Halskov
Startside Op Buus Dines Frandsen Halskov Hansen Lundquist Siggaard Stjernfelt Teilgaard Forord Deltagere

 

A Guide to Metaphorical Design(1)

 

Kim Halskov Madsen(2)


Currently there is a growing interest in addressing the role of metaphors in the design process. By addressing metaphorical design from an empirical, a methodological and a theoretical perspective, this paper provides the basis for a systematic guide to metaphorical design. Rather than using a structural or formal mapping definition of metaphor, a pragmatic approach is applied, emphasizing use of metaphor as a particular kind of "seeing as" governed by previous situations and examples rather than by rules and fixed categories.


Introduction

Whereas the desktop metaphor previously was the metaphor, many different metaphors are increasingly becoming a prominent part of commercially available software (Carroll, Mack & Kellogg 1988:68). Most of the earlier research on computers and metaphors emphasized ease of learnability and ease of use, see for instance (Carroll & Thomas 1982) and (Carroll & Mack 1985). But currently there is a growing interest in addressing the role of metaphor in the design process, though only little research on systematic use of metaphor in this area exists (Carroll, Mack & Kellogg 1988) and (MacLean et al 1991).

This paper provides the basis for a systematic guide to metaphorical design by addressing metaphorical design from three different perspectives. First, after a brief review of previous research on metaphor, a collection of cases where metaphors consciously or unconsciously have been used are reviewed. These examples serve as the empirical basis for the generalizations about metaphorical design. Second, based on the cases, a set of guidelines for metaphorical design is offered. These guidelines provide a methodological basis for metaphorical design. Finally, based on the cases and the guidelines, a set of characteristics of metaphorical design is offered as a theoretical basis for reflecting on what metaphors are and how they work.

Previous Research on Metaphor

Within linguistics, a metaphor may be defined as a concept from one linguistic category (the source) used about a phenomenon normally referred to by concepts from a different linguistic category (the target), (Andersen & Madsen 1988). The major linguistic categories are normally considered to be animate, inanimate, human, animal and physical objects. So for example, the phrase "Bruce is a lion" is metaphorical since "Bruce" (a human) is referred to by the concept "lion" (an animal). But, referring to something in terms of something else which is different in some way, is often considered to be metaphorical as well (Ortony 1979) and (Lakoff & Johnson 1980).

Carroll, Mack and Kellogg (1988) have "identified three streams of current research on metaphor". Operational approaches focus on how and to what extent metaphors have a measurable effect on learning. Structural approaches, represented for instance by Gentner's structure-mapping theory (Gentner 1983), address metaphor by developing formal representations of relations between primitives (including relations among primitives) in the source domain and the target domain. Pragmatic approaches acknowledge that in the context of real-world situations, metaphor inevitably involves incompleteness and mismatches and that the power of metaphor may be attributed to such disparities between the source and the target domain. This paper follows the latter approach to metaphor. In particular, rather than using a structural or formal mapping definition of metaphor, a pragmatic approach is applied, emphasizing use of metaphor as a particular kind of "seeing as" governed by previous situations and examples rather than by rules and fixed categories (Lanzara 1983, Schön 1983).

A collection of cases

Metaphors have been used in many areas such as 1) design of a small command language, 2) a design task where users can define links between parts of different computer documents, 3) design of a bank Automated Teller Machine, 4) initial design of a production planning system, and 5) discussion of the impact of the use of computers on the service provided by libraries.

Case 1: A small command language

Although a metaphor may not be explicitly supported by the computer system, users often understand the system in metaphorical terms. An investigation of the language usage of employees at a Danish library reveals that they understand the structure of their computer system in terms of at least three different metaphors: the physical space metaphor, the conversation partner metaphor, and the organism metaphor (Andersen & Madsen 1988).

Investigation of language usage was based on tape recordings of conversations with three employees who were asked to describe how they used the various computer applications at the library.

