The Syntax of Links

Being me, I tend to apply the rules and standards of print to the web--not because I want to constrain the web to my antediluvian grammar-wonk ways, but because it gives me a reference point I'm familiar with. And, well, because I can't look at words in sequence without grammar-wonking. Anyway, it gets me to asking questions like: what is the proper grammatical usage of a link? What are the functions of a link? It links, first and foremost: it can cite or it can trigger. But before the reader follows where it leads, it sits there on the screen, highlighted, set apart. What does the eye make of that?

Links as Punctuation

links interact with quotation marks

Being a new protocol, hypertext is oddly untouched by the exacting grammatical rules that govern writing in print. Let's say, for example, that I'm writing about Al Capone, and I want to breezily refer to him as "Scarface" at the beginning of a new sentence about his indictment. Til now I've only referred to him as "Capone," or maybe "Mr. Capone." I could say:

"Scarface" was charged with tax evasion...
and I'm fine, I've implied with the quotation marks that the name refers to the man I've been talking about. In hypertext I might skip the quotes and write:
Scarface was charged blah blah...
and that works fine as well--in fact, it's a standard use of a link: the first instance of that new term becomes the anchor text for a link that explains its meaning in case you don't know.

But what if I want to use the possessive? In print that presents me with an uncomfortable choice. I can't write

"Scarface"'s career in crime came to its end when...
because it's unreadable; I suspect it's so patently ugly that no style-guide writer has even thought to forbid it. But that's what I mean, really. I want to quote the name, and then make it possessive. Curses. Alternately, I could mark the whole possessive:
"Scarface's" career in blah blah...
But that's still kind of visually confusing and ugly. I don't want to suggest that Capone's nickname was "Scarface's." Since I'm not quoting any specific speaker, and I'm sure someone somewhere has said "Scarface's" before, I'm within my rights to put it in quotes--but it really isn't what I mean. My third option, and the one I'll choose every time, is to rephrase my sentence--even though it would be a perfectly viable sentence if spoken out loud. Frustrating.

Now, look what happens if I'm writing the same piece for the web. Along comes my troublesome sentence:

Scarface's career blah...
and here I've neatly sidestepped the problem. I can anchor whatever text I want without interrupting the flow; I can anchor half a word if I feel like it, and I can certainly leave the apostrophe and the s out of the anchor. This way the link clearly pertains to "Scarface" rather than to "Scarface's" and I get to write my sentence beginning with "Scarface's" anyway.

The important point is that all this is true even if the reader never follows the link. The fact of the link, the visual highlighting of the anchor text, carries meaning by itself. It's functioning as punctuation on a page. And interestingly, it obviates other punctuation that would be obligatory without it. As in the Scarface example, I can introduce a new term without quotes if I use a link.

Well, that's not true for every usage: If I'm actually quoting someone's words, I still need the quotation marks, link or no link. But for just that reason, I'd prefer to quote in plain text and anchor another phrase, or else <blockquote> the entire quote to avoid using quotation marks. Why? Again, there's no rule-book on these things that I know of. But I think "quotation marks as part of a link" just look dumb and "an anchored passage within unanchored quotation marks" looks even dumber. So I'd rather avoid it.

The <blockquote> option is questionable, too, because it tends to lead to long anchors. Anchor text draws the eye, and like any eyecatching thing, too much of it is just annoying. The appropriate length of anchor text is inevitably subjective; I would suggest that a link should be long enough for someone with wobbly hands to hit with a mouse, and short enough for all anchored words to be easily held in the mind at once, as a phrase. Loosely, I'd recommend that the length of anchors be proportional to their importance for the reader, especially for multiple links close together in a text passage.

links interact with titles

Here's at least one other case where a link trumps the need for traditional print markup: titles. If I write about a novel, like Always Coming Home by Ursula LeGuin, I need to underline the title; if I refer to short story, or a poem, the title gets quotation marks: "His Coat So Gay" by Sterling Lanier, "Brown's Descent" by Frost--unless it's a poem long enough to fill its own book, like Beowulf. Old hallowed rules. Introduce hypertext and they go by the boards: I can shamelessly recommend Nabokov's Glory, or Jack Gottlieb's In Love by Ellen Bass, without any such efforts. The link is marking enough to the eye. (Not just because the default representation of a link is blue text with an underline; it has the same authoritative force on a page whose links are not underlined.)

more questions and few sources

Mind you, these are rules I'm divining or formulating on my own. Strunk & White doesn't touch this. The Chicago Manual of Style takes a jaw-dropping hard line: why, certainly you may use a link, if it's within a full and proper print-style citation, same as always!

