but computer-based concept mapping tools can do more than this and Beyerbach and Smith (1990) showed that concept mapping techniques enhance preservice teacher thinking about effective learning strategies. This view is also supported by Lederman and Latz (1995) who maintain that while some preservice teachers may posses the basic subject matter knowledge needed to effectively teach science, they find it difficult to organise their knowledge so that they can be effective instructors. The research of Lederman and Latz (1995) and Ferry (1996) support the view that the use of computer-based concept mapping tools assists preservice teachers to organise curriculum content into powerful integrated frameworks. Further, the process of concept map construction can stimulate a dialectic between curriculum content and pedagogical practice as often preservice teachers experience tension between "the curriculum" as the content to be covered during instruction, and the pedagogical process adopted to assist learners to understand the content. Thus this paper argues that the concept mapping process can be used by preservice teachers to actively develop the understandings that help them to link curriculum content with appropriate pedagogical practice.

When preservice teachers construct concept maps they identify and define important concepts or ideas and graphically represent interrelationships among concepts (Armbruster, 1979; Goetz & Armbruster, 1980; Holley & Dansereau, 1984; White & Gunstone, 1992). The resulting two-dimensional or three-dimensional map represents a spatial organization of their knowledge about curriculum content and pedagogy. As the links in this structure are labeled, and the concepts fully described, the map becomes of greater value to the learner. While this part of the process is crucial, it is not an easy task and White and Gunstone (1992) report that learners find this part to be a "most irksome... and would skip it if they could" (p. 18).

Holley and Dansereau (1984) recommend that learners employ five or six steps when they create concept or network maps:

1. select key concepts. This is a recognition process that activates relevant knowledge, and assists in topic identification;
1. write the key concepts;
1. make an attribute list of the key concepts;
1. relate key concepts in a spatial relationship;
1. rearrange spatial representations;
1. compare representation to the text.

Researchers have suggested that a period of direct instruction is necessary before learners can successfully employ this process (Harlen, 1992; White & Gunstone, 1992). Thus the instructor must model the process of concept map creation to students, and provide assistance that can be in the form of named links, structured hierarchies, chains, or clusters of concepts. The instructional steps that follow are recommended by White and Gunstone (1992):

1. begin with a simple topic, familiar to students so that it is easier for them to concentrate on the learning process. Ensure that a small number of terms are involved;
1. model the construction of a concept map to the class. This can be done with an overhead projector or computer with projection facilities;
1. encourage students to think of all possible links and to write down the nature of each link;
1. it is unlikely that students will produce good maps on their first attempt. Provide constructive criticism;
1. you may provide a suggested layout the first time, but it is important to remove these prompts from subsequent maps;
1. tell students that there is not a single correct answer to the task.

In the past, learners may have used a set of cards as concept labels and pieces of string for links. They arranged these on a large sheet of cardboard and attached them with tape. Finally, the relation between pairs of concepts was written next to each link. Computer technology can replace cardboard and string; concepts and links can be more easily manipulated on a computer screen, which, in turn, has made it easier to construct and revise concept maps.

The concept map tool used in this study is considered to be a cognitive tool. Cognitive tools have been defined by Jonassen (1991, p.2) as: "generalizable tools that can facilitate cognitive processing." They are both mental and computation devices that support, guide and extend the thinking processes of the users. Such tools are external to the learner and engage the learner in meaningful processing of information.

The concept mapping tool was created in a HyperCard™ programming environment because Apple Macintosh ™ computers were used across the campus and all preservice teachers had at least one course on the application of information technology to teaching and learning (fourteen two-hour tutorials).

The concept map displayed in figure 1 was constructed by Barbara during her first tutorial session with this tool. It is a screen capture of a particular stage of concept map construction. The functions of the nine buttons shown on the palette in the lower right-hand corner are described below:

• concept allows the user to type concept labels and to move these labels in two dimensional space;
• link allows the user to draw link arrows between concepts. Arrows indicate the direction of these links;
• note allows the user to attach explanatory notes to the links or concept labels. Therefore concepts and links can be explained. The concepts in figure 2 that have bold boxes around them have notes attached which can be displayed by clicking on the concept labels (two are displayed on screen and the others are hidden). The link lines that have labels N1 or N2 are displaying their attached notes;
• eraser allows the user to correct mistakes;
• help provides access to a simple help screen;
• print allows the user to print out the concept map and the notes;
• save allows the user to save the map to a disc;
• load allows the user to load a map from a disc.
• clear screen allows the user to clear the whole screen

The study was designed to answer the following research questions:

1. How do preservice teachers use computer-based concept mapping tools to organise their curriculum content knowledge?
1. What features of the computer-based concept mapping tools help preservice teachers to link effective instruction with curriculum content knowledge?

