Back in the day (i.e. the 1990’s) I managed a web site for an educational institution here in the United States. When you do that you of course have all of the usual concerns about having as good design, gathering and presenting all of the information needed by all of your audiences, and so on. But one thing that people don’t always think about is accessibility. In the U.S. there is a law called the Americans with Disabilities Act that says that any organization that receives federal funds must make its facilities accessible. Now I personally would have wanted the site accessible even without a law, but as a practical matter it really helps to have a law when you are dealing with your own executive leadership who might question why you are putting so much effort into something that does not seem apparent when you look at the site.
Achieving accessibility required investigation and learning some things. And one of the most profound things I learned had to do with the separation of presentation from content. What this means is that the actual information you are presenting should be independent of the way the information is presented. In the simplest case, this means that a poem should have the same meaning whether it is typed out or hand-written. And when I say it that way you might respond “Well of course, what difference does that make?” And yet, there is an issue here which anyone trained in graphic design would see right away. We use visual cues all of the time to signal something about the meaning of text. For instance, the book I am reading right now has chapters. How do I know when a new chapter occurs? Well, the previous chapter tends to end with the text not filling the page, that is clue number one. So when I see a page with text only going half-way down, and all of the rest of the page blank, I infer that the chapter is ending, and I would anticipate a new chapter beginning on the next page. And if that next page has a line at the top that is separated from the rest of the text, is in a larger font, and is in a bold font, that will confirm that a new chapter has begun. And within the chapters this particular book (right now I am reading Don Tapscott’s book Wikinomics) there are sub-sections, but they are clearly marked by separate lines, bold fonts, etc. This is an example of a visual language that is communicating information that is not in the raw text. Another example of this is the use of italics to provide a sense of emphasis to a word or phrase within a piece of text. In my teaching days I used to point out that italics were an author’s way of saying “Pay attention, this is important!” And again the point is that this is actual information that is being communicated not through the raw text itself, but through the presentation of the text.
In the world of print this works pretty well. Talented graphic designers can use the tools of visual presentation to create compelling content that informs and entertains us. And yet the World Wide Web Consortium (aka W3C) has said this visual language should not be used this way. Why is that? Well, as I said earlier there is that accessibility issue. Using the visual appearance of text to communicate meaning is going to leave out one large group, the visually impaired. But in fact the issue is even deeper than that. Web content needs to be interpreted and displayed in a variety of ways. For instance, you might be looking at this page with a large monitor, like my 25″ desktop monitor. Or you might be looking at it on a 12″ laptop. Or on a 7″ tablet, or a 4″ phone. This poses a problem for those talented graphic designers that they don’t encounter in print production where the dimensions of the product are fixed.
But that is not all. People are not the only ones who need to access and understand this data. Every day I expect every person reading this post uses a search engine of some kind (I like Google) to find Web pages. This is done using software to “read” the page and abstract from it the basic ideas of what the page is about. This is a specific case of the more general case of machine reading. Another example is software that turns text to speech, which originally was developed to assist the visually impaired, but which is now becoming even more broadly applicable. For instance we are starting to use software to take electronic texts and turn them into spoken word for people whose vision is fine. As an example, you might want to listen to a book you have while driving, while doing housework, or during any task where your vision needs to be freed up. This is addressed in a good article at the Universal Usability web site.
Jakob Nielsen, in his classic text Designing Web Usability (New Riders, 1999) talks about this in terms of what he calls semantic encoding. By this he means encoding the text in terms of what it means and how it functions within the document, not in terms of how it will look. So to Nielsen it would be very wrong to take the title of your page and place tags that make the font bigger and bolder as a way saying “This is the title”. Instead, we should us a tag called <H1> that directly communicates “This is the title of the page.” And the benefit of doing it this way is that any person or device that accesses this page will unambiguously know what the title is because it has been properly tagged. Screen readers for the visually impaired, text-to-voice software, search engine “spiders”, it doesn’t matter, all will now exactly what the page title is by the proper tag.
This becomes important when searching, for instance, because it helps improve the quality of the search results. If a piece of software simply reads the page and records the appearance of a word, how significant is that word to the meaning of this page? If the word is in the title, it probably means this whole page is about that topic, whereas if it appears somewhere in paragraph 5 it may well be less important. Semantic coding helps solve this problem. If you have ever had the experience of doing a web search and winding up on a page where you wonder “Why did they send me here? This is completely irrelevant.” there is an excellent chance you are the victim of a page designer who never understood semantic coding.
OK, you may be thinking, but suppose I want my titles to be in a bigger, bolder font? And suppose I like using italics to emphasize words or phrases? After all, this is a visual language that the majority of us have grown up with, it is one that we are fluent in, and it really helps us to build more compelling pages. And that is why W3C did not say we should get rid of visual language, only that we keep the visual presentation of that content separate from the semantic meaning of the content.
The way we handle this on the Web is though the use of style sheets. Style sheets (known as CSS for Cascading Style Sheets when used on the Web) are a concept that pre-dates the Web by many years. In print production this was often used to create a distinctive appearance. Some time, if you want to have some fun seeing how people think of this, I invite you to watch the movie Helvetica. Some folks with a passion made an entire movie devoted to a font. And if you have any feeling for graphic design or just appreciate geekery this is actually a great movie. I recommend it. Anyway, many publishers of magazines and newspapers would select certain type faces, a color palette, and other elements of visual design, and combine them into a style sheet that told the designers “This is how we want to appear.” You may not have noticed it before, but it is already all around you. I bet you can identify many different magazines at a glance just because you know how they look, even if you never consciously thought about it before. In terms of Web design, a style sheet is a way of saying “Every time I put this kind of element on a page, here is how I want it to look.” A style sheet lets you specify, for instance, that every <H1> element on your site (remember, the semantic meaning of this tag is that this is the title of a page) should appear in Arial font, bold weight, size 16 points, and in the color blue. And every sub-head (e.g. <H2>) gets its own font specification, every sidebar link gets a specification, and so on. With style sheets you can have control over appearance and still do proper semantic tagging of your content. Because you can apply a single style sheet to the entire site, it is a big time saver. And further, if at any point you decide to redesign the visual appearance of your site, all you need to do is write a new style sheet to get instant results. All of your actual content can be unaffected. To see how this works, check out CSS Zen Garden. This site lets you swap out a variety of very different style sheets with exactly the same content, so you can see how this works.
You might think that is all to be said on this topic, but actually, I plan to extend this insight further in another post.
Listen to the audio version of this post on Hacker Public Radio!