The geeks responsible for the web provide us with a style guide, naturally enough, but it's really more concerned with writing style than with mechanics. Use important and relevant words for anchor text, you don't have to bother saying "click here," that kind of thing. This sort of advice is easy to find. The more academic of coders are highly concerned with usability in writing style; often, as we'll see, they can be ominously didactic about it.

But nobody, in short, seems to be talking about the nitty-gritty usage questions that attend the task of writing a document studded with links. Do you enclose in an anchor the same text as you would enclose in quotation marks, including some incidental punctuation? Or do you, on the principle that the link pertains to the word itself, but not the comma, leave the punctuation outside the anchor? This latter seems to be the tacit standard, but I've yet to find it codified explicitly.

Beyond punctuation, it seems plain enough to me that when certain words in a document are glaringly marked, rendered in a different color, this will function as an emphasis in and of itself. The style guide of the Library of Congress has noticed this as well, and sadly concedes it to be a necessary evil: "Because hypertext links appear in blue and often are underlined, they may unintentionally highlight and emphasize the link text. This will have the same effect as inappropriately making text bold. Although this cannot be avoided, be careful not to overlink."

Well, maybe. I would prefer to think that the emphatic quality of anchor text is simply one more tool in the writer's box--not a troublesome symptom of HTML but a technique that can be used quite deliberately and to good effect. But on this matter, too, there is precious little discussion to be found at the time of this writing.

Forthwith, into the breach.

Links as Emphasis

emphasis on the web

To emphasize text in print, a writer uses italic, bold, CAPITALIZED, or underlined characters. To emphasize text in an HTML document, all these things are possible (obviously) but officially discouraged. Instead, the standard tells us, we should use the <em> and <strong> tags:

Strong emphasis!!
For non-geek readers: I know, emphasis probably looks the same as italics and strong emphasis probably looks just like bold. That's what's standard and it's been standard for a long time. But the difference is that when I type <i> into an HTML document I'm making the choice, as the author, of what you'll see. When I type <em> I'm leaving it up to the judgement of whomever coded your browser. And there are still, by the by, a surprising number of text-only browsers out there that show them differently, and an even more surprising number of people who are quite contentedly using those browsers, and will no doubt continue.

I live at an imperfect peace with the utilitarian geek aesthetic. It's not that I don't see the crucial value of usability; it's just that the more academic geeks tend to get oddly prescriptive about their taste for spartan simplicity. Here, for example, is the first paragraph of the introduction to the WWW Consortium's style guide:

You are going to write (or generate ) some online hypertext. Because hypertext is potentially unconstrained you are a little daunted. Do not be. You can write a document as simply as you like. In many ways, the simpler the better.
Note it says you can write as simply as you like, not in whatever way you like--indeed this simplicity is offered as a relief from the intimidating lack of constraint that is presumed to be bothering you. The cynical paraphrase might be, "You are totally free to write HTML the way we think you should."

I am not ready to concede that a total indifference to style is somehow correct. The emphasis tags are a perfect example. I think I can see what they're intended for. A search engine, no doubt, can be made to assign a higher priority to emphasized words than other words, and a higher priority yet to strongly emphasized words. This characteristic of emphasis, one imagines, will tend to correspond with the importance of a word in its document (though I do wonder how reliably). So, voilá! the <em> and <strong> tags will amplify the power of searching and thereby increase the effectiveness of the web as a whole.

Sounds great. I don't mean that sarcastically: it really does sound smooth and elegant. But there is a problem. The whole plan rests on the idea that anything you have to say can be reasonably represented by two heirarchical levels of emphasis and nothing else. Italics are to be an amplification of print, and bold print simply the same phenomenon in a greater degree? This, please, is a little like expecting "good" and "plusgood" to express every kind of approval. Or, rather, every kind of approval [someone thinks] you have any legitimate need to express.

flavors of emphasis in print

But emphasis is not a simple continuum. The language of print is steeped in nuance, which writers and readers have made and absorbed and challenged and extended for many generations. A style of typeface is, in essence, a voice. Readers know every typeface cue intuitively and whether you intend it or not they will respond.