Sixty-nine preservice teachers volunteered to take part in this study. They were enrolled in their third year of a Bachelor of Education (Primary/Elementary Teaching) Degree at the University of Wollongong. The mean age was 24.4 (SD 5.8), and the age range was 19 years to 49 years. All subjects had prior instruction with the Apple Macintosh ™ computers and with HyperCard ™ presentations of learning materials. However, while students might have successfully completed an information technology subject, they may not always feel confident and competent using computers (Haywood & Norman, 1988).

The preservice teachers were given a period of direct instruction in a lecture theatre that contained computer projection facilities. The concept mapping tool was presented and the process of concept map construction modeled by the researcher. Tutorial sessions were conducted in a computing laboratory, and at each session, the researcher again used computer projection facilities to introduce and revise procedures. Each preservice teacher was allocated a computer, but they were free to collaborate with their peers as they constructed their maps.

After the first tutorial copies of completed maps were collected and analyzed. Feedback was provided on how their maps might be improved, and again demonstrated in the following lecture. This procedure then repeated for a second tutorial.

The data, in the form of concepts maps and interview audio-tapes, were gathered at three stages in the process: after the first tutorial session, after the second tutorial and after final tutorial sessions. At each stage, concept maps were collected for analysis and twelve volunteers were interviewed. The volunteers were interviewed three times: after their first tutorial session, after the second tutorial session, and when they completed their final map. Completed concept maps were examined by the researcher and then returned before interviews commenced. All interviews were audio-recorded and later transcribed for further analysis. This procedure allowed the researcher to gain an understanding of how the volunteer preservice teachers used the concept mapping software to create their concept maps and to understand how they revised their maps.

The final interview with each preservice teacher took place in the computer laboratory and began with an on-screen display of their completed concept map. They were asked to describe and demonstrate how they constructed their concept map. At various points the interviewer asked clarifying questions or re-stated their descriptions. If necessary, the preservice teachers could re-affirm their statements or elaborate further. This strategy helped the researcher to understand how the preservice teachers used the software to create their final concept maps. It also helped to identify the features of the software application that supported preservice teachers in linking effective instruction with curriculum content knowledge.

Twelve reflective journals written by the interviewed preservice teachers were also examined. The interview transcripts and written responses were coded and analyzed for trends.

Various methods of analysis of concept maps have been described by a series of researchers including: Beyerbach, 1988; Harlen et al., 1990; Jonassen & Reeves, 1996; Lloyd, 1990; Novak and Gowin, 1984; Shavelson, 1993; and White & Gunstone, 1992. A commonly adopted scoring procedure is that proposed by Novak (1984, pp. 36-37) and it consists of the following rules:

• any relationships that are valid score 1 mark each;
• every valid level of hierarchy scores 5 marks each;
• cross links if valid score 10 marks each. If the cross link is valid but does not illustrate a synthesis between sets of related concepts and propositions it only scores 2 points;
• examples score 1 mark each.

Twenty concept maps constructed during the first and last tutorial sessions were scored in this way and two experienced science teachers independently applied this criteria to score the same sample of twenty concept maps. Their scores agreed for 14 of the twenty maps. Whilst the difference between total scores for the other 6 maps was never more than 7 marks on any occasion, the procedure was still considered to be only fairly reliable and was not adopted. After this trial it was decided that reliance upon any such scoring system would be problematic in this context so other procedures were explored.

Shavelson (1993), and White & Gunstone (1992) have reported that some researchers use a criterion concept map for the material to be mapped. Then they divide student map scores (obtained by a procedure such as described previously) by the criterion map score to give a percentage score which can be used for comparisons. With this procedure it is possible that some students may score better than the criterion map and receive a percentage score that is more than 100%. Rye (1995) reported that a criterion map could be used with success when students studied the same topic (Global Atmospheric Change). In his study, the expert maps were used to identify central concepts that could be used with a technique known as Pathfinder Network Analysis. This approach restricts the topic of the map to one domain of knowledge but for this study preservice teachers were not restricted in their choice of topics. Therefore, such an approach was again not suitable in this study.

Extensive experience over a long period of time led White and Gunstone (1992) to state that "our personal preference is not to score concept maps" (p.39). Instead they have argued for a more general approach that examines the detail contained in the maps, the variety of relationships (particularly cross relationships) and links. A study by Shavelson (1993) reviewed the methods used to score concept maps and he concluded that "the variation in concept map assessments that we have described is enormous. Further, these assessments are unlikely to produce equivalent scores" (Shavelson, 1993, p.18).

