IMS_Blog

My initial solution, which I actually used for a while, was to simply enclose the Roman number in a span tag and use the title attribute to convey information. This is fine for visually sited users who happen to mouse over the text but for those unable to do so, who don’t think to do so or who aren’t “reading” the site it’s obviously of not much use. It’s also not really offering anything in the way of a semantic addition (span is supposed to be semantically empty after all). I’ve tried Googling around the topic but I couldn’t actually find much discussion. What little I did find seems to hold with the use of abbr or acronym to enclose the number. The other semi-obvious approach would be to take the route used by Colin in his ALA article and use the dfn tag. To tell the truth I initially wasn’t really happy with any of these solutions.

In grandest tradition, let’s take the second option first. The HTML specification says of the dfn tag Indicates that this is the defining instance of the enclosed term. I fail to see how having a Roman number (or Roman ordinal) can be a “defining instance” of that Roman number. Even if one links it to a glossary, as Colin does, then surely only one of those can be the defining instance. In any case, I’m unhappy with the idea that I’m somehow defining something (indeed, Colin doesn’t define the thing he’s enclosed within the tag. Instead his glossary is a definition of Roman numerals in general. This just doesn’t sit well with me).

Having dismissed dfn we’re left with abbr and acronym. For the sake of brevity I’ll just refer to the abbr tag and leave a debate about abbreviations and acronyms (and indeed IE support) until another day. My first reaction was basically, no, that can’t be right. A number is not an abbreviation. And I still believe that reaction was essentially correct. However, in my usage case my reaction is noticeably wrong. I’m not using the Roman numeral as a simple number but rather as a shorthand version of “The Third” (to continue the original example). This is an obvious abbreviation, and so it seems that the abbr tag is the correct thing to use to indicate this. In fact this seems so obvious now that I wonder what my initial confusion was.

However, I cannot agree that Roman numbers should be marked in the same way. Semantically there is no difference between 3, III, and 11 (binary). I suppose there is an argument that III is a shortened form of the spelt out word three and should have an abbr tag but in that case so should 3. It’s a symbol, not an abbreviation. Much like we don’t tend to mark © in an abbr tag (though perhaps we should?). Really the spec needs a symbol tag. Perhaps it might be possible to claim that such symbols are merely unusually “shaped” abbreviations for a phrase (but then the words making that phrase are themselves symbols in a sense) but sometimes a symbol encapsulates a concept as well as a phrase.

As the previous paragraph probably demonstrates, my thinking on this point is not particularly clear. I don’t know the answer. While I think abbr is incorrect it might be the best available. Though as it seems likely (but is difficult to confirm) that screen readers will correctly handle Roman numerals then perhaps a simple, semantically empty, span is the best way to go.

Addition: Just in passing I thought I’d mention that whilst writing this post I wondered what to do about the abbr and dfn tags. Should they get surrounded by abbr? In the end I decided not. They’re code segments. That’s what the code is, it’s not abbreviated. I did stick a title attribute on the code tag though, on the basis that it wouldn’t hurt.