Some Bold (and Italic) Thoughts on HTML 5
Internationalization Activity leader Richard Ishida comments on the
HTML 5 draft, The formerly deprecated
i elements of HTML are defined in terms of their appearance: for example, the name of a ship can be bracketed with
<i>…</i>, because ship names are italicized. The problem is, of course, that this applies to English, but not to Japanese, for example. Should these tags be suppressed in favour of something ‘more semantic’ (whatever that means)?
Where it Started
HTML inherited its inline elements from the Texinfo format. In Texinfo, there were
directives for the types of emphasis expected in computer documentation (
@var, ...), many of which have no equivalent in non-technical
prose. To be honest, most of them are not used in most technical documentation, because of
ignorance or unconcern induced by the use of Microsoft Word, whose default styles do not
include anything for computer text.
Because there were no tags for book titles, foreign tags, binomial names (as in Homo
sapiens), and all the other things we conventionally use italic for, the advice was to
<em>…</em>, which frankly is no more semantic than using
anything it is worse, because it means
em is no longer consistently representing
It Gets Worse
To make things worse, Microsoft Internet Explorer’s default rendering of
dfn element shows they misunderstood their purpose, hastening their
var element was originally for ‘variables’ such as x and y in mathematics or placeholders in syntax definitions, as in the following example:
Type the command
rm -rfdir to delete the directory dir and all its contents.
The usual convention is to distinguish the varaiable (dir in this example) using italics. MSIE instead used the monospace (‘typewriter’) font used for computer text, seemingly thinking that
var was for identifiers found in computer programs.
dfn element was for defining instances of a term, as in the following example (stolen from Wikipedia):
A definition is a statement of the meaning of a word or phrase. The term to be defined is known as the definiendum (Latin: that which is to be defined). The words which define it are known as the definiens (Latin: that which is doing the defining).
The typographical convention used varies from book to book; I have seen italic, bold, and
bold italic used. Italics makes sense to me because it corresponds to the
emphasis that is given to the term when speaking a definition aloud. Boldface makes sense
because it allows one to find the term later when skimming backwards to remind yourself of
the definition. Most web browsers do not distinguish text enclosed in
Both of these problems are non-issues nowadays, since we can override the default styles using CSS. But during the late 1990s this was not possible, so even the vanishingly small percentage of the writing population that knew and cared about such distinctions had to drop them.
Is the Solution to Add More Tags?
When HTML started being used more widely they could have added more elements for embedded foreign words, words mentioned but not used, etc., but never quite got around to it. It is probably just as well: this process could go on forever as technically-minded folks are liable to invent distinctions that would never be used by normal people. Donald E. Knuth, for example, used two different forms of italic in his books (traditional italic for emphasis, slanted for book titles), and I have seen people who give different meanings to ‘…’ and “…” quotation. The problem is that this makes it even harder for a non-specialist to know which tags to use to mark up a given phrase, which makes it more likely that they will throw up their hands and refuse to try to learn.
And let’s not forget that most people who write do not write their own tags, but just mash ⌘I to make their word processor switch to italics mode.
What the HTML 4 standard did do was introduce the
q tag to represent quotation. This
was, in my opinion, a serious error, and has cost web-browser implementers hours of
labour that could have been better spent doing something useful.
The HTML-5 approach is to take the opposite tack, and un-deprecate the
i tag, giving it the meaning roughly ‘something that would normally be italicized’. In scripts that do not use italics this might be taken as meaning that the
i tag is not useful with those scripts, or that it might be used to mark up text whose connection to italic fonts is more tenuous.
It this seems too simple consider that the Guardian newspaper goes one further, and distinguishes book and movie titles, ship names, and all that jazz with the use of capital letters alone: they do not use italics for these. In the busy environment of a news office this probably gains them hundreds of person-hours’ productivity a week by eliminating a whole class of argument between sub-editors.
One question Richard Ishida raises is that this may mean that a translator of a passage has less information to work with than if more detailed semantic mark-up were used. While I can see the point I don’t think it is likely to make a big difference practically.
For example, suppose our hypothetical translator has the following paragraph to translate to Japanese:
Captain Sulu beamed up to the Excelsior and handed Krorg’s bat’leth to the transporter chief to dispose of. I almost didn’t make it this time, he thought ruefully.
Already she needs to know whether to render Sulu’s name as Kato (as used in the Japanese version of the TV show) or some transliteration of Sulu, the Japanese transliteration of Krorg’s name and the Klingon term bat’leth, and for that matter a decision on whether Excelsior’s name is to be transliterated or translated. At this point I imagine that recognising that Excelsior and bat’leth are italicized for different reasons and these correspond to different punctuation marks in Japanese is the least of her troubles.
This is what XML is for
There are still situations where the complete and accurate marking up of text is desired—generally in an academic or other technical context. For these purposes, an application-specific XML vocabulary seems like the ticket. Trying to add all the possible extra meanings people might need to HTML will only bloat HTML (which is already too complex) while probably omitting exactly the special meaning you need now.
The alternative if sticking to HTML would be to use
class attributes in the obvious way to augment the information conveyed by the
<p><span class="nameprefix rank">Captain</span> Sulu beamed up to the <i class="ship">Excelsior</i> and handed <span class="personname transliteratedfromklingon">Krorg</span>’s <span class="foreignword transliteratedfromklingon">bat<span class="glottalstop">"’</span>leth</span> to the transporter chief to dispose of. <i class="internalmonologue">I almost didn’t make it this time</i>, he thought ruefully.</p>
This saves us the inconvenience of having to add a definition to the HTML 5 standard describing the tag used to represent telepathic speech in sf novels.