All of this is subjective, naturally. I'm formulating perceptions that are highly fluid and contextual even for me alone, and I can't hope to speak for all readers. But though I'm probably oversimplifying, I think I'm in a recognizable ballpark for readers of English. There really are conventions about this stuff, even if they're rarely made explicit.

If you think one kind of emphasis can simply substitute for another, imagine swapping bold for italics in the case of foreign words.

"It has a certain je ne sais quoi, I suppose," the colonel demurred.
It's subjective territory, but I say this just looks stupid. Why is he blaring that way? Is this a language lesson? Italicizing words has the effect of obscuring them slightly, as foreign words may be obscure to an English-speaking reader, or to whomever the colonel is addressing. Bolding words makes them jump out at you, makes them become clearer, more apparently central; italicizing words makes them seem to recede, makes the reader work harder for them. It's different qualitatively, and the visual effect should match the intent.

hypertext and colored text

The effects of these various typefaces are more or less the same on a screen as they are on paper (I'm not sure how much argument I should expect on that point) but now the hyperlink is added to them. By default, and most commonly, the link is rendered in blue underlined letters--but it can be styled to look like whatever a web author wants. Rarely, though, is anchor text the same color as basic text.

The barest effect of the link's appearance seems comparable to bold text--it stands out. Probably more so: if there's a bit of bold text somewhere on the page, and a link nearby, the eye will be drawn to the link. I think it's reasonable to say that some part of this is simply because a word in a different color is more striking than a bold one. And that, in fact, is a little puzzle of its own.

Coloring text is another way to emphasize it, of course, and it also comes down to us from the long history of print. But a critical difference makes it more widely used on the web than in print: it doesn't cost extra. Unlike a print publication, a web page can feature 256 different colors of text without costing a penny more to produce than a simple black and white version. So there's some reason to think that colored text should be rampant on the web.

To an extent that's true. There is a countervailing force, though: hyperlinks themselves. A web reader quickly learns that a link can be any color the author likes; generally the appearance of links will be consistent throughout any given site, but most readers spend at least some portion of time in an ongoing exploration of new sites, so we are trained to watch for any text that looks different because text set apart has some chance of being a link.

To avoid confusion, many web authors generally avoid using colored text for inline emphasis, reserving color tricks for headings or fine print. And there's feedback in that process; the more authors reserve in-line color changes for links, the more readers associate the two.

Even so, that's the lesser part of a link's power to pull the eye.

links become emphatic

More importantly, the effect of a link on the eye is experiential, learned by the habitual surfer of the web. A link has power: it is an unlocked door that leads somewhere. It's magic.

Loosely, one might say there are two sorts of links--central and supporting. A central link carries the point of the writing; the reader is meant to follow it, and may not be able to understand the rest of the piece without first reading (or watching, or whatever) the referenced document. The passage may have been written for no other reason but to frame the link. Central links are all but obligatory, naturally, in a weblog, which exists precisely to call found pages to the public attention and discuss them. Broadly, they are to be expected in reviews, recommendations, commentary of any sort, ads, lists of things, navigation bars. They are a sort of idealized links, the ones a reader is fully intended to follow.

But more interesting, here, is the link cited along the way as support. A supporting link is present to provide more information in case the reader needs it. If I'm writing about Henry Kissinger, and I make a passing reference to Salvador Allende, the relevance is going to be instantly familiar to some readers and totally opaque to others. Readers who don't get it, if they want to know what I'm talking about, can click on Allende's name to be spirited away to an article that I picked out to explain the background. Those who recognize the reference, or who they feel they can follow my piece without knowing the relationship, they read merrily on. It's the reader's choice; the link is illustrative, supplementary.

To continue the earlier parallel between written words and spoken, hyperlinks are like a speaker tapping a diagram with a pointer for emphasis. The sentence rolls on as usual, but the speaker is simultaneously brandishing references to reinforce it.