As the total scores that can be obtained for a whole map are subject to marker variation, it was decided to analyze the main attributes of the concept maps (concepts, links, notes, and hierarchical levels). The process required the following categorization for each map. The links between and among concepts were allocated into one of the following categories: those with 2 links, those with 3 links, those with 4 links and those with 4 or more links. The frequency for each of these categories was then recorded. Concept notes were analyzed in 2 ways: by recording the mean of the word count for the concept notes and the mean of the teaching strategies mentioned in the notes. The links were also analyzed in two ways. First each map was examined for the presence of absence of hierarchical links; that is where a more inclusive concept (e.g. galaxy) is linked to a related but less inclusive concept (e.g. Milky Way). Second the link notes were examined to determine if the text clearly explained the relationship between the linked concepts. This data could then be used to compare maps and to make descriptions of general and specific changes. This method for organizing and analyzing the data is based upon the approach reported by Beyerbach (1988), who used frequencies and descriptions to compare the main attributes of concept maps.

The results are organized by the two research questions. Part 1 presents and discusses the data that helps to clarify our understanding of how the preservice teachers used computer-based concept mapping tools to organise their curriculum content knowledge. Part 2 presents and discusses the data that helps to identify features of the computer-based concept mapping tool that helped preservice teachers to link effective instruction with curriculum content knowledge.

It was observed that the construction of concept maps is a skill that initially requires careful instruction and this supports the claims of Novak & Gowin (1984) and White & Gunstone (1992). After the preservice teachers received instruction that modeled the process and completed a one-hour practice session in the computer laboratory, they needed little support with the process. This observation also supports the research reported by Jonassen (1996). Indeed many subjects reported that they felt comfortable with the process after one tutorial session and the only concept map construction tool that needed further demonstration was the link tool as some preservice teachers needed to practise the "click-and-drag" skills required to draw links (to draw a link users must hold the mouse button down "click" to start the link and then keep the mouse button down as they drag it across the screen to create a link line. Releasing the button ends the link).

The following journal entries support this interpretation:

During the initial tutorials, three independent observers (the director of the computing laboratory, a PhD student, and an instructor) each spent thirty minutes observing the preservice teachers as they constructed their initial maps. They reported that few subjects met with any problems when they used the software. Indeed they felt that the concept map tool quickly became transparent and this enabled users to focus on the cognitive processes involved in constructing the concept map.

Initially, most of the preservice teachers collaborated with a peer. This supports research by Ramsden (1992) which showed that supportive peers were beneficial for adult learning. However, three experienced computer users in the group preferred to work alone, and as one graduate in computer programming said: "I find that I can't play around with the program if I work with a partner."

Interview transcripts from the less experienced computer users revealed that initially they consulted with a peer who helped if they forgot how to use the software. Also it allowed them to discuss their ideas. By the second tutorial session, they were ready to work independently and the following quote from Michael (aged in his late 20's) describes his experience with his collaborator Julie (age 21):

It appears that Michael was looking for support when he began to use the computer and sought out a supportive peer. Also Michael realized that working with Julie stimulated him to think about his ideas. Furthermore, he realized that concept mapping could be applied to other subjects. In Michael's case, Julie acted as a peer tutor and provided the initial scaffolding needed before he worked alone.

The following comments from Glen describe his first attempt with the concept mapping tool:

In Glen's case the support of a peer was not needed so he and partner Joanne decided to work independently. Whilst he admitted to feeling a 'little nervous' at the start, he soon realized that the application was easy to use and that he was capable of using it. Although Glen did not need the support of a peer, as he was an experienced computer user; he was observed to help others in later tutorial sessions. For this sample, prior experience did affect the way that preservice teachers preferred to work and those that were experienced computer users (those who were experienced in using HyperCard) preferred to work alone.

One strength of HyperCard-based software is that it continually updates the currently displayed screen. If user actions or other events (such as a network malfunction) accidentally causes a computer to 'crash', the current concept map is saved and re-displayed when the application is restarted. It was found that an early demonstration of this feature tended to allay fears about accidental data losses. Thus when the inevitable occurred, most preservice teachers calmly waited for the system to re-boot.

The following journal entry by Ruth (age 22) sums up many of the comments made by the preservice teachers about the value of the instructional process.

Ruth's comments about the importance of teacher demonstration and practice in a non-threatening environment were echoed in nine other transcripts. Thus instructional strategy was an important factor. The instructional sequence was based upon Cambourne's model of instruction (Cambourne, 1988). His model suggests that the instructional sequence described in table 1 is effective for instruction in literacy.