Identification of the physical space metaphor was based on the criterion that a logical or functional part of the system was referred to in terms of concepts normally used about a physical space:

then I could go in and make back-up copies
when you want to get over from the circulation control system to the information retrieval system
that is we move a search between the bases
since the borrowers' numbers must lie in the circulation control system anyway

Identification of the conversation partner metaphor was based on the criterion that the computer is the sender or receiver in a sentence containing a speech act verb normally expressing linguistic acts of a human being:

well, then it asks me for the number of my borrower's card
and then I get an answer like that
it answered in a way I'm not used to seeing

Identification of the organism metaphor was based on the criterion that the computer is the subject of a sentence containing verbs like know, make and do.

it means that it knows itself what it shall do
I'm not sure whether it has shifted
and what it did now was that it dropped the connection

As an attempt to explicitly support the physical space metaphor, the tape recordings from the library formed the basis for designing a small command language consisting of commands like "go in", "go to IRSystem" and "go back" (Andersen & Madsen 1988).

Case 2: Links between documents

Erickson (1990 b) presents a design task where users can define links between parts of different computer documents so that when changes are made in one part the other parts are automatically changed. Three constraints are imposed: 1) links are directional, i.e. information can only be sent from the source to the destination, 2) links are one-to-many, and 3) the destination may not be instantly updated. Candidate metaphors include the 'TV broadcasting metaphor', the link metaphor, and the pointer metaphor.

Case 3: An Automated Teller Machine

MacLean et al. (1991) tell a story about how analogy or metaphor played an important role in the design of a bank Automated Teller Machine. In one case, the designer saw the ATM as an Express Checkout Counter at a supermarket, triggering the idea of having ATMs that could shift between an express mode with limited services offered and a full service mode. In another second case ".... the designers had personal experience of a bagel store which handled its lengthy queues by having an employee work along the queue, explaining the choices available and helping fill out their order on a form. The customers would hand over their forms when they reached the counter, enabling their requests to be processed more speedily" (Ibid. p. 169). The familiarity with the bagel store arrangement lead the designers to the innovative idea of having bank cards which the customers could pre-program while waiting in line.

Case 4: Production planning

A major Danish corporation's production planning was discussed in a workshop by using a metaphorical design approach. Production planners and systems engineers from the MIS department participated. The purpose of the workshop was not to develop a new production planning system but to illustrate how metaphor may stimulate new ideas about how computers could support production planning. After a brief introduction to metaphorical design, brainstorming about potential metaphorical views on production planning started. The participants in the workshop suggested many metaphors. Among other things, it was suggested to see production planning as:

a travel agency
a bakery
house cleaning
cooking
a soccer match
transportation
conversations
a power plant
cattle raising

 

Case 5: Service at libraries

In a technology assessment project at the Danish research libraries, the impact of the use of computers on the service provided by the libraries was one of the key issues (Etzerodt & Madsen 1988) and (Madsen 1989). In order to stimulate discussion about the impact of computers, the staff was challenged by three different metaphorical views of what a library is or could be. The three metaphors were "the warehouse", "the store" and "the meeting place". Each of the three metaphors provide a different account of what a library is, leading to different computer applications and eventually different kinds of service. Two of the metaphors, the warehouse metaphor and the meeting place metaphor, are presented below.

Seen as a warehouse, the library is first and foremost a place where books and other publications are supplied. Focus is on:

stock in trade (book stock)
orders (requisitions)
loss (lost books)
purchase of goods (accessions)
delivery of goods (lending of books)

Seen as a warehouse, the most import tasks of the staff are to find the goods quickly, to have precise knowledge about the stock and to avoid loss. The common computer systems for libraries, like information retrieval systems and circulation control systems, support the tasks reflected by a warehouse view of the libraries.

Seen as a meeting place the focus is on conversations between people about books and subjects related to books. The conversations could for instance be about

books available at the library, particularly new books
reviews of books
books useful in the context of specific meetings or cultural events
conversations take place between
the staff and the borrowers
the borrowers
the staff at the library
the staff from different libraries

Typical computer applications in this perspective could be:

a news catalogue announcing new books available at the library
a review catalogue for formal reviews as well as reviews by
borrowers
a calendar announcing events where specific books are relevant
an electronic bulletin board facilitating communication between borrowers.

