Papers CHI 99 15.-20 MAY 1999

- Designing Multimedia for Learning: Narrative Guidance and Narrative Construction

Lydia Plowman Rosemary Luckin Scottish Council for Research in Education School of Cognitive & Computing Scieaces

15 St. John Street University of Sussex Edinburgh EH8 8JR UK Brighton BNl 9PH UK

-l-44 (0)131557 2944 +44 (0)1273 678647 Lydia.Plowman@ scre.ac.uk rosel@cogs.susx.ac.uk

Diana Laurillard, Matthew Stratfold, Josie Taylor’

31(

ABSTRACT Narrative is fundamental to the ways we make sense of texts of all kinds because it provides structure and coherence, but it is difficult to see how this works in the context of multimedia interactive learning environments (MILES). We tested our hypotheses about the form and function of narrative in MILES by developing three versions of material on CD-ROM which had different narrative structures and analysed the impact of the different versions on learner behaviour. We present a theoretical framework in which we explain the concepts of narrative guidance and narrative construction and their application to the design of MILES.

Keywords Narrative, multimedia design, education

NEW NARRATIVES AND LEARNING Narrative is not simply an aesthetic issue. It is fundamentally linked to cognition and understanding, so an understanding of how it works is particularly relevant to the design of multimedia for learning. The introduction of hypertexts has seen a resurgence of interest in the form and function of narrative because the authorial process and how it is understood by the reader or user accentuates the roles which are frequently left unstated in other media [ 1, 9, 19, 211. There has, however, been less interest in the role of narrative in non-literary texts and, particularly, of educational media. Our aim is to develop an approach which emphasises the importance of structure but recognises that narrative cannot be studied in isolation from the dynamic processes of interaction if it is to be applied to multimedia interactive learning environments (MILES). We discuss the relationship between narrative and usability and how MILES can provide the narrative guidance necessary to facilitate the user’s construction of narrative.

There are precedents for suggesting that books, films, drama and other narrative media can inform the design and usability of multimedia [5, 10, 151 but we focus on how the form and function of narrative facilitates or impedes learning, heeding the ways in which prior exposure to other narrative forms shapes our understanding but also examining how it works in a specifically interactive medium. Permission to make digital or hard topics of all or part of this work for personal or classroom USC is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the tirst page. To copy otherwise, to republish, to post on servers or to redistribute lo lists. requires prior specific permission and/or a fee. CHI ‘95, Pittsburgh PA USA Copyright ACM 1999 0-201-48559-1/99/05...$5.00

Because nafiative can be suspended or altered at various decision points, or foci of interactivity, the rearra$kgement of discrete elements generates new texts and new meanings [24]. Narrative structures as traditionally understood are not therefore appropriate for interactive multimedia but an Aristotelian concept of narrative (at its simplest, the concept that texts should have a beginning, a middle and an end) is locked into western European thought and shapes our expectations. Students can discern narrative more easily when using textbooks or video because these media generally conform to generic narrative expectations. There is not yet this heaitage to draw on for MILES and so students often have :problems engaging with materials and making sense of them [25].

The two main forms of MILES observed in use in the early phase of this study and previously [ 12,231 are those presented in an encydopaedia format, designed to support both Gree exploration and specific information enquiries, and those which use a more guided and focussed tutorial approach. Both formats have structures, the issue is the extent to which they are perceptible to users. Although encyclopaedia-style software is more commonly supposed to be lacking in structure, at least at the highest 1,svel it is recognisably an encyclopaedia, even if the specifics of navigation are not clear. Tutorial-style multimedia software does not have a comparable provenance: because it is based on live interaction with a teacher, so although it may seem to lend itself to a narrative structure, the lack of precedents makes design decisions difficult, especially given the lack of empirical work in this area. Either way, comprehension problems can be exacerbated for users when compared to books, videos or live tutors, mainly because the procedures and operations involved in using them get in the way of understanding. Where tealchers are not on hand to provide guidance, it is even more important that it is provided by the software.

NARRATIVE AND COGNITION Narrative is typically thought of as being synonymous with ‘story’, but whereas ‘story’ implies both form and content our observations suggest that, for MILES, it is generally the structure rather than the content which is problematic. Narrative can be seen as a macro’-structure which creates global coherence, contributes to local coherence and aids recall through its network of causal

’ All at Open University, Milton Keynes, MK7 6AA, UK

CHI 99 15-20 MAY 1999 Papers

links and signposting. The structure provides a linear dynamic which can accommodate diversions and tangents and allows learners to maintain their plans and goals. It has both cognitive and affective impact, performing an essential organising function for the learner by shaping the creation of meaning from texts of all kinds.

