In a review of interface issues, an essay by Lindeman (1991) points out that "as the reality that can be manipulated by the user gets more complex, it becomes even more important that interface designers communicate a cogent model of that reality to the user." (p. 3) But when the arena of functions to be modeled involve the tasks and concepts of online information retrieval from the Internet, no single metaphor seems to suggest itself. Lindeman, in fact, states that the success of end user information retrieval has been limited because a good metaphor was never found. Now that the user must select retrieval options in a more hyperlinked and distributed environment, the complexity of that reality is particularly difficult to capture in a single metaphor.
The following study examines three literatures to gather a broad sample of the many Internet metaphors used by authors in library and information science (LISA/ISA databases), authors of more technical articles (Computer Database) and finally authors of more general articles (Magazine Index). The metaphors found (through the titles of articles exclusively) are grouped inductively into broad general categories and discussed using a variety of metaphor evaluation criteria.
A decade later, Wozny (1989) provides an excellent introduction to the application and empirical study surrounding the use of metaphors, analogies and conceptual models in computer systems. In addition to a clear distinction between metaphor and analogy, she provides a review of some of the novice training research that focused upon the use of metaphors. Broadening the case for the importance of metaphors, she points out that gaming software has been quick to couple the physical metaphor to the structure of a game largely because users report that the metaphor allows them to know how elements of the game work. That sense of understanding the underlying structure of the game provides a sense of knowing that makes the game fun to play--reported as a "what is where" knowledge. Further, Wozny explains that the "cut and paste" metaphor used for functional modeling of the movement of text from one application to another clearly demonstrates to the user "how it works" knowledge. The physical and organizational metaphor of "windows" is also shown to be an effective mechanism for describing a multi-tasking operating system. (p. 277) With the benefit of these common, well understood spatial metaphors, the user no longer has to exert effort to produce a mental structure of how the system will function and is freed of the particular exertion of memorizing the individual commands that were formerly required to produce the desired result.
Spatial metaphors are often chosen to organize information, but the newer hyperlinked environments challenge the user's ordinary sense of space. How far is it from one linked item to another on the Web? Vaananen (1993) reviews the difficulties that arise with a hierarchical metaphor in a hypermedia environment. Offering the metaphor of a "house" with "rooms" containing "nodes" with "documents" attached, the users' evaluation of the interface is provided, along with various "guides" created to further structure the "rooms." The author clearly sees the use of a good metaphor as essential to the transference from the author to the user the responsibility of structuring the information in a meaningful way, but adds that hypermedia tends to create a new, difficult-to-structure reality.
From studies in psycholinguistics, there are instructive efforts to provide evaluative norms for a wide variety of literary and nonliterary metaphors. One such study undertaken by Katz, Paivio, Marschark, and Clark (1988) built upon 10 psychological dimensions they developed for the perception of a metaphor derived from earlier work (see Katz, et.al., 1985). They tested the 10 dimensions by using a large sample of several hundred literary metaphors taken from various poetry texts and nonliterary metaphors conceived by the authors. Literary metaphors taken from poets like Byron, Hardy, and Thomas included phrases like "blushes of youth are roses spread by joy" or "an autumn storm is the funeral song of the dying year." Phrases like "hard work is a ladder" or "education is a lantern" represent typical nonliterary metaphors devised by the authors for testing.
The list of over 400 metaphors created was then rated by subjects using a Likert scale to measure the appropriateness of the 10 criteria for each metaphor--characteristics like comprehensibility, ease of interpretation, metaphor goodness, felt familiarity, semantic relatedness, and number of alternative interpretations. The authors provide the norms derived and identify a category of metaphor they call "frozen" which have been used so often that their metaphoricity has been forgotten. An interesting example of what the authors called a frozen metaphor was the word "orient" which was originally used to mean "facing East" but now is universally understood to mean either "facing in any direction" or "the Far East." The Internet may provide an opportunity to test or norm in a similar manner the various Internet metaphors currently in use. (Note: A similar effort was created by the author [1] of this paper using the metaphorical categories derived from this effort along with a 7-point Likert scale at http://www.amigobus.utexas.edu/~jhassay/metaphors.html/ but it has yet to receive many hits. It is hoped that there will be sufficient data for discussion before this effort is presented.)