Ruth's transcript also shows that while the software makes the physical task of construction easier, and provides an effective visual display, the learner still has to acquire and organise their knowledge. Further, it appears many preservice teachers did not realize how inadequate their understanding of the subject matter knowledge was until they began to create concept maps. Often they had superficial knowledge of a few key concepts, but had little or no in-depth understanding of related concepts and how they linked together. It is likely that this would have impacted upon their ability to effectively instruct young children.

The concepts maps created during the final tutorial session were compared to those created during the first session. A total of sixty-nine were analyzed. Table 2 displays the results of an analysis of these maps.

In most cases, concept maps created in the second tutorial contained more concepts, and 24 were organized in branches that contained more than 4 linked concepts. Only 8 of the first set of concept maps contained a branch with 4 or more linked concepts. Also the number of maps with clearly defined hierarchical links had increased and this may indicate that the preservice teachers had a better understanding of the relationships among the concepts in their map. Other improvements were: greater use of link notes to clearly explain links, more comprehensive descriptions in the concept notes and a more extensive discussion of teaching strategies in the concept notes. The descriptions of teaching strategies became more extensive and varied in later tutorial sessions. Further, interview transcripts and journal entries showed that most preservice teachers preferred to concentrate on constructing their concept map to a stage where they could understand and link their concepts into an integrated framework that they were confident with before they added teaching strategies to maps. Thus they developed their curriculum content knowledge to a point that was sufficient for them to feel confident before they added teaching strategies.

The concept map in figure 2 shows a screen capture of Barbara's map as she revised it during the second tutorial session. It shows that she described the planet Jupiter as "a large gas planet. It has many moons and a thin layer of rings... and a red spot". She goes on to say "the swirling clouds of Jupiter can be simulated". Concepts were clearly linked with labeled lines and the concept 'stars' (figure 2) indicate she understood that stars formed part of the Milky Way and were also present in globular clusters. This map also indicates more extensive notes about all aspects of the topic, and describes a variety of instructional strategies, including a visit to a local planetarium. This map was again revised and further teaching strategies added in subsequent tutorial sessions.

The following interview transcript is from an interview with Barbara who created the map in figure 2. She talked about her attempts to create a concept map about an astronomy topic.

Her interview transcript indicates that the software provided a visual display that could be easily altered. Like all of the others interviewed she revised and refined her concept map several times before she became satisfied with the final product: a more organized cognitive framework containing powerful integrated patterns with well-labeled links. As the revision process was easy to perform, it was not as "painful" as if she had used pen and paper. The ability to rapidly create, alter and manipulate concepts was mentioned by all teachers interviewed and in several journal entries.

Later in the interview, Barbara explained that:

This pattern of work was reflected in the structure of the maps collected during the three stages. Many of the maps collected at the end of the first tutorial demonstrated limited understanding of the topic and poor construction. For example, thirty-six of the original maps contained concepts that were limited to those presented in common textbooks, while thirty-three took the trouble to find more information. However, all maps, except one, had some notes attached to the concepts, which appears to indicate that individual concepts may have been understood. However, the links among concepts were not understood, as only thirty-one contained labeled links. Notes attached to the concepts included descriptions or definitions, and a few brief comments about teaching.

At the second tutorial session, most preservice teachers brought a revised concept map (the form varied from those constructed on paper with pen or pencil, to a computer printout, or a disk file) and support materials such as books and summaries. It appeared that the process activated secondary support strategies (Holley & Dansereau, 1984). Many made efforts to improve their subject matter and pedagogical knowledge, as well as concept map construction skills during the intervening week. Indeed, computing laboratory records showed that many preservice teachers used the computing laboratory during their spare time to improve their maps. The main improvements observed in the maps were: the use of more specific concepts, and the greater use of notes to describe concepts, instructional strategies and explain links between pairs of concepts.

The final set of maps showed that most preservice teachers had organized and linked concepts into integrated frameworks which were easy to understand. The supporting notes contained extensive teaching strategies.

While further investigation is needed in order to fully understand why the preservice teachers went to considerable lengths to improve their concept maps, Barbara's previous comments provide us with a clue as it suggests while she was constructing her map, she was unconsciously thinking about appropriate teaching strategies. Evidence from the books and summary diagrams brought to the tutorials by the preservice teachers showed not only were they trying to find out more about the topics, but they were also thinking about appropriate teaching strategies. Further, the resource librarian confirmed that the preservice teachers borrowed materials that related to pedagogy and to content knowledge in the weeks between tutorials. Thus the preservice teachers were thinking about structuring subject matter knowledge related to their chosen topic and about how they would teach it. They had used the concept mapping process to help them to link curriculum content with appropriate pedagogical practice.