Whether we have a predisposition to finding and creating narrative [3, 171 or whether we are enculturated into it, the role of narrative is central to our cognition from earliest childhood [6, 201. It helps us think, remember, communicate, and make sense of ourselves and the world. More specifically, previous research [2, 22, 283 suggests that texts which are unfamiliarly structured because they do not conform to mental models of narrative make excessive demands on our cognitive processes. Learners constantly adjust their understanding in accordance with their exposure to conventional narratives, making the construction of narrative a central cognitive goal.

Learners favour clearly structured and navigable texts so MILES generally produce cognitive costs in terms of narrative construction. The very nature of a medium which brings with it the benefits of interactivity and a multimedia format also results in a fragmentation of the narrative which is normally present in linear media such as books: the narrative can be suspended and altered and may thwart or confuse our expectations, producing a multiplicity of pathways and disruption of the flow of the user’s experience. The freedom to explore the material and choose its pace and sequence is liberating but it can also be confusing because it results in frequent, sometimes gratuitous, interactions with the computer. This can mean that the interactions which are integral to the experience of using MILES impede learning. In such contexts narrative not only exists in the software with which students interact but must take account of the interactants, what they bring to its interpretation, and the processes of interaction. We describe the two main interdependent processes as narrative guidance and narrative constrzztion.

Background to the study The focus of the research reported here is how students make sense of their learning with multimedia by constructing their own narratives in conjunction with the narrative guidance presented by the text. Previous observations [26] of school use of commercially available CD-ROMs showed that group discussion was associated with operations and procedures rather than the educational content, meaning that students were unable to construct a sustained narrative because their learning was so fragmented. One of the problems in designing MILES is the tension between providing integral learning activities and providing information and resources for externally provided tasks. This can manifest itself as a series of units with varying degrees of relationship to each other and to the overall structure. We have characterised this in terms of the macro-narrative of the global structure and the micro-narratives of the lower-level task units.

Some software is designed with an explicit narrative which operates at a macro-level to give coherence to the

content. An example of this is Clkmentine, a CD-ROM designed to teach French to poorly motivated teenagers. The macro-narrative involves Clementine having to raise a million francs to save the Bengali tigers and all the sub- tasks, such as designing posters, arranging coverage on the local radio station, and Clementine’s appearance on a TV charity quiz show, relate to this. The structure is relatively linear because this was seen as more appropriate for the guided progression needed for language skills.

But an explicit macro-narrative does not necessarily lead to a coherent learning experience. School Disco had a strong narrative of organising a school disco but the tasks involved (such as calculating cumulative fmquency) did not always bear a clear relationship to the narrative, or were so complex and time-consuming that it became forgotten. In such an example there is a tension between the macro-level of the narrative and the lower-level task units. Students need to be able to relate the constituent units to the high level structure but also perceive the relationship between sub-units.

NARRATIVE GUIDANCE AND NARRATIVE CONSTRUCTION The design elements presented by the software constitute narrative guidance and can be a combination of features specific to interactive media, such as the need for clear navigational procedures, with features associated with traditional media, such as recognisable narrative and a clear relationship between tasks and the macro-narrative. We have observed ways in which teachers compensate for its absence by providing their own narrative guidance. They mediate its use for the classroom context by constraining the range of responses, offering interpretative possibilities and elaborating the task. An example of this is a task in which students aged 11 had to use Encarta to find information relevant to a forthcoming trip to a Roman site. They were going to act out, in costume, a play they had written based on a newspaper account of the discovery of a skeleton at the site; they were able to locate and use information more purposefully than others who did not have this guidance.

This example of the teacher providing narrative guidance is unusual because it involves a lot of preparation and familiarity with the materials. More often, students use MILES in a context where they do not enjoy such committed teacher input and the narrative guidance needs to be embedded in the software design.

Narrative construction is the active process of meaning- making, stimulated by the text and the environment, combined with the vast reservoir of knowledge that each person brings with them to the experience. It is a process of both discerning and imposing a structure on the materials, making links and connections in a personally meaningful way.