The most remarkable aspect of metaphor is its potential variety. There is a possibility of using practically any semantic field as a metaphor for any other. In a highly cited effort, Lakoff and Johnson (1980) enumerate the wide variety of metaphors that they argue are fundamental to the way that speakers create meaning out of the everyday activities of life, down to the most mundane aspects of personal reality. Through the classification of many metaphorical speech examples, they provide a variety of metaphor types. For example, we understand argument as war, love as a journey, life as a game of chance, or theories as buildings to name only a few. Lakoff and Johnson represent the study of metaphor as a conceptual mechanism and argue that these conventional conceptual metaphors belong to the common knowledge of all language users.
So how do we judge the usefulness of a metaphor as a sensible surrogate for activities on the World Wide Web? There can be an almost limitless variety of them and some provide more insight or sense making possibilities than others. From the human computer interaction literature, two assessment checklists for the evaluation of a metaphor in the realm of modeling computer systems are found in Madsen (1994) and Butler (1986). These provided some guidance for sorting out the metaphors that occurred among the literatures sampled for this study. Madsen developed a set of usefulness criteria that require a good deal of evaluation by users, but his characteristics seemed insightful. A good metaphor is one that has the following characteristics:
Butler (1986) provides criteria for evaluating the usefulness of metaphor in a sense very similar to the intent of this study and the choice of literatures examined. His interest was in the success of a metaphor as a communications vehicle for the explanation of technical material. His basic classification of metaphorical types include:
The following search strategy was used. After restricting each search statement to the title field only, the search statements were ORed together to form the final set from which the top 100 titles were printed. They were provided in reverse chronological order, newest first. The goal of 100 titles was easily reached within the date range of 1995 publications in all three of the databases chosen. So there was uniformity in all three databases as to the strategy, the number of titles in the sample, and the date range of the articles from which the titles were taken.
NREN OR NATIONAL()RESEARCH(1W)EDUCATION()NETWORKIn detecting the presence of a metaphor, Katz, et.al. (1985) was used to provide a good operational definition. A metaphor was described as an expression in which one thing is compared to another in a nonliteral way. As an example, the article titled "Noorda-backed start-up says Linux will roar," compares the Linux operating system to some roaring animal. Since Linux cannot literally "roar" the occurrence of "roar" becomes an eligible instance of a metaphor. A sample of some of the titles from the Computer Database is given in an Appendix with the metaphorical references highlighted.
NII OR NATIONAL()INFORMATION()INFRASTRUCTURE
INFORMATION()SUPERHIGHWAY OR INFORMATION()HIGHWAY
INTERNET OR BITNET OR ARPANET
INFOBAHN
CYBERSPACE OR WORLD()WIDE()WEB OR WWW
The proportion of titles containing metaphorical references was highest in the Computer Database (70%), second highest in frequency was the Magazine Index (65%), and finally in third place was Information Science Abstracts (55%). It seemed likely that the Computer Database would prefer literal expressions over non-literal references to explain more technical detail, but upon reflection, it seemed to be not particularly surprising that the Computer Database had the largest number of titles with metaphors since a very high proportion of the articles are written to describe the current computing arena to the personal computer owner and the business user.
In attempting to group the metaphors into families or types, a very inductive approach was used. First, categories or families of metaphors were generated from the metaphors found as each title containing a metaphor was considered. This was done by the author and continued until a group of categories that seemed to include all occurrences within the ISA titles sample was established. This group of categories was then provided, along with the titles containing metaphors to a second coder, a graduate student unfamiliar with the author's interests. The second coder's sorting of the metaphors into those same categories was attempted. Disagreements were discussed and categories were collapsed or renamed to settle the majority of differences found between the two coders. When a percentage of agreement of 90% was achieved between both coders, the categories were used to examine the other two literatures-titles from Computer Database and Magazine Index. [2] The following families were established for the titles from Information Science Abstracts without the need for the ubiquitous “Other” category. Coding the other two literatures was less successful with these general families, but more on that shortly.
Space (12%) e.g., "outer limits," "cyberspace," "field of dreams"
Travel (20%) e.g., "road map," "highways," "toll booth," "road kill"
Frontier (12%) e.g., "homesteading," "tracking," "true grit,"
Fire/Water (6%) e.g., "surfing," "stream," "flames,"
Commerce (14%) e.g., "marketplace" "profiting" "publishing"
Buildings/Politics (15%) e.g., "village," "town hall," "library"
Animals (6%) e.g., "spiders," "spinning," "dragon," "roar"
Anthropomorphic (15%) e.g., "rich-poor," "dreams," "wet feet," "fear & loathing"
These eight metaphor families formed the basis for the Web-published survey that did include the ubiquitous "Other" for respondents to list their particular favorite if it seemed to fall outside of these eight.
