Rhetoric of Hypertext Writing – ENGL 382, Fall 2012

  1. Psychological rules of thumb about attention
  2. Chunks:Growing from an on-line documentation scheme ("Info") when computers had very limited resources, HTML hypertext still tends to be built of many smallish chunks of text (or other media) multiply-interconnected with each other. In addition, HTML mounted on the Web can have connections to documents housed anywhere in the connected world and so is in principle boundless. One can of course follow a reference on a printed page to other pages in other texts which in turn can point to other texts, and those texts to yet others, in a similar cascade, but on the Web these potential journeys are not time-consuming but virtually instantaneous. One may well weigh the cost of travelling to a library to follow out a reference, but an on-line reference is only a click away and is available 24/7/365.

    The limitation of even a large display screen makes brevity a great virtue. While one certainly can put all one's document into a single 'page' and expect viewers to scroll through it, or navigate internally within in the page by internal href links, when we encounter such long pages, we strongly suspect the document was converted from a printed text. The long scrolling page is the standard format for academic articles, whether they are co-published in print or not. But academic articles are usually assigned reading (if only self-assigned); when viewing out of curiosity, most people don't like to scroll the display and the writer is well advised to limit the page-chunks to about 1 1/2 screensful each. This means that more linking of pages is required. Of course, one can simply put a [next] link at the end of one page to the top of the following one, and if that is the only link on the page, there is little chance of getting lost or losing one's place in the document, especially since retracing with the Back button is always possible as long as pages are loading into the same tab (which is the default).

  3. Optimal vs. multi paths: One of the five canons of Classical rhetoric was dispositio – the ordering of a discourse. Deciding on an optimal order of presentation was crucial for ancient oratory, which was of course orally delivered; listening to a speech, one cannot mark passages or insert to flip back to to review a point. As writing replaced oratory, being able to read a discourse in print did ease this rather severe limitation of channel, and to some degree invited and facilitated excursions from the main line of argument via footnotes and cross references. There are two main kinds of side-tours in a printed document: footnotes and cross-references to other pages in the document. Footnotes themselves can provide several different kinds of information: a citation, a gloss, or a commentary and discussion on the term at hand. Writers are usually advised to minimize footnotes (or endnotes) that are more than a citation or the gloss of a term for fear of distracting/drawing the reader away from the point under discussion in the main body of the text. In many kinds of discourses, we want the viewer to stay on track, and that means having a well-defined and optimal track/path (i.e., sequence) in mind. Every writer has experienced the struggle to decide how the material should be put in order to maximize the effect the writer wants to have on the viewer. Often saying everything there is to be said on a particular topic would distract from the line of argument; cross-references are ways of saying 'more will be said on this concept/term later/earlier.' So in a sense the cross-reference does not want you to take it, but if it is implemented as a href link, it is very easy to take it, and it is very well established that viewers scan a page for links, sometimes before they read the whole page. So cross-reference links should be used somewhat warily. They are like squirrels to distract the casual viewer. It is ironic that the father of the concept of hypertext (Ted Nelson) was himself a long term sufferer of ADD.

    Along these same lines, the original, default – and for a while nearly the only – way to present an (unvisited) ANCHOR HREF link was to color it blue and underline it. One of the earliest CSS features was the ability to suppress the underlining of ANCHORS ("text-decoration:none"). Many felt that the blue and underlining made links entirely too prominent and fed the eagerness to click in unwholesome ways. Typographers and designers hated to be held to the clunky layout and single color of blue, even though it was indisputably a standard and universally recognized, and so the practices of indicating a link in some other way, or in advanced cases, not indicating it at all in, say, menu lists, where it can be assumed that all items are links.

    HREF-linked chunks of text can be written according to this plan of a single, main, preferred line of march, and many have, with a rather sparing use of links in each chunk. This is the best way to mount a narrative sequence or a sustained line of argument or instruction, where the order is crucial and carefully worked out. None of these work well if the component pages are viewed in any of several possible orders. In fact, it has been noted that a sequence of pages with links only to the next page (a "tunnel") restricts the viewer even more than a printed page does, since no skipping is possible. Sometimes brief side-excursion links are added, turning a tunnel into an axial structure like a hotel corridor. But whether tunnel or corridor, the main line of march is clear and it is also clear when you have passed each of the way-stations and arrived at the final page. If the movement of a viewer through a site is called a traversal, then these optimized sites are traversals as journeys.

    Following these two basic structures, Lynch and Horton describe what is probably the most common organization of a site which is into hierarchical divisions and subdivisions – the sort of thing that can be traditionally outlined with either Roman and letters and numbers or the dot-divided 1.3, 1.3.1, 1.3.2 etc. scheme. In a strict hierarchical ("tree") structure, there is no crossing over from one branch to another, and each page ("node") in the site has a unique path from the top page. (Mathematically, it is 'acyclic'.) The simplest of these structures is a simple "star" (LH: hub-and-spoke) pattern of a central ("top") page with several links to pages all one level 'down' from the top. That is, all the "rays/spokes" terminate in pages with no linking to other pages (and so are "dead-end pages"). On the next level of complexity, the rays continue in a string of pages, but still these pages do not cross over to pages on another ray.