Although this active process of meaning-making is cognitive, it cannot be divorced from cultural context, whether this is the classroom or exposure to other media. Narrative is a way of connecting what goes on ‘in the head’ (internal representations in fig.1) with what goes on

311

Papers CHI 99 15-20 MAY 1999 --

in the world (external representations in fig.l), hence our shift in emphasis from the text itself to the process of constructing narrative. Prior knowledge may prove to be counter-productive if, for example, learners’ experiences of paper-based texts lead them to assume they can interact with multimedia in ways which do not translate from one medium to another or if they expect the strong narrative typical of computer games. Or prior knowledge may be beneficial if learners are able to make meaningful links to other texts and sources of knowledge.

teacher support

software support

cons true tion

Figure 1: The cycle of narrative guidance and narrative construction

Although we have separated them for the purposes of explication, narrative guidance and narrative construction are not discrete processes but, in a productive learning experience, a dynamic cycle. The construction of narrative is a collaboration between designers, teachers and learners.

The learner’s inclination to supply narrative continuity, to bridge the gaps at the foci of interactivity, is based on their own input in addition to what is presented by the software and their co-learners. The relationship between the designer and learner is symbiotic, each dependent on the other, and not so different from Steiner’s description of the ‘full act’ of reading books as being one of ‘dynamic reciprocity’ [27]. In this sense, there is little difference from books and other media with which narrative is more closely associated. The notion of an active reader is nothing new: Sterne played with such conventions in his eighteenth century novel Tristram Shandy and, more recently, reader response theory [8] has promoted an understanding of the ways in which author and reader jointly construct narrative. In an educational context, teachers are also included in this process.

THE DESIGN OF GALAPAGOS We have developed three versions of Galapagos as a research tool, based on extended observation of students using commercially available CD-ROMs in schools. We hypothesised that students need some form of presented narrative guidance from the software because the narrative in MILES is not easily discernible and there may be multiple narratives depending on the route taken. Students also require support for narrative construction as this increases understanding of the material. The purpose of the manipulations was to see to what extent different

forms of narrative guidance affect learner behaviour, including the ease with which they accomplish tbe task and how well they can recall their learning, and to establish implications for the design of multimedia for learning.

Galapagos is a CD-ROM in which we have manipulated the type of narrative guidance in each version and prcovided a model answer, script of the audio track, notepad and easily accessible statement of the task to support learners in their narrative construction. Galapagos uses Darwin’s experiences in the Galapagos Islands to teach the principles of natural selection and aims to stimulate users to think about how wildlife arrived on the newly formed volcanic islands given they are so distant from the nearest land mass. This in turn stimulates thinking about how variation in the islands’ bird population has arisen and all learners are given the task of explaining the variation in the wildlife observed by Darwin and constructing their answer to this task using the on-line notepad. All l.earnen have access to a model answer once some text has been saved in the notepad. All versions have the same content (eight sections, each of which deals with a different aspect of the Islands such as their formation, location, or the effects of trade winds), use the same video sequences and audio tracks, but have different structures.

The linear version was designed to present an easily identifiable narrative so we could explore the extent to which MILES should emulate more familiar narrative structures. The resource-based learning version was designed to reflect existing commercially available CD- ROMs we had observed in classroom use, and the guided discovery learning version was designed to c.ombine guidance of the type offered by classroom teachers with the benefits of an interactive medium.

The linear version The linear version is designed in such a way that :students are led through the eight sections of the CD-ROM in sequence and it is closest to a traditional narrative as presented in, say, educational television. There are no search mechanisms and users cannot change section, except by going backwards or forwards, until they have seen all eight sections once. The interface ,is active compared to the other versions: video segments start as soon as they are on the screen and learners are relatively passive. After the voice-over welcome users are invited to proceed by selecting the ‘Next’ button. The educational rationale for this design was that it would be more suitable for novice users as the scope for machine interaction was limited and it would encourage them to see enough resources to be able to answer the set question; left to their own devices, they may not see everything available. There is no explicit external support for narrative construction other than that provided by all versions; narrative guidance is provided by the familiarity of a traditional linear structure.