As indicated above, in the interest of an across-literatures comparison, the same categories were used to code the titles from both Computer Database and Magazine Index. As mentioned, however, the classification of metaphorical references into these families was not so easily achieved. It was necessary to create new categories of metaphorical instances for many of the titles from Computer Database and Magazine Index. By further collapsing categories, some improvement of the intercoder reliability problems were relieved, but at the expense of achieving any degree of finely drawn categorical distinctions. For example, Space, Travel and Frontier categories were collapsed into a single Travel/Explore category to accommodate items like "passport," "guided tour," "sails around," etc. Commerce and the Buildings/Politics categories were collapsed to accommodate items like "house hunting," "robber barons," etc. For the Magazine Index, it was necessary to create a category called Communication containing references to "mailboxes," "access," "Yellow Pages," "talk," etc. And for both the Magazine Index and the Computer Database, there seemed to be a heavier occurrence of Tool/Machine/Process related metaphors such as "sledgehammer," "tool kit," "control," "spins," etc. These additional categories are included in the chart below that compares the three database literatures by their respective proportions of metaphors by type.
| Metaphor Family | ISA | Computer Dbase | Magazine Index |
|---|---|---|---|
| Travel/Explore | 44% | 15% | 15% |
| Fire or Water | 6% | 1% | 2% |
| Commerce/ Politics | or Place29% | 16% | 2% |
| Animals | 6% | 6% | 8% |
| Anthropomorphic | 15% | 43% | 23% |
| Communication | --- | --- | 15% |
| Tools or Machines or Process | --- | 10% | 10% |
| Other | --- | 9% | 25% |
| TOTAL | 100% | 100% | 100% |
In trying to classify the metaphorical references in the various titles into Butler's categories of metaphor assessment--location/place, action, qualities, and model (containing two or more of the preceding categories), there was much poorer intercoder agreement. The best PAI score for intercoder reliability reached was a poor 58%. Nevertheless, the categories are shown below, but they should be considered only a poor approximation to establish the proportions found within the titles drawn from the three databases.
| Butler's categories | ISA | Computer Dbase | Magazine Index |
|---|---|---|---|
| Location/Place | 35% | 23% | 39% |
| Action | 29% | 42% | 35% |
| Quality | 22% | 28% | 23% |
| Model | 16% | 7% | 3% |
| TOTAL | 100% | 100% | 100% |
The difficulty encountered in sorting the metaphors into Butler's categories seemed most noticeable when trying to detect the metaphors that might qualify for the Model category. To be included in the Model category required that the metaphor carry the potential of being included in two or more of the preceding categories. For example, when choosing a metaphor like "library" for the model category, “library” needed to qualify as a place/location metaphor and also as a metaphor which would be seen as appropriate to the actions and/or qualities category. An information professional would understand the actions and qualities attached to the metaphorical reference to "library" but a user might not. Similarly, examples like "town hall," "village," "marketplace" are each strongly associated with a place, but can also be seen to be associated with actions and qualities as well. There was strong agreement between the coders that metaphors like "commerce" and "marketplace" form a strong, model-like metaphor for the Internet. This same opinion has been expressed often among the responders to the Web based survey.
When considering the Computer Database, a surprisingly large 43% of the metaphorical references in titles could easily be characterized as Anthropomorphic. In the more general literature of the Magazine Index, the single most identifiable category was also in the Anthropomorphic area with 23%, but there were fully 25% of the metaphors in the general literature area that defied classification and were assigned to the inevitable Other category. What conclusions can be drawn from these differences between literatures? Few, but one that seems likely is that there may never be a "best" or most appropriate metaphor for this new hyperlinked information environment. It may simply be viewed differently by the various groups of users who find their way to the World Wide Web. Clearly, if LIS hopes to serve those users, more data must be gathered. It must also be gathered directly from the user, as well. However, the present effort can hopefully indicate the breadth and scope of what should be expected.
Striking a more editorial tone, there seems to be a greater care taken when choosing metaphorical references in the LIS literature. As a excellent example, a recent article by McMurdo (1995) states that in choosing the name for the Lycos search engine developed at Carnegie Mellon and made available in July, 1994, the name Lycos was taken from the lycosa kochii or wolf spider. The name was deemed most appropriate since the wolf spider is active at night and catches its prey by pursuit rather than by creating a web and waiting. The Lycos search engine essentially emulates its metaphorical reference by skipping from server to server gathering documents as it goes.