The staff may take part in the conversations directly or they may act as consultants facilitating contact between the borrowers.

Guidelines

Guidelines derived from the cases are organized along the three main activities of metaphorical design: generating metaphors, evaluating metaphors and developing metaphors. Generating metaphors concerns getting ideas of potential metaphors to be used in the design process. Evaluating metaphors concerns choosing among the potential metaphors the one (or the ones) that in a productive way may fit the particular design task. Developing metaphors addresses using the metaphors chosen in the actual design task.

Generating

Listen to how users understand their computer systems(3). Making tape recordings of conversations with users about how they use their current computer systems, as done in the case of the small command language, is one way of capturing metaphors used by the users themselves. The analysis and identification of the metaphors were facilitated by establishing precise criteria for what constitutes an instance of a particular metaphor.

Build on already existing metaphors. Building on metaphors already used by the users, as in the case just mentioned, increases the chance that the metaphor will be assimilated by the users. Metaphors reflecting physical structure are often successful(4).

Use predecessor artifacts as metaphors. Using predecessor artifacts, like the physical card files at libraries, for instance, is a way for the user to benefit from his or her previous experience. Similarly, predecessor tools may be used as metaphors for the new computer-based system (Carroll, Mack & Kellogg 1988:77).

Note metaphors already implicit in the problem description (Erickson 1990b:69 f). For instance, links are directional, i.e. information can only be sent from the source to the destination. The link metaphor could give rise to more specific metaphors like "pipes" and "paths". Information could flow in the pipes or information could be carried along the paths.

Look for real world events exhibiting key aspects (ibid. p. 70). A key aspect of the link case is directionality, part of many real world things like rivers, TV broadcasting, newspaper publishing. Rivers end up at one point whereas TV broadcasts and newspapers end up in many places.

Evaluating

Choose a metaphor with a rich structure (Erickson 1990b:70ff). The "link" does not have much structure in the sense that it is rather general or abstract and does not suggest any of the specific aspects like the directionality or the one-to-manyness. In contrast, the "TV broadcasting" metaphor is more concrete and comes with a rich vocabulary like stations, channels, reruns, serial, receivers, VCRs, etc., all part of the everyday experience of most people.

Evaluate the applicability of the structure (ibid. p. 71). Consider whether the structure offered by the metaphor covers relevant aspects of the problem. For instance the "TV broadcasting" metaphor may be misleading since it implies that information is transmitted instantaneously.

Choose a metaphor suitable to the audience (ibid. p. 71f). For the metaphor to support the users' understanding of the system, the literal meaning needs to be familiar to the users. For instance, for a non- technical user community, a "pointer" metaphor in a computer science sense would be an inappropriate metaphor for links.

Choose metaphors with well understood literal meanings. Since one of the key elements in metaphorical design is to understand something in terms of something else, it is essential to know the source domain at a certain level of detail. For that reason "a travel agency", "a bakery", "a power plant" and "cattle raising" were eliminated as metaphor candidates in the production planning case.

Choose metaphors with a conceptual distance between the source and metaphorical meaning. The power of metaphorical design is to see things in a new way, which obviously is easier if they are seen as something fairly different. For that reason, the conversation metaphors were eliminated in the production planning case. It could actually be argued that the conversation view is not a metaphor at all.

Have at least one bridging concept. Since the literal meaning of the metaphor and the thing it is used as a new term for need to be significantly different, a concept bridging the target domain and the source domain is needed. For instance, the idea "that you always have to clean up the mess made by the others" bridged house cleaning and production planning.

Do not necessarily explicitly incorporate the metaphor in the final design. In the case of the bagel store, success was due to the designers' specific knowledge about a particular bagel store, unfamiliar to most other people.

Developing

Elaborate the triggering concept(5). The triggering concept may either be a key concept in the target domain, as "service" is in the case of the warehouse metaphor, or the triggering concept may be a key concept in the source domain, as "to meet" in the case of the meeting place metaphor. In the case of the meeting place metaphors "to meet" and "to have conversations" are elaborated to cover activities normally not understood in those terms(6).