The resource-based learning (RBL) version The resource-based version offers no guidance thrlough the CD-ROM sections and leaves students to define their own

312

CHI 99 15-20 MAY 1999 Papers

route. Although on a smaller scale, its closest equivalent is a multimedia encyclopaedia. The eight sections of the RBL version can be accessed from a main menu or the simple word-search mechanism at the top level. When resources appear on screen, they are not activated until the user clicks on them. The style is exploratory and users are free to go where they want; the only constraint is that they cannot access the model answer before they have entered some notes in the notepad, as is true for all versions. The interface is static and more complex as learners have to make selections for control and navigation. This could mean a cluttered and intimidating screen, so we provided sufficient features to enable effective RBL but from a clean and attractive interface. The educational rationale for this version was to encourage independent learners who are able to search through the resources to research their answers and complete the task. In this design manipulation learners need to construct their own questions in order to decompose the task into manageable chunks. There is no additional support for narrative construction and no explicit narrative guidance.

The guided discovery learning version The guided discovery (GDL) version offers guidance in breaking down the task by providing paths through the material, questions to stimulate enquiries, and direction to specific resources. Of the three versions, this is closest to the support given by a teacher. It does not present the narrative as strongly as the linear version but provides more guidance than the RBL version. The eight sections can be accessed from one of three text guides which focus on aspects relevant to completing the task such as formation and location of the Islands, effects of tmde winds and ocean currents, and Darwin’s exploration of the

islands and the bird populations. The guides are not animated or intelligent agents although this is an area for future development. Users do not have the same Worn to move around as provided by the RBL menu but, as with the RBL version, users are expected to be active, although the range of choices they can make at any given point is constrained. After the voice-over welcome and introduction to the package, task, and toolbar buttons, attention is drawn to the ‘Guides’ button. The educational rationale for this version was to encourage and support learners in locating appropriate resources to complete the task. There is explicit support for narrative construction through the provision of this guidance.

The approach learners adopt to complete the given task is therefore determined partly by the version they use. In addition to the narrative guidance manipulations which are outside the learners’ control, other features such as the availability of a model answer and notepad editor scaffold the task [14] at different points in the narrative construction process. Use of these supports depends on whether students choose to adopt them because, as table 1 shows, they are available in each version.

DESIGN OF THE STUDY Four groups of three students, aged between 15 and 21 years, used each of the three versions of Galapagos. The 36 students were based in two different institutions and all were studying for a national examination in Biology. A session using Galapagos to complete the task to their own satisfaction typically took about 45 minutes. The study maintained ecological validity inasmuch as use of the CD-ROM was integrated into their studies at an appropriate point in curriculum delivery and provided an authentic task within the syllabus. It took place in their own environment and their teachers both introduced the

LINEAR

NARRATIVE GUIDANCE

recognisabte, linear strucme easy navigation limited interaction

SUPPORT FOR NARRATIVE

CONSTRUCTION

notebook model answer easily accessible statement of task script

STUDENT RESPONSES

navigation: access all sections, do not alter sequence task: do not take notes during 1st linear sequence, then become more like other groups; write chunks in notepad w/o reference to sections; all access model answer and subsequently make revisions

RESOURCE-BASED LEARNING (RBL)

GUIDED DISCOVERY LEARNING (GDL)

talk: talk low during 1st linear sequence, then both procedural & task increase slightly

no explicit narrative notebook navigation: do not access some sections; use guidance model answer menu or search for navigation

some implicit guidance in interface easily accessible task: start written response before completing

design (eg order of statement of task access to content; low level of text revisions; all access model answer - not all revise

items) script subsequently

talk: less talk relating to answer construction

three text guides offer notebook navigation: access all sections; use online guide routes through model answer as navigational tool material and stimulate enquiry easily accessible task: start response before completing access to

statement of task content; high level of text revision; text entered some implicit in small sections; frequent reference back to guidance in interface script content design talk: low level of talk in early stages; increases

after access to guides -.. _ . *. . ^ ..^^ a-. -I

‘IatJle I: summary ot design and use ot ditterent versions 01 ciakzpagos

313

Papers CHI 99 15-;!O MAY 1999 -

topic and debriefed them afterwards. However, each group was randomly assigned to one version of the CD- ROM and none of the students was aware of the different versions. It was not our intention to measure learning (as there are too many variables if studying a naturalistic environment is prioritised) but to identify ways in which learning can be supported.