Interview transcripts and journal entries provided four suggestions about how the preservice teachers could improve their concept maps. They suggested that they

• needed to improve their content knowledge,
• could practice constructing concept maps,
• needed to include more specific content matter knowledge, as they acknowledged that their maps were too broad and lacked specific details, and
• needed to reorganize their concepts maps in integrated networks which contained descriptions of a of instructional sequences and linked the curriculum content to the learner.

The benefits of the computer-based tool was described by Leanne (aged in her mid 30's). She said:

Thus Leanne viewed the concept mapping software as a cognitive tool and later in the interview she compared the computer tool to pen and paper:

She saw the tool as more efficient than pen and paper and the following comment indicates that she was also thinking about how she would use the tool to support children learning:

Leanne's comments underline the value of applying concept mapping to the classroom. In particular she sees the technique as a tool to display and record children's thoughts to guide learning. Her comments support the claims of Beyerbach and Smith (1990), and Lederman and Latz (1995) who suggest that concept mapping techniques could enhance preservice teacher thinking about effective teaching.

This study focused on preservice teachers use of concept mapping software to organise their curriculum content knowledge. The findings suggest that tool made it easy for most preservice teachers to construct and revise a concept map of their subject matter knowledge. While the physical part of the processes may have been easier than pen and paper, preservice teachers still had to apply their current knowledge when they constructed their map. The early attempts at creating concept maps were generally poor and this finding supports the research of White and Gunstone (1992). However the preservice teachers persisted and revised their maps many times. They reported that the revision process was less irksome than with pen and paper and this attribute appears to be one reason why they persisted with their maps.

The study suggests that because the concept mapping tool was easy to use, it quickly became transparent to the preservice teachers. This enabled them to focus on the cognitive processes involved in the construction of their concept maps. However, the use of the concept mapping tool requires careful instruction as the construction of concept maps remains a complex skill (Novak & Gowin, 1984; White & Gunstone, 1992). The preservice teachers involved in this study received careful instruction that was based upon a model developed by Cambourne (1988) before they used the software application. Too often it is assumed that just providing the hardware and the software is enough.

Interview transcripts and analyses of concept maps demonstrated that most preservice teachers needed to reach a stage where they were confident that the cognitive framework, presented in their maps, was comprehensible and teachable to their targeted class, before they were able to fully develop learning strategies. That is, first they needed to be confident that they understood the knowledge domain and the relationships among the concepts. Then it was easier to select and attach appropriate learning strategies to their map.

Data from the interview transcripts supported the research by Beyerbach and Smith (1990) and the findings of Lederman and Latz (1995) who have suggested that concept mapping techniques may be used as a tool to enhance preservice teacher thinking about effective teaching. This study suggests that thinking about effective teaching and learning strategies occurs at all stages and the formal linking of strategies to concepts was a culmination of an ongoing process that continued throughout the development of the concept map.

Preservice teachers believed that the computer generated display of linked concepts and the underlying notes made it easier for them to identify parts of their map that needed revision, or to identify deficiencies in their subject matter knowledge. When they were away from the computing laboratory, many preservice teachers made additional efforts to revise their maps and to improve their subject matter knowledge.

Previous research supported the contention that the use of computer-based concept mapping tools assisted preservice teachers to organise curriculum content into powerful integrated frameworks, but the evidence from this study suggests that the process of concept map construction can do more than this; it can stimulate preservice teacher thinking about the link between curriculum content and appropriate pedagogical practice. Preliminary data from follow-up studies of twenty-one members of this group of preservice teachers (now full-time teachers) suggests that once the process of concept map construction has been experienced for a sustained period of time (in this case 14 weeks) and applied to a school context, it continues to be used as a planning tool. However, most preservice teachers did not use a computer-based concept mapping process in schools; instead they went back to using pen and paper in schools because they did not have sustained access to computers.

This study should encourage educators to continue to experiment with similar cognitive tools as it is not claimed that the software used in this study is an "idea tool". Nevertheless it challenges instructional designers to consider incorporating metacognitive tools that support the process of concept map construction in quality computer-based instructional materials.

(c) Brian Ferry, John Hedberg and Barry Harper

The author(s) assign to ASCILITE and educational and non-profit institutions a non-exclusive licence to use this document for personal use and in courses of instruction provided that the article is used in full and this copyright statement is reproduced. The author(s) also grant a non-exclusive licence to ASCILITE to publish this document in full on the World Wide Web and on CD-ROM and in printed form with the ASCILITE 97 conference papers, and for the documents to be published on mirrors on the World Wide Web. Any other usage is prohibited without the express permission of the authors.