Look for new meanings for the concept(7). Metaphors suggest new meanings for existing concepts. For instance, "a meeting place" does not need to be a physical location but could be a virtual location like an electronic bulletin board where people do not physically meet.

Restructure the perception of reality. Be aware of new features and relations, regroup, reorder and rename those relations. The meeting place metaphor brings to awareness the relation between the borrowers. By renaming(8), hidden or forgotten features are highlighted or created. By labelling the staff "consultants", the role as someone who advises or assists others is emphasized.

Elaborate assumptions. Make explicit what the metaphor hides and what it highlights. The warehouse metaphor hides the borrowers and highlights the books, whereas the meeting place metaphor highlights the borrowers rather than the books.

Tell the metaphor's story. Talk about the target domain as if it was the source domain. When telling the warehouse story above, the library concepts have been put in brackets, and when telling the meeting place story the concepts from the source and target domain are intermingled.

Identify the unused part of the metaphor. Look for aspects, features and properties in the source domain and consider how they may play a role in the target domain. The better the source domain is understood the better the target domain is understood.

Generate conflicting accounts. By generating conflicting accounts based on different metaphors reflection and critical awareness are stimulated. The library metaphors reflect different views of what the purpose of a library is.

Characteristics

The characteristics and the nature of the role of metaphors in design are not generally associated with a single one of the activities: generating, evaluating and developing. By considering all of them, we can make several theoretical observation about the role of metaphor in design.

Physical structure plays an important role. Several of the cases reflect the importance of physical structure. Holt has pointed out the importance of the physical structure of the workplace. Each area has its dedicated function, for instance the storage place, the assembly stations, the sales office. In the non-electronic environment, physical proximity is a prerequisite for doing tasks. For instance the person, the tools and the materials have to be near each other to do a task (Holt 1986).

Metaphor is an inherent part of everyday language. Even when a metaphor is not explicitly represented in the exiting interface, as in the case of the analysis of the conversations at the library, users understand the structure as well as the functionality of the system metaphorically. In some cases the metaphors are so deeply rooted in our language that we hardly notice them.

Metaphors often originate from everyday experience. The physical space, conversations, and interactions with other people are all part of our everyday life. As observed by Lakoff and Johnson, the new and unfamiliar is understood in terms of the old and familiar, (Lakoff & Johnson 1980). As pointed out by Mountford, among others, metaphors ".... can anchor users' understanding of the computer to something which they are already familiar with" and "Much of the ease of use of the Macintosh interface is attributable to the correspondence between appearance, uses, behaviours of such interface objects as documents and folders and their real world counterparts, and user difficulties are often attributable to differences between them" (Mountford 1990:25).

Abstract concepts are understood in terms of concrete and familiar things (Lakoff & Johnson 1980). Many of the "link" metaphors suggested above provide an opportunity to understand an abstract and unfamiliar feature in terms of more familiar and concrete things such as rivers and TV broadcasting.

Metaphors provides detailed and specific design options. The supermarket checkout metaphor generated the option of having a "fast cash" light above the ATM.

Metaphors may provide the basis for justifying design decisions. For instance, by pointing to the success of separating receiving services and placing requests at the bagel store, a similar separation could be justified for the bank.

A metaphor provides the user with a model of the system. The metaphor provides an expectation about what can be done and how it works.

Seeing something as something else. By definition, metaphorical design involves seeing something as something else. Seeing as is governed by previous situations, examples, and pictures rather than by rules and fixed categories (Schön 1979). As observed by Schön, rather than looking for standard design categories, designers pay more attention to the uniqueness of a situation than is generally acknowledged. When seeing the library as a meeting place, we are as interested in the specifics of this particular kind of meeting places as in the general characteristics of meeting places. By seeing something as something else we benefit from previous experience(9).