Data collection Every session had two video-recorded sources: one recorded the group of learners at the computer to capture talk, movement, gesture and machine interaction; the other was the screen image, taken from the computer via a scan converter. The videotapes were mixed in an editing suite, transcribed, and used for very detailed analysis of learners’ talk and behaviour and their path through the material [ 131. The three-way software design manipulation has enabled us to analyse similar events across versions to see how design features, such as the narrative structure and provision of software support for learners, interact with and support students’ learning.

Following completion of the task, individual learners were given a tape-recorder and asked to record their responses to questions which ranged fi-om checking on accuracy of factual recall to asking them to recount the lesson for an absent friend. We analysed the notepad answers constructed during use of Galapagos, as they provide the most accessible means of insight into their emerging understanding, and a teacher evaluated how well they fulfilled the task requirement of explaining the variations in wildlife. We also collected information about each learner’s prior experience with computers, their predicted examination grades for Biology and the teacher’s assessment of their oral abilities.

STUDENTS USING GALAPAGOS Because the students used Galupagos in groups their experience is most clearly revealed through their talk. This is particularly useful for illuminating the processes of narrative construction, especially discussion involved in producing an answer to fulfil the task requirements. This data is supplemented by ’ navigational information from the video recordings. As the route taken is determined to some extent by the version used, this gives us information on narrative guidance, although neither source of data can be analysed in isolation.

For analytical purposes, each group was conceived of as a small case study and we did not attempt to find quantitatively defined relations between groups and the version used. Our focus of interest was in local relations between design features and behaviour for the various sources of data we had. Some of these were generated by the group (talk, notepad responses), others were generated by individuals (self-recorded responses to probes). The following discussion is based on very detailed analyses [ 131 of learners’ interactions with the software and with each other but has been simplified for illustrative purposes.

Interacting with Galapagos The linear version presents a high degree of n;zrative guidance and little opportunity for learners to decide their own narrative path so they have relatively little control. This is more apparent in the early part of a session as all eight sections were viewed in sequence, even though learners were able to move backwards and forwards. Once they had seen all eight sections Ilearners exercised more control but still retained a fairly linear path through the material.

For the RBL version, the reverse is true. There is very little narrative guidance offered and learners have to construct a narrative by making decisions about sequence, so there is a high degree of user control and heavy use of the menu to decide the route. In some cases, this meant that they did not access all available sections. The GDL version was designed to offer a balance between narrative guidance and support for narrative construction and this is reflected in a more even balance between user and system control: learners are able to determine sequence and their course of action but are offered guidance in doing so.

Given that the main purpose of Galapagos was for students to represent their knowledge in a Iwritten response to a question, one measure of difference between the versions is how many sections they access in composing their answer. There is a total of eight subject-related sections and a model answer available. Attempts to open the model answer before sufficient text has been entered into the notepad results in learners being advised that they can only access the, model answer when they have typed 50 words into the notepad. Analysis of video recordings shows how many sections of Galapagos were accessed from the point at which they started composing the answer, ie after first opening the notepad, and before accessing the model answer.

The mean for students using the linear version was only two sections (two groups did not revisit any sections at all) and they did not return to any sections subsequently. Students using this version did not take notes, and they did not interrupt the flow of the narrative or alter the order of the presentation, although their behaviour often changed when they reached the end of the sections with its reminder of the task. At this point they began to behave more like students using the other versions, going back over other material and focusing on the construction of their answer.

Most sections (seven) were accessed by the GD:L users because the guide had encouraged them to be interactive in their approach and to use the material to support their response. This was confiied by teacher analysis of the completed written responses. The RBL users accessed a mean of three sections. This could be because they were forced to be more interactive than the linear version users but did not receive the guidance of the GDL users.

Interacting with each other All group talk was transcribed and utterances were classified using NUD*IST but quantitative results across

314

CHI 99 15-20 MAY 1999 Papers

versions and different talk categories show no regular pattern. Task talk was highest for linear version users at 73% of the total number of utterances, but there was no difference between RBL (65%) and GDL (63%) users. For procedural talk about navigation and operations there was no difference across versions, all three being between 17% and 20% of the total number of utterances. We cannot read too much into this given the small number of groups, but the minimal difference in procedural talk could be attributed to the fact that all versions featured the same explicit learner support and this may have overridden problems inherent in different structures. The same features appeared in each version so as not to favour a particular narrative structure. By doing this, however, we lost some of the clear differences in the types of talk we had identified in earlier classroom observations of commercial CD-ROMs which did not offer such support. Analysing simple differences in the amounts of talk generated by different versions also risks eliding the role of individuals and, in some cases, individuals were a more important factor in group behaviour than the version to which they were exposed.

