Over the past two years I have been attempting to increase the proportion of pictures to text in three book length manuscripts. I have been asking the question: How can we replace text with appropriate images? Can we develop a "rhetoric" that will accommodate a ratio of 50% or more visual material? Some of the issues that have emerged are: What are the functions that the word elements perform best and what are the functions best served by the visual element? How can we avoid the usual repetition in the text of explanation that occurs in the table or diagram. What kinds of new rhetorical units are useful when one begins to think about composing such fully integrated visual-text books? How are the overall canons of page design affected by this high visual-to-text ratio? What are the limits we are pushing when we try to develop communications in this manner? The paper will be illustrated with overhead slides from the three manuscripts. Since there are still far more issues than answers at this point, the author looks forward to contributions from the audience as he poses and illustrates the issues.
This story begins with my observation over a number of years that visuals/ graphics/ illustrations/ diagrams
than prose all by itself. I don't have to elaborate on this at a conference like this.
But the feeling was that most of the time we don't quite get it right. There is still too much text and not enough visuals in our documents and on our computer screens. Or perhaps a better way of putting it would be that the text still dominates our approach and visuals get stuck in for one reason or another (often for very good reasons).
So, my conclusion is that over the years we haven't taken advantage of the capacity of visuals to improve and enhance our communication. The reasons for not using sufficient visuals have usually been:
But I saw in my mind a vision of a book where what was best expressed visually would be visual without all of these reasons standing in the way.
I wanted to explore what would happen if we tried to put in as many visuals into a technical book as we needed. This of course is not an objective criteria--it depended upon what visuals I was capable of inventing and executing.
Let's look at the outcomes. In the first place, we have approx. 50% visual material measured by square inches covered.
What I want to do in this talk is to show you what it look likes and itemize some of the issues that came up for me as I made the book and as I look at it now. We have not done any particular research, evaluation, or testing on the book as yet.
I will describe some of the central elements of our approach and what the implications and issues are.
Visual language. We used an approach that we have come to call visual language. Visual language is the combination of words, images and shapes into integrated communication units. Those of you who know our Information Mapping approach will realize that what we have done in this book is to overlay a much more highly visual approach on top of our our analytic rhetoric. Some of the imperatives for book design flowing from this approach are:
1. Integrate Words and Images. Illustrations and diagrams should be as close to the words that they go with as possible: there should be an integration of meaningful elements rather than a separation.
2. Page Designs Convey Structure of Information. Use different page designs that aid the reader see the structure of the information.
3. Chunk and Label with Information Standards. Use Information Mapping's method of analysis of information, especially the analysis of text into blocks and maps, for the elementary chunking of the information.
4. Use Appropriate Elements for Different Functions. Visuals do certain things best. Prose does other things best. Shapes do something better than either. Use the appropriate visual language element for the communication job.
5. Avoid Repetitive Explanation. Instead of the frequent repetition in the text of explanation that occurs in tables or diagrams, place the diagram, illustration, or table immediately in the text so the eye doe not have to scan at all for the information.
6. Repeat Chunks on Different Pages if Needed There. Repeat diagrams and illustrations and tables (or parts thereof) rather than sending the reader backwards or forwards to find them again.
7. Avoid Requiring Eye to Track Unaided. Lead the eye with lines or arrows rather than using "callout" devices such as matching letters or numbers and forcing the reader to scan a block of explanatory or labeling text that accompanies an illustration.
8. Use Tables Rather than Prose for Appropriate Material. For material that lends itself to regular tabular arrangement, use tables rather than paragraphs.
9. Design With Double Page Spreads. We have used both single and double page designs in our work. The single page ones simply accept that the page facing may in one way or another not go with its opposite page. But as we have worked more at the books we have found ourselves designing double page spreads more and more.
steps make a difference.
Attention to many small incremental steps make bigger differences to both the reader and the writer. My example here is the slowness of having to look at an illustration and trace down in the block of text how it operated. Look at this example:
Notice how slow and frustrating it is to look up at the diagram and then down at the text to see how the thing operated. Observe how easy it would be to skip a step. Notice how how there a kind of "scanning drag" in looking at the illustration. These problems, while small, if repeated in many places in a book lead to cumulative frustration, slow task completion, and increase the number of errors made in using the document on the job. This is easily fixed by arranging the prose around the illustration and running arrows to the appropriate parts of the illustration. There are many such small incremental improvements, which if implemented with consistency, make extremely large differences in the efficiency and effectiveness with which a book is used. I will mention one other that came up incidentally to our developing the book. The use of a large proportion of the space in the book for illustrations gives the book an entirely different texture. Most people who have seen it have called it an "inviting texture." I feel that I could dip into any page and get something out of it, even though I realize this is a technical book.