Provide a novel view of reality(10). Using metaphors provides an opportunity to see new things and to give new explanations, and consequently to get ideas for new computer systems. For instance, the meeting place metaphor directs attention towards the interplay between and among the staff and the borrowers. This shift of attention leads to different explanations as to how service may be improved by using computers at the library.

Provide a shift in focus of attention. Often what was otherwise in the background comes into the foreground and vice versa. The warehouse metaphor brings the books to the foreground, and the meeting place metaphor brings people at the library to the foreground.

Problem setting(11). The problem to be solved by computerization is not known beforehand and its identification is taken as an explicit issue to be considered. The meeting place metaphor points out several problems, including lack of contact among the borrowers, and suggests several different candidates for computerization, including an electronic bulletin board facilitating communication among borrowers.

Discussion and conclusion

Alan Kay (1990) advocates that "metaphor is a poor metaphor for what is needed" and calls for illusions or magic. For instance, his point is that we do not need computer screens that are as hard to erase or change as paper is, and we would like to have folders that are easier to find things in than physical folders. And challenging the HyperCard metaphor he points out that a specific card can only be in one stack, you can't have a stack in a card, and variables (though they are actually also called "containers") can not contain stacks, cards, and buttons etc., etc.

Kay's criticism is apparently based on the "formal mapping between source and target" approach to metaphors. Whereas if we take the "seeing as" definition of a metaphor, a relation between objects in the source domain does not necessarily hold in the target domain. For instance, if "cards do not contain stacks" in the source domain it does not necessarily imply that "cards do not contain stacks" in the target domain.

The issue that Kay implicitly addresses is the dilemma between tradition and transcendence(12). On the one hand, when designing computer system we strive for systems resembling the previous work environment (for instance having folders on the computer desktop) but, on the other hand, we would also like to benefit from the power of the technology and provide opportunities not available in the current work environment (for instance being able to move folders in and out of each others independently of their size).

As indicated in the introduction and further documented by this paper, there is currently a growing interest in addressing the role of metaphors in the design process, though only little research on systematic use of metaphors in this area exists. Documentation of the benefits or the effectiveness of using metaphors in design is either stated in rather general terms (Lanzara 1983) or based on isolated, thought very convincing, cases like in (Schön 1979) or like those reviewed in this paper. To conduct a more systematic qualitative study on the effectiveness of using metaphors in design seems an obvious candidate for future research.

To conclude, the purpose of this paper on metaphorical design has been to create a greater awareness of the use of metaphors, how they can be used and how they work.

Acknowledgements

This paper has been improved by comments from John Carroll, Arne Kjær, and the referees. Language revision in Denmark by Marianne Johansen.

References

Andersen, P. A. (1990) A Theory of Computer Semiotics. Cambridge: Cambridge University Press.

Andersen, P. B. & K. H: Madsen (1988) 'Design and Professional Languages', In: P. B. Andersen & T. Bratteteig Computer and Language at Work. Oslo Norway: Department of Informatics: 157-196.

Carroll, J. & J. Thomas (1982) 'Metaphors and the Cognitive Representation of Computing Systems', IEEE Transactions on Systems, Man, and Cybernetics v. 12, 2 : 107-116.

Carroll, J. & R. Mark (1985) 'Metaphors, computing systems and actice learning', International Journal of Man-Machine Studies v. 22: 39-57.

Carroll, J., R. Mack & W. Kellogg (1988) 'Interface Metaphors and User Interface Design', In: Hellander Handbook of Human Computer Interaction. Amsterdam: Elsevier Science Publishers: 67-85.

Ehn, P. (1989) Work-Oriented Design of Computer Artifacts. Hillsdale: Elsevier Science Publishers.

Erickson, T. (1990) 'Interface and the Evolution of Pidgins: Creative Design for the analytically Inclined', In: B. Laurel The Art of Human-Computer Interface Design. Reading Massachusetts: Addison-Wesley Publishing Company: 11-16.

Erickson; T. (1990) 'Working with Interface Metaphors', In: B. Laurel The Art of Human-Computer Interface Design. Reading Massachusetts: Addison-Wesley Publishing Company: 65-73.