The points at which talk takes place and the effortfulness involved are more revealing. There was a low initial level of procedural talk for users of the linear version because there were fewer procedural and operational decisions (where to go, what to do next, how to do it) to be made. However, there was also a relatively low level of task talk (relating to completion of the task and relevant information) early in the session as the students simply sat back and watched the multimedia sequences unfold. During this time them was minimal student collaboration. The level of procedural talk increased once they had seen all eight sections because they had to assume some control, but this subsequently levelled off.

Although the level of procedural talk was similar for the resource-based (RBL) students they spent less time on answer construction and we have already noted that they referred back to fewer sections. This is reflected in an initial analysis of their self-recorded individual accounts after the event, with RBL students using less than half as many words (an average of 31) than either the GDL or linear users (76 and 77, respectively). Procedural talk tended to relate to using either the search or the menu facility as a central spine around which navigation to other sections of the CD-ROM was managed.

The guided discovery (GDL) users demonstrated low levels of talk in the early stages, but once the guides had been identified procedural talk decreased and task talk increased because the written guidance helped them to focus on the task and suggested routes. There was a tendency for the text to be entered in small sections, none longer than two short sentences, and to refer back frequently to other material, whereas some of the linear system users constructed large parts of their answer without referring back, an approach which was only observed with the linear version.

DISCUSSION We are not basing universal design guidelines on these findings because narrative is constructed both by learners as individuals and within groups, is culturally defined [4, 71, and there are important gender differences [ 181. Because the subject matter of Galapagos was evolution, there are also religious differences of interpretation. We have not, so far, analysed our findings from these perspectives although support for narrative construction should provide for individual differences to some extent.

The presence or absence of narrative affects whether learners interact productively with MILES and so affects usability. Teachers provide narrative guidance and support for narrative construction because they are able to assimilate digressions, repeat points and tie up threads in a highly interactive way which takes account of the individuality of learners, the social context, relevant artefacts and the environment. They are able to elicit knowledge from students and respond to what they say, to initiate, confirm, evaluate, reformulate, and give feedback. They provide what Mercer [ 161 calls the guided construction of knowledge. Narrative guidance and narrative construction are interdependent because the guidance provides the means for learners to construct their own narratives and the additional support for narrative construction leads to greater engagement with the narrative guidance (fig.1). MILES are not as sophisticated as teachers in this respect but our findings suggest that improved design and awareness of this interdependence would lead to benefits for learning.

Narrative guidance The linear version of Galapagos demonstrates a relatively traditional narrative structure but this is not necessarily a solution to the design of MILES. Although navigation is straightforward and learners access all parts of the content, this design encourages students to enter text in the notepad without reference to other sections, thus undermining a potentially beneficial feature of interactivity - being able to move easily between different sections.

A strong narrative can be motivational and contribute to the dynamic of the text but it can also detract from the real learning content. A previous example is Stowaway, a CD-ROM which provides a database of information about fighting ships in the 18th century. One observed problem was that the game of trying to spot the stowaway became a diversion from the information even though it provided a strong narrative link [23].

The presence or absence of the teacher is also a key variable. Students are more able to cope with multimedia encyclopaedias, for instance, if the teacher is available to offer guidance. Where teachers provide a strong narrative structure for the task, we observed fewer problems and noted that low-level interface and navigational problems can be more easily accommodated [23]. If more effort is involved in interacting with the text, mental resources available for comprehension am reduced. Low narrative guidance leads to problems with

315

Papers CHI 99 15-20 MAY 1999

narrative construction: even though the resources were very limited compared to a multimedia encyclopaedia the RBL group did not access all the sections and started answer construction before viewing all the sections. Not all groups made revisions after accessing the model answer. This suggests less involvement and the possibility that the procedural overhead meant that their attention was diverted from the task, as demonstrated by the minimal narratives offered to explain their experiences after the event.

It is not usually possible to design a software-embedded narrative structure similar to the linear or GDL versions for multimedia encyclopaedias but it is possible to provide the means for learners to develop narratives which encompass the task and the route required to complete it satisfactorily. The role of teachers in mediating the information can also be critical.