If a piece of writing is making a string of improbable claims without citing any supporting materials, readers might be quick to abandon it. The formidable Jakob Nielsen, in a paper on writing for web, notes the cachet that links lend to a piece of writing:

Users rely on hypertext links to help assess credibility of the information contained in websites. This point was made by 4 participants. "Links are good information. They help you judge whether what the author is saying is true," one said. While reading an essay, one person commented, "This site is very believable. The author presents several points of view, and he has links for each point of view." Another person made a similar statement about a different essay: "Because the writer is referencing other links, it's probably relatively accurate information."
That last quotation sums up the unfortunate side of this authoritative aura; people are ready to believe a page studded with links, even before they've followed the links. It may be that some readers are mollified by the appearance of groundedness and never do bother to verify anything for themselves.

This works on two levels. To the reader's conscious mind, a lot of links imply that much research underpins a document, that due diligence has been done by the author, and a crowd of somehow support the document that cites them. But because of that, and more insidiously, there has come to be a more visceral and immediate expectation of validation associated with the very appearance of a link in the eye of the habitual web reader. That is the sort of emphasis anchor text provides: an almost subliminal impression of veracity and informedness.

Writers understand this, and it becomes tempting to enhance one's credibility with an array of linked references. And for a writer, especially a rhetorical writer, it's easy to slip into a habit of placing those links right on the most powerful, most forceful phrases. And it's easy to slip beyond that into seeking out some validating page to link to--even if it's not particularly vital, or maybe, particularly reliable. Readers and writers alike need to remember the old adage: a chain is only as strong as its weakest link.

I've been able to find only a handful of sources that recognize and discuss this emphatic use of the link (let alone in depth), and all but one point it out only to warn against its abuse. One of these notes one more critical and obvious factor in the reader's reaction to a link: "Links ... force readers to think about the anchor text and decide whether to follow the link or not."

Soul of the Web: the Message Is the Medium

readers come from somewhere and only some are arriving

Of course the most immediate reaction a reader has to a link is that it represents a choice. The effect of a link as an inflection on its context cannot be separated from its function. The web ain't linear; that's the premise and the promise of it, and the trouble with it too. Every link is a fork in the road.

The information available on the web is already more than any one reader could paw though in a lifetime, and that volume increases exponentially. There's no one way to try and read through it, or navigate it, or map it. So different sorts of readers, or readers of different strategies, will encounter any given page (and depending on the site, any of them might be the intended audience):

Some readers will read an entire document; more will skim for keywords, perhaps settling in for a deeper read once full relevance has been established. Some readers check a circuit of sites they know and little else; others bounce in and out of Google all day long, sifting immeasurable reserves of information the way a baleen whale strains the sea. For every HTML document there are readers who constitute an intended audience, and many more unintended readers who could never have been predicted, and who often as not are just passing through on their way elsewhere.

Ideally, every HTML document would rapidly give each of these readers what he or she wants: retain the interested, refer seekers to relevant material when it can, and shoo the uninterested back whence they came. Chiefly, this is a matter of HTML documents identifying their nature and intent clearly at the outset, and plenty has been written about that already. But to a lesser extent, every link seeded along the way is a renewed question: will the reader stay or investigate? If the latter, will you ever get that reader back?

every user is a cartographer

In some primal vision of the web, a browser is a single window that shows one document at a time. This is a little like walking through the Labyrinth without leaving a trail. But people map the maze in different ways; some just lean on the Back button, like a breadcrumb trail retracing an exact route. Most also use bookmarks, but there are ways and ways to use bookmarks. They work well as a short list of daily stops or other chronically useful pages, but for combing through a lot of information, they're limited; it's nice to be able to slap a keystroke and be certain of keeping the url, but do it too often and you'll be forced essentially to work just as hard sorting bookmarks as you did finding them originally--unless you're prepared to invest some time and energy in keeping a large bookmark file organized and up to date.

More immediately, a user might hold two places by opening a new window, exploring one fork in the road without committing to it. Or the link might do it automatically. This tends to be confusing to some users (maybe mostly newer users) and can be tremendously helpful to information-sifters. There's a lively debate on the subject, and rather than wade in I'll point you to arguments in favor and against. I will note, though, that in sites by designers the new window is de rigeur and in the sites by more technical geeks it's nowhere to be seen. Likewise, designers' pages tend not to differentiate visually between links that have been visited and those that haven't, where code-geeks' pages stick with the default of rendering them differently.