Etzerodt, P. & K. H. Madsen (1988) 'Information Systems Assessment as a Learning Process', In: N. Bjørn-Andersen & G. B. Davis Information Systems Assessment: Issues and Challenges. Amsterdam: North_Holland Publishing Company: 285-301.

Frey, N. (1966) Anatomy of criticism: Four Essays. New York: Ateneum.

Gentner, D. (1983) 'Structure mapping: A Theoretical framework for analogy', Cognitive Science 7: 155-170.

Holt, A. (1986) Primitive Man in the Electronic Work Environment. Milan, Spain: Presented at The Conference on Electronic Work Environment.

Kay, A. (1990) 'User Interface: A personal View', In: B. Laurel The Art of Human-Computer Interface Design. Reading Massachusetts: Addison-Wesley Publishing Company: 191-207.

Lakoff, George & Mark Johnson (1980) Metaphors We Live By. Chicago: The University of Chicago Press.

Lanzara, G. F. (1983) 'The Design Process: Frames, Metaphors and Games', In: U. Briefs & C. Ciborra & L. Schneider Systems Design For, With and By the User. Amsterdam: North-Holland Publishing Company: 29-40.

MacLean, A., V. Bellotti, R. Young & T. Moran (1991) 'Reaching through Analogy: A Design Rationale Perspective on Roles of Analogy', In: S. Robertson, G. Olson & J. Olson Reaching through Technology. New Orleans: Addison-Wesley Publishing Company: 167-172.

Madsen, Kim Halskov (1989) 'Breakthrough by Breakdown: Metaphors and structured Domains', In: K. Klein & K. Kumar System Development for Human Progress. North-Holland Publishing Company: 41-53.

Mountford, S. (1990) 'Tools and Techniques for Creative Design', In: B. Laurel The Art of Human-Computer Interface Design. Reading Massachusetts: Addison-Wesley Publishing Company: 17-30.

Ortony, Andrew (ed.) (1979) Metaphor and Thought. Cambridge: Cambridge University Press.

Richards, I. A. (1925) Principles of Literary Criticism. New York N. Y.: Harcourt, Brace and World.

Schön, D. (1979) 'Generative Metaphor: A Perspective on Problem-setting in Social Policy', In: A. Ortony Metaphor and Thought. Cambridge: Cambridge University Press: 254-283.

Schön, D. A. (1983) The Reflective Practitioner. New York: Basic Books.


1. A version of this material originally appeared in Communications of the ACM. Volume 37, number 12 (December 1994), 57-62. Tilbage

2. Kim Halskov Madsen is associate professor at The Department of Information and Media Science, Aarhus University. Tilbage

3. A general introduction to work language analysis as the basis for design may be found in (Andersen 1990). Tilbage

4. For a discussion of the usefulness of physical space metaphors in the context of ease of learnability see (Carroll & Thomas 1982:110). Tilbage

5. This is similar to what Erickson (1990 a) calls Design by Symmetry, which works by first establishing a symmetry between concepts from two different domains and then extending the symmetry further. Tilbage

6. Related to 'Magnify-Add to the Object', a general technique for creating new ideas (Mountford 1990). Tilbage

7. Adapted from "New Uses for the Object" and "Modify the Object [or the meaning of the term] for a New Purpose", (Mountford 1990:27). Tilbage

8. Using words in a more general meaning is a typical way of developing metaphorical ideas (Erickson 1990a:15). Similarly, as a general technique for creating new ideas Mountford (1990:27) suggests "Modify the Object [or the meaning of the term] for a New Purpose". Tilbage

9. See also Carroll &Thomas (1982:107). Tilbage

10. Schön (1979:259) uses the term 'generative metaphors' to characterize metaphors 'generating new perceptions, explanations, and inventions'. Tilbage

11. Referring to Frey (1966:123) and Richards (1925:240), Carroll et al (1988:74) note that in the traditional literary view metaphors serve to raise questions that lie beyond the text. Tilbage

12. For an elaborate discussion of the importance of tradition and transcendence in design see Ehn (1989). Tilbage