1. The Grid. As I understand the grid approach in document design, you set of some kind of set of areas arranged in a box or matrix like structure. Then you adopt the rule that you will place type or illustrations strictly within the boxes or adjacent boxes. This results in a feeling of solid structure and predictability for the reader. I felt compelling reasons to put illustrations and diagrams in the text, not located outside the flow of the text in some grid slot. This resulted in what I would call a modified grid approach. We modified the rule. We kept some of the grid structure in width or horizontal dimension for most blocks of type. We used it for placement of the title of each double page spread.
We violated it by letting the information run as long as it needed, rather than stopping at the end of a block.
We further violate the basic gird rule by placing visuals in or close to the text as they were needed in the flow of explanation or meaning.
Putting this in a positive form, we used the rule: Text should go for as long as it needs to determined by the writer.
2. Busy Look. Some commentators, especially professional designers have criticized the book as graphically "too busy." In part this has to do with the proliferation of drawings. In part it has to do with the crowding of some pages to get everything in to a double page spread. Yet, we felt that the tradeoff of either leaving out material or artificially dividing the topic into two and running it over onto another double page spread was worse than the solution we selected.
3. No "visual elements." Sometimes we encountered pages where there was not "good" or "logical" illustration, diagram, or other visual element that could be used. This meant that the double page spread would be all prose. That frankly, looked funny. Our approach has been to put some relevant illustration somewhere on the page to keep the texture similar to that of the rest of the book.
4. Uneven Amount of Information on Different Pages. People have subjective measurements for the amount of information on a page or a double page. A Couple of our early readers complained about the large differences between the amount of information on different pages. Usually the great amounts were there same but the differences in amount of text seemed to determine what they were noticing.
5. Style. The problem here is the question : How do you devise a visual style that people in commerce will accept? Another way of putting it: How to avoid looking like a comic or a children's book or like advertising in a magazine. Our solution has been to follow the illustration style pioneered by graphics artists working for many of the large newspapers, the so-called information graphics style. We have much to learn from these pioneer speakers of visual language.
Levels and Meaningful Units.
How shall we analyze visual language. What is it's rhetoric? What are the basic visual language units? What corresponds to syntax in purely verbal language? What can we say about the semantics of the language? How do we chunk this task? How do we divide our analysis into levels?
In prose we usually take words and sentences to be the minimum meaningful units. Of course we have to incorporate this analysis of prose into our visual language rhetoric.
But we must skip the paragraph because it does not have sufficient precision. A better start in developing clusters of meaningful units for some domains is the typology used in the Information Mapping approach. This divides clusters of one or more sentences and or diagrams into carefully delimited and defined "information blocks." (Horn, 1989) Each of the blocks is a particular type and it resembles all other blocks of that type along specific dimensions for each block. Certain of the blocks appear with great frequency in particular discourse domains (e.g. training manuals as one domain and proposals are another) and in particular information types (procedure, process, structure, concept, principle, fact, classification). Among their other properties, information blocks are always labeled usually with a label that tells the reader the function they perform, but sometimes with labels that do other things. Because of the patterns (both taxonomic and empiric) the Information Mapping method has been called the closest thing we have to an engineering of documentation.
We can illustrate this with a block we call a parts-function diagram. This is a semantic unit. There are different ways of showing such a diagram. One is as a table:
Another is with the parts assembled and names and functions described in prose linked by lines to the diagram.
The next level of clustering is the linking of blocks around a topic. In the Information Mapping method, we call this larger group of (usually) 2 to 9 blocks an Information Map. In the book I am discussing in this paper, we usually arranged to have one Map correspond to a double page spread. This is not the only format that we use, or even the most frequently used format. But it seemed to work well for a book on which we were going to superimpose the visual language approach on top of the basic Information Mapping analysis.
Inside blocks, we can see different structures. Some blocks are entirely prose. Some contain or are comprised entirely of tables. Others contain or are comprised entirely of illustrations accompanied by their labels and perhaps other chunks of text.
In the example we have used above, the parts-function diagram can be further analyzed into the drawing(s), the labels, the function descriptions, the lines for the table or the lines or arrows that connect the drawing to the prose.
Diagram Clusters With a Double Page Spread
Sometimes we found that we were not simply putting information blocks in a single column down the page, which is perhaps our most common format. But rather, we grouped the blocks into a larger diagram or shape that gave a visual shape to the entire double page spread. We don't have a taxonomy of these at yet, but I would like to show you some that are quite different. We don't have a name for them either, perhaps we should call them "double page clusters."
Horn, R. E. Mapping Hypertext: Analysis, Linkage, and Display
of Knowledge for the Next Generation of On-Line Text and Graphics.
The Lexington Institue: Lexington, MA, 1989,