Narrative construction The design dilemma is that building in support for narrative construction requires more tools and a more complex interface. The three versions varied in the amount and type of guidance offered but they all offi similar help for narrative construction (model answer, notepad, script, statement of task) as we hoped that comparisons could be more easily drawn. The existence of a model answer motivated learners to start constructing an answer of their own and, once opened, the model answer itself prompted revisions.

Even though all versions have the same features supporting narrative construction, the take-up of these features varies according to the version used. For instance, users of the linear version do not access the notepad until they have seen the whole sequence of eight sections and do not use the back/forwards buttons, even though they are clearly accessible and their function is described in the spoken introduction. It is not that the linear version precludes them from using these features, but the interface design affects their responses so they do not actively engage with the task. The design also affects their interaction with each other, as a lower level of collaborative talk was observed for this group.

Although users of the linear version did not need to cb as much work at narrative construction they did not seem to use their cognitive resources to tackle the task more effectively than other groups. It is as if a strong, recognisable narrative structure encourages users to jump aboard and be transported to the final destination but they can’t easily get off. This narrative seduction works well in traditional linear media such as films but is less appropriate for MILES because it doesn’t maximise the potential benefits of an interactive medium or the interactivity offered by a good teacher and so is less suitable for learning.

The impetus to narrativise is so strong that in the RBL version cognitive resources are directed at constructing a route and finding structure rather than a content-rich narrative. The GDL version offers the same support for

narrative construction as the other versions, balanced by a presented narrative which offers guidance to scaffold users’ learning [ 141. It appears to level out some of the individual differences noted with other versions., giving support most consistently. Design features which we identified as effective in engaging learners, supporting meaning-making processes and eliciting e,ffortful activity included:

a clearly defined goal to encourage productive collaboration;

a balance between narrative guidance provided by software design and the provision of facilities which allow learners to determine their route and construct a meaningful narrative;

easy access to the task and notepad enabling learners to search for information relevant to their task and use it to construct a narrative of their emerging understanding;

a model answer which motivates learners both to start constructing an answer of their own and, once opened, prompts revisions.

CONCLUSIONS These are small-scale, exploratory studies and we: do not claim a simple causal relation between a version of Galapagos and effective learning. Rather, we have explored a dynamic relationship in which the learners’ experiences of design features affect both group and individual behaviour and this, in turn, affects their experience of design features. We have explored this relationship with particular reference to the cyclical processes of narrative guidance and narrative construction because they are central to learning.

The generation of narrative is an active process of meaning-making. It is a process through which we create a structure for interpreting and understanding connections and links between people, texts and events. But although we are skilled at imposing order in this way, this is not necessarily an efficient use of (cognitive resources when engaging with complex texts. Texts which appear to have a strong surface narrative. can be problematic if there is no support for narrative construction.

Even for more experienced users, if effort is involved in interacting with the material, mental resources available for comprehension and achievement of the learning goal are reduced. MILES do not yet accommodate d:iversions, interruptions and repetitions as human teachers do, so although students are usually able to follow i.ndividual units of content they are not given the help they need in getting the bigger picture of how units interrelate. Productive learning experiences benefit from the interplay between the processes of narrative guidance and narrative construction. Multimedia interactive learning environments need to be designed so learners are able to both find narrative coherence and generae it for themselves.

316

Cl-II 99 15-20 MAY 1999 Papers

ACKNOWLEDGEMENTS This research has been conducted as part of MEN0 (Multimedia, Education and Narrative Organisation), funded by the Economic and Social Research Council’s Cognitive Engineering Programme, grant no. Ll2725 1018. The Galapagos CD-ROM was developed by Matthew Stratfold. We are indebted to the schools, teachers and students who made this research possible and to the reviewers, for detailed and valuable comments.

REFERENCES Publications marked with an asterisk are available from http://meno.open.ac.uklmeno/

1.

2.

3.

4.

5.

6.

7.

8.

9.

Bolter, J.D. (1991). Writing Space: the Computer, Hypertext and the History of Writing. LEA, Hillsdale, NJ.

Bower, G. H., & Morrow, D. G. (1990). Mental models in narrative comprehension. Science, 247(4938), 44-48.

Bnmer, J. (1996). The Culture of Education. Harvatd University Press, Cambridge MA.