To click on a link is to make a choice, and to take a chance. The user doesn't know how long the next page will take to load, whether it will open a new window or not, or whether it will be there at all, or simply how long a page will be. And after all that, there's the question of how relevant it is. What web reader hasn't had the experience of following a link and immediately regretting it?

the signpost is the road

As authors we do what we can to mark the trail clearly. For the most part (obviously this would be less true on a comedic site) it's a great help to the reader if the context of the link makes fairly clear what's to be expected from following it. And no context is more sensitive than the anchor text itself, which becomes a sort of gloss for the referenced document, almost an unofficial title, and in a potentially insidious way, a lens through which the document may be seen if the link is followed. Before the reader clicks that link, the anchor text and its context are the only information available about a document somewhere.

(That is, unless the author is making conscientious use of link titles, those little mouse-over descriptions of links and images. These aren't really in overwhelming use at this point, and browsers aren't always very generous about displaying them in full, but they can certainly help.)

The link itself is interestingly identified with the page to which it leads; the reader can either ignore it or take a single step toward it, a step which will traverse the whole of the road.

every step is a crossroads

In a pleasantly remarkable concession to the Sisyphean prisoner's dilemma of routing readers through the inexhaustible trove of the web, authors and their pages have acquired some interesting habits designed to declare themselves and their contents as fast as possible. Notably, the table of contents--on the web, a vastly more powerful device, half catalog and half caddy--is routinely merited by documents far too short to have rated such treatment in print. This essay is a case in point; it's no longer than a typical nonfiction chapter, might well be shorter than average. But it has its table of contents, made all of live links to very small and specific sections.

Even had I not included a TOC, I would have anchored the sections for the convenience of anyone who might wish to make reference to a small point made here. I've too often been frustrated by long pages without this courtesy. My expectations, maybe, have been raised.

It's important to remember: the rarified speed and precision of cross-referencing that well-tuned web documents offer us is not a necessity. It's a power we've only had at all for less than a decade; we've survived for untold generations without it. I don't believe our attention spans have so far atrophied that we can no longer read full chapters without skipping magically back and forth. But--what is necessity anyway? We can do this, and we've gotten used to it quickly. It works.

Maybe inevitably, that broad offer of expediency--given to the reader by spelling out the contents of the page so clearly--is also a surrender of power by web authors. Readers may reach their destinations faster, and so the web as a whole will be working faster, better, for everyone. But to bring this about, the authors must loosen what grip they have on their readers. A document with its contents laid out is far less likely to be read straight through, ever, than one made of mundane text from one end to another.

and you want to try and control this?

Some practical recommendations might be devised that could clear things up for readers: distinctly different styles for links that open new windows and links that don't, for example, or (better) different styles for central links and supplementary ones. If differentiated links became at all familiar to the general user--it needn't be uniform practice worldwide--people would adjust easily, and the bulk of pages on the web might be much more amenable to users and their objectives. In the meantime, readers need to make their choices, and writers need to understand that this is what readers do. What else would they do? The organization of web writing is wound into the writing itself. Structure itself becomes a thing that might be added on the fly for reasons of style. When making a reference becomes a choice of font rather than of phrasing, the mechanics of writing coincide with the text itself. Reading becomes multitasking, whether we like it or not.

Of course, we as writers can feel that reading hypertext will be different from reading paper, because writing it is different. We have two options in writing hypertext: either we write until we hit a statement that ought to be backed up with further information--and then plug in the appropriate link, or maybe seek one out, before moving on--or else we write a piece straight through, and then go back to pepper it with links. To taste.

Readers have a parallel choice--to read straight through and follow up on references later, or to follow each one as it comes, with the relevant discussion fresh in mind. It's no new question; reading any heavily annotated text in a book forces the same choice on us. Footnotes just never used to be so enticingly easy to find. In a way, that's the only difference, but it's a huge difference. Reading a web page, I'm often interested and anticipatory to see a link, and I was certainly never excited by footnotes. It's even worse now; reading through a heavily researched and annotated book recently, I was horribly frustrated not to be able to follow up on the authors' research as I went along. All at once I was disappointed by the physical reality of the book. I wanted that glued-together stack of paper I was cradling with my free hand on the subway to show me some blue text and display for a moment the grounding evidence behind the litany of fantastic assertions I was reading.

That, of course, is precisely what hypertext offers. And now that experience, and readers' reactions to it over time, have become a new and permanent layer of meaning, part of the art of writing.