Another recent commentary on creating a modeling or process-oriented metaphor for the Internet comes from Cronin and Hert (1995). They put forward and elaborate with actual search examples the metaphor of "scholarly foraging." They describe the Web as an "intellectual bazaar" with offerings both serious and whimsical, but suggest that the "foraging" metaphor is a most appropriate expression of the process required because of the uneven offerings. A similar note was struck by one of the titles retrieved for this study which used the phrase "picking among the clutter." Another voice raised for this sort of metaphor came from Jim Harris who from his own Web site at http://www.davis.k12.ut.us/knowlton/LDSBC/metaphor.html suggested that the Web could be likened to a library with all the books thrown on the floor. This image, he suggested, was derived from the pictures of the Dade County Public Library after hurricane Hugo.
It also seems appropriate to echo the sentiments of Henry (1993) who examined some of the government's metaphorical offerings that likened the Internet to "Gutenberg's press" or a "national co-laboratory" and then cautioned us to be careful of promoting a "myth woven in fiber optic strands." (p. 31) Clearly, the futures ahead we can only glimpse, but in training the user to explore and exploit the Internet, it will be essential to chose our metaphors with care.
Carroll, John M. and Robert L. Mack. (1995) Learning to use a word processor: By doing, by thinking, and by knowing. In Baecker, R.M. et.al., Readings in Human-Computer Interaction: Toward the Year 2000. San Francisco: Morgan Kaufmann Publishers, Inc. pp. 698-717.
Cronin, Blaise and Hert, Carol A. (1995) Scholarly foraging and network discovery tools. Journal of Documentation 51(4):388-403. December, 1995.
Henry, Charles. New technology, more technology: NREN metaphors. Wilson Library Bulletin November, 1993 pp. 30-32.
Katz, Albert N.; Paivio, Allan; Marschark, Marc. (1985) Poetic comparisons: Psychological dimensions of metaphoric processing. Journal of Psycholinguistic Research 14(4): 365-383.
Katz, A; Paivio, A.;Marschark, M; Clark, James M. (1988) Norms for 204 literary and 260 nonliterary metaphors on 10 psychological dimensions. Metaphor and Symbolic Activity 3(4):191-214.
Lakoff, George and Mark Johnson. (1980) Metaphors we live by. Chicago: University of Chicago Press, 1980.
Lindeman, Martha J. (1991) Interface styles. In Dillon, Martin, ed., Interfaces for Information Retrieval and Online Systems: The State of the Art. New York: Greenwood Press, 1991.
Madsen, K.H. (1994) A guide to metaphorical design. Communications of the ACM. Dec., 1994 37(12):57-62.
McMurdo, George. (1995) How the Internet was indexed. Journal of Information Science. 21(6):479-489.
Ortony, Andrew, ed. (1979) Metaphor and Thought. Cambridge University Press, 1979.
Stempel, G.H. (1955) Increasing reliability in content analysis. Journalism Quarterly. 19:449-455.
Vaananen, Kaisa. (1993) Interfaces to hypermedia: Communicating the structure and interaction possibilities to the users. Computing & Graphics 17(3):219-228.
Wozny, Lucy Anne. (1989) The application of metaphor, analogy, and conceptual models in computer systems. Interacting with Computers 1(3):273-283.
The trees for the forest. (using the World Wide Web) (Multiple Access) (Network Edition) (Column)
Netscape Navigator 1.1N sails around the World-Wide Web. (First looks) (Software Review) (Evaluation)
Hassle-free Internet. (includes related articles on Editors' Choice, glossary) (overview of four evaluations of netWare Internet gateways)
Web expected to change the face of business: analyst. (World Wide Web)
Microsoft opponents go for blood; after Intuit victory, rivals take aim at Internet services bundled with Win 95.
Getting your feet wet on the Net. (alternative strategies for developing an Internet presence)
Real progress: the Internet as information utility.
Multimedia, Internet technologies tie knot.
Square-dancing online services change partners en route to the Internet.
2 Stempel’s (1955) Percentage Agreement Index (PAI) was used which creates a proportion of the number of classification decisions that were in agreement compared to the total number of decisions made.
© 1996, American Society for Information Science. Permission to copy and distribute this document is hereby granted provided that this copyright notice is retained on all copies and that copies are not altered.