Chafe, WL (ed.) (1990). The Pear Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Production. Ablex, Nonvood NJ.

Clanton, C., F. Iannella & E. Young (1992). Film craft in user interface design. Tutorial Notes, CHI’92, Monterey, Ca. May 3-7, 1992.

Engel, S. (1995). The Stories Children Tell: Making Sense of the Narratives of Childhood. New York: WH Freeman.

Heath, S. (1983). Ways with Words: Language, Life and Work in Communities and Classrooms. CUP, Cambridge.

Iser, Wolfgang (1989). Prospecting. From Reader Response to Literary Anthropology. John Hopkins University Press, Baltimore.

Landow, G. (1992). Hypertext: the Convergence of Contemporary Critical Theory and Technology. Johns Hopkins University Press, Baltimore.

lO.Laurel, B. (1993). Computers a.s Theatre. Addison- Wesley, Reading, Mass.

11. Laurillard, D. (1998). Multimedia and the learner’s experience of narrative. Computers & Education 31 pp.229-242.*

12.Laurillard, D., L. Baric, P. Chambers, G. Easting, A. Kirkwood, L. Plowman, P. Russell & J. Taylor (1994). Teaching and Learning with Interactive Media. National Council for Educational Technology: Coventry. ISBN 1 85379 302 7.*

13. Luckin, R:, L. Plowman, L. Gjedde, D. Laurillard, M. StratfoId & J. Taylor (1998). An evaluator’s toolkit for tracking interactivity and learning. In M. Oliver, ed. Innovation in the Evaluation of Learning Technologies, pp.42-64. University of North London, London.*

14. Luckin, R., L. Plowman, D. Laurillard, M. Stratfold, J. Taylor (1998). Scaffolding learners’ constructions of narrative. In A. Bruckman et al. (eds.) Proceedings of Third International Conference on the Learning Sciences (ICLS-98), pp. 181-187. Georgia Tech, Atlanta, GA.*

15.McKendree, Jean & John Mateer (1991). Film techniques applied to the design and use of interfaces. In Proceedings of the IEEE Twenty-Fourth Annual Hawaii International Conference on System Sciences, (pp.32-41). IEEE, NY.

16.Mercer, N. (1995). The Guia’ed Construction of Knowledge. Multilingual Matters, Clevedon.

17. McNeil, L. (1996). Homo inventans: the evolution of narrativity. Language and Communication, 16(4), pp.331-360.

18. Millard, E. (1997). Differently Literate: Boys, Girls and the Schooling of Literacy. Falmer Press, London.

lg.Murray, J. (1997). Hamlet on the Holodeck The Future of Narrative in Cyberspace. Free Press, NY.

20. Nelson, K. (1996). Emergence of the storied mind. In Language in Cognitive Development, ed. K. Nelson. CUP, Cambridge.

21. O’Donnell, J. (1998). Avatars of the Word: From Papyrus to Cyberspace. Harvard University Press, Cambridge, MA.

22. Olson, DR (1990). Thinking about narrative. In Narrative Thought and Narrative Language, e&. Bruce B&ton & Anthony Pellegrini, LEA, Hillsdale, NJ.

23. Plowman, L. (1996). What’s the story? Narrative and the comprehension of educational interactive media. Proceedings of the 8th European Conference on Cognitive Ergonomics (ECCEB), Granada, Spain, eds. T. Green, J. Ctias, C. Warren, pp.167-172.”

24. Plowman, L. (1996). Narrative, linearity and interactivity: making sense of interactive multimedia. British Journal of Educational Technology, 27(2) pp.92-105.*

25. Plowman, L. (1994). The ‘Primitive Mode of Representation’ and the evolution of interactive multimedia: some design issues for group use. Journal of Educational Multimedia and Hypermedia 3 (3/4) pp.275293.”

26.Plowman, L. (1992). An ethnographic approach to analysing navigation and task structure in interactive multimedia. In People and Computers VII, 4s. A. Monk et al., pp. 271-287. CUP, Cambridge.*

27.Steiner, G. (1996). ‘The Uncommon Reader’. In No Passion Spent, Essays 1978-1996. Faber & Faber, London.

28.Thorndyke, P. (1977). Cognitive structures in comprehension and memory of narrative discourse. Cognitive Psychology, 9, pp.77- 110.

317