    Lynch and Horton point out that the little band of title and greater-than carets that we often see across the top of a page ("breadcrumbs") are ideal for marking the place of the page in a hierarchy.
    These are coded as part of each page and are not dynamic, so they only work reliably if the path to the current page is in fact unique. (If more than one path is possible, which do you code for the page? If you arbitrarily choose one path and it is not the one some viewers actually followed, their attempt to "go back up the tree" will not retrace their steps.)

    Some websites are built for different purpose – not to expound upon a subject or argue for a position on an issue (or sell a product), but to explore an idea or reflect on an experience and its rammifications and connections to other ideas. Not all journeys are business trips; some are for site-seeing. Sites offering a place for exploration are usually worked out in networks with considerable cross linking so that multiple paths can be traversed, and the kind of thinking supported by this style is associative rather than directed toward a single goal or outcome. Such sites are usually called webs. One such site is Kreg Wallace's Miller Walks.

    Since webbed sites with cross linking allow the viewer to come to a page from more than one direction, web-composing eases off the effort to choose 'one best context' for each page. It sometimes happens that the first few times a viewer arrives at a page they have already visited, they are disappointed or even disoriented, especially if they are trying to function as efficient information seeking machines. And when that happens, viewers may thinks they have seen all there is to see on the site. Sites with a good bit of cross linking ("webby") are usually provided with considerable navigational assistence (an overall map, a 'return to Home' link on each page, etc.); lacking these, the viewer may experience the site as a tangled maze.

    OTOH, coming at a page from two different directions could result in its being seen different ways – as being a meaningful node in more than one thread, a knot of signification, if you will. An enjoyment of this kind of experience is basic to the enjoyment of literature.

    Reading: Locke Carter, "Arguments in Hypertext: A Rhetorical Approach"

  4. Links as Choice Points: With hypertext, the immediate context leading to a page is the link, and considerable attention has been given to the phrasing of an informative link. For example, a general purpose here link is often criticized for being almost vacuous. It gives no clue as to what will come up ("where you will jump to") if you click on it and thus no guidance on whether to do it. Indeed, the HREF link triggers a wide array of outcomes ranging from a jump to another part of the same page, to another page on the same site, a page on an entirely different server, or a secondary window, and a variety of other javascript events. A few devices have been developed to narrow this wide range of uncertainty.

    1. link-typing icons:

      Links to pages on another site are the most consequential; they are the basis of Ted Nelson 'Docuverse' of everything linked eventually to everything else, and they are (usually) links to pages the current writer does not control. They have not been used as liberally as they used to be because we have come to realize that URLs change a lot and without notice. If not found and repaired or deleted, they become the dead links that are unfortunately common. The marker of an off-site link is usually There is also a fancier animated version of this icon that pokes out the arrow and can be used when the link is hovered over. Another much less common one is a little blue world globe (and a version that rotates when the link is moused over) . And there is an elegant little conditional CSS gambit that will insert these automatically if the URL of the link begins HTTP. (see the stylesheet for the calendar page of this site for a sample installation). You can use these to mark links to .pdf files, popup windows, , accordion-drop 'more' sections – all these can be coded by for each item, or automatically via CSS.

    2. Reading: Venkatesh Rao, The Rhetoric of the Hyperlink .

    3. tooltips (and marking of same):

      A relatively short message about not only a link but any element can be included as the value of the "title" attribute and will appear immediately over the cursor when the cursor hovers over the element. Support for this feature developed rather slowly in browsers over the last decade and now is reliably built in to all the major ones. Because it is not widely known, hovering over elements waiting for a tooltip to pop up is not often done, and suggestions have been made that this very handy little feature would be more widely used if it were clearly marked. Some have suggested dotted underlining (which may be too much of a shout-out); another option would be perhaps a little note icon or even just a little 'info' icon (). The only problem is that there is no standardization on the meanings of these icons. An alternative suggestion has been to use the CSS cursor-changing code to turn the pointer to a question mark if the viewer mouses over the item. As as in this phrase. Of course, that won't work unless you pass the cursor over the word. I suppose that viewers could acquire the knack for doing that. Obviously you can do lots of other things with a tooltip than just indicate the type of href link, such as provide a gloss on the item, identify a source, add a brief comment, and so on. There are more elaborate implementations using JQuery and other Javascript, but the lowly out-of-the-box tooltip is a nifty and underused device. [BTW--I have implemented the dotted underline for a link with title in the style sheet for this suite of pages: a[title] {border-bottom-style:dotted}. In fact, it is implemented for all title attributes, even those on other elements. (=*[title])

    4. ezLinkPreview (Chrome) and CoolPreview (FireFox) Extensions:.

      Much has been made of the paralyzing effect of too many links on a page. Marie Laure Ryan, for example, quotes Michael Heim's observation that "Hypertext thinking may indeed reveal something about us that is agitated, panicky, or even pathological. As the mind jumps, the psyche gets jumpy or hyper" (Narrative as Virtual Reality, 261). But some of this anxiety of choice can be reduced for viewers who install the exLinkPreview extension in Chrome or the CoolPreview extension for FireFox. These allow you the viewer to pop open a floating small browser window into which the link site loads, so that we can preview without "leaving" the sending page.

      True, not many people have installed these extensions, and many others still rely on Safari or some version of Exploder, so you can't be sure your viewers will have this extension installed and use it. But try installing it and see if it changes how you feel about and how you use href links.