Dynamic Text Presentation

Until this point in the present chapter, it has been assumed that presentation of text is static. In dynamic text presentation, an attempt is made to optimize reading by utilizing some of the special capabilities of the computer. Two methods of dynamic text presentation that have been tested are rapid serial visual presentation (RSVP) and the Times Square format.

RSVP

With RSVP, text is presented at a fixed location on the screen, one word at a time or a few words at a time. Several researchers have demonstrated that readers can perform approximately as efficiently with RSVP as with normal page-format reading (e.g., Juola, Ward, & McNamara, 1982). There are several potential uses of RSVP:

• when display space is limited
• scanning and skimming, which constitute a high proportion of cognitive processing of text
• reading by users with impaired peripheral vision, for example, retinitis pigmentosa (Williamson, Muter, & Kruk, 1986)
• studying cognitive processes
• with certain kinds of poor readers, who perform better, after practice, with RSVP than with regular page format (Juola, Haugh, Trast, Ferraro, & Liebhaber, 1987)
• as an efficient way to present continuous text in general, when optimal parameters are established, because there is no need to expend cognitive capacity on controlling eye movements

The optimal conditions for RSVP seem to be the following (Juola, Haugh, Trast, Ferraro, &
Liebhaber, 1987):

• about 12 characters per window on average
• two or three words per window
• idea-unit segmentation
• 250-500 msec blank window between sentences

Giving the user control over RSVP presentation, e.g., over regressions and rate of presentation, sometimes has adverse effects on performance (Chen and Chan, 1990; Muter, Kruk, Buttigieg, & Kang, 1988). But, of course, under many circumstances, people will prefer to have this control.

Times Square Format

Kang & Muter (1989) found that smooth (pixel-by-pixel) horizontal scrolling with a small window (Times Square format) produced performance at least as good as RSVP, contrary to earlier studies which did not use pixel-by-pixel scrolling. In addition, subjects preferred the Times Square format, which is often used in electronic billboards.

Interactions

A major difficulty in research in text presentation, whether static or dynamic, is that the various independent variables often interact, sometimes in extremely complex ways: The effect of one variable depends on the level of other variables. For example, this has been a problem with to respect to typographic variables such as type size, line length, and interline spacing (Frenckner, 1990). Following are some possible approaches for handling the problem of intractable high-order interactions:

• Perform a huge number of factorial experiments. (This is usually not practical.)
• Sample randomly from a large range of several factors.
• Re-construe the problem: "What is sometimes required is not more data or more refined data but a different conception of the problem" (Shepard, 1987, p. 1318).
• Seek "robustness in variation": Try to find evidence for a principle in many relevant settings (Landauer, 1988).
• Prune the alternative space; i.e., somehow reduce the number of possibilities under consideration.
• Use an algorithm such as Simplex (Nelder & Mead, 1965) to determine which values of several independent variables to experimentally test next in order to maximize a dependent variable such as reading rate.
• Use a "kitchen sink" approach for practical problems (Muter & Maurutto, 1991): throw into a single condition every feature that might have a beneficial effect, based on theory or data, and compare it to a control condition.