APPLIED COGNITIVE PSYCHOLOGY, VOL. 2, 87-95 (1988)

The Use of Computer Displays to Improve Reading Comprehension

JAMES F. JUOLA Department of Psychology, Fraser Hall, University of Kansas, Lawrence,

Kansas, 66045, U.S.A.

SUMMARY There are a number of different ways to present text on computer-controlled displays, and these vary in their relative readabilities. ‘Electronic pages’ can be made to be almost as readable as high-quality printed pages. Alternative text display formats can make use of some of the flexibility inherent in computers-for example, by presenting text that appears to move upwards from the bottom of the screen or from right to left along a single line. Research has shown, however, that the most viable alternative to the standard page format, both as a tool for the study of reading processes and as a means to promote reading efficiency, is the rapid, serial visual presentation (RSVP) technique. In RSVP the text is divided into small units of one or a few words that are presented successively for brief durations to a common location on a screen. Text presented in this way can be read rapidly and with good comprehension, while the need for eye movements is eliminated. The data from several studies indicate that RSVP can actually improve the comprehension abilities of less-skilled readers. The present paper explores reasons for this apparent advantage of RSVP for some readers, and proposes ways in which the method might be applied in instructional settings.

An increasing percentage of what we read at home, at work and in schools will be presented on CRTs or other electronic display devices. For this reason alone it is important to learn how best to present text on such displays in order to promote rapid reading for comprehension. Other reasons for conducting research on computer-displayed text include: (1) novel methods of text presentation offer new ways to evaluate theories of reading processes in that unique and complete control can be gained over which aspects of text are viewed and the length of time they are available for inspection, (2) optimal methods of text presentation have been developed that actbaily are improvements over printed pages, at least for some readers and task environments, and (3) some presentation methods show promise as comprehension aids for less-skilled readers and could be incorporated into reading curricula for beginning and disabled readers.

A common method of displaying text on a computer screen is to present paragraphs of 200 to 300 words as single ‘pages’ that fill the display screen and can be read with normal eye movements. Electronic pages can generally be read about as well as printed pages, although at equivalent levels of comprehension they are read a bit more slowly (Kruk and Muter, 1984; Muter, Latremouille, Treurniet and Beam, 1982). Displays under computer control offer a number of other possibilities, and some of these have only recently been tested in laboratory

0888-4080/88/010087-09$05.00 0 1988 by John Wiley & Sons, Ltd.

Received 2 February 1987 Revised 2 June 1987

88 J . F. Juola

settings. For example, text can be moved more or less continuously up from the bottom of the screen by lines (‘scrolling’) or from the right to left along a single line (‘leading’). Both scrolling (Kolers, Duchnicky and Ferguson, 1981; Oleron and Tardieu, 1978) and leading (Granaas, McKay, Laham, Hurt and Juola, 1984; Sekey and Tietz, 1982) have been shown to be generally inferior to standard pages in studies of reading speed and comprehension. This result is not surprising in that, in reading moving text, normal eye movement patterns must be replaced with those requiring anticipations and compensations for the movement of the text itself.

A more promising method of presenting text on computer screens is one that eliminates the need for eye movements altogether. This can be done by using RSVP, or rapid serial visual presentation, in which successive text segments of one or a few words are presented to a common location on the display (e.g. Bouma and deVoogd, 1974; Forster, 1970; Gilbert, 1959; Raygor, 1974). In this way it is possible to simulate a succession of typical eye fixations in normal reading by showing a new, short segment of text every 250 msec or so. Most studies of RSVP reading have used display conditions that approximate the results of typical eye movement and fixation patterns and have found that RSVP reading performance is no worse, and sometimes better, than that obtained for normal pages (Bouma and deVoogd, 1974; Juola, Ward and McNamara, 1982; Potter, Kroll and Harris, 1980; Raygor, 1974; Ward and Juola, 1982).

At first it might appear counterintuitive that the elimination of normal eye movements could in any way be conducive to reading efficiency. Normal deviations from invariant scanning patterns and fixation durations have been argued to reflect changes in processing load or decisions about how to sample information from text (e.g. Just and Carpenter, 1980; Rayner and Duffy, 1987; Levy-Schoen and O’Regan, 1979; McConkie, 1979). Thus variations in saccadic extent and fixation duration are thought to be necessary concomitants of normal reading processes, much as are regressive eye movements and pauses, none of which are present in the methodical RSVP format. Because most previous research is consistent in showing that reading comprehension is no worse for RSVP displays than for normal pages, any cost due to the invariance of visual parameters and surrender of control in RSVP must somehow be compensated by benefits due to the elimination of the results of normal eye movements.’

Are there possible inefficiencies, information losses, and attentional demands resulting from the normal saccadic control system which might be reduced in the methodical RSVP format? An analysis of what might entail from the elimination of the need for eye movements suggests that at least some benefits might be expected. For example, it is possible that typical eye movements are inefficient because readers normally tend to be conservative, moving their eyes forward in smaller saccades and in more frequent regressive movements than are necessary for clear resolution of the text (Crowder, 1982; Hochberg, 1976). Eye movements themselves sometimes result in inaccurate fixation locations, particularly in return sweeps to new lines of text, necessitating corrective saccades (McConkie, 1983).

It is known that vision during eye movements is suppressed (e.g. Volkmann, 1976), and this suppression extends into the time of adjacent fixations. Although saccadic suppression might not be an important factor in normal reading (Rayner, Inhoff, Morrison, Slowiaczek and Bertera, 1981; Wolverton and Zola, 1983), Breitmeyer (1983) and Kolers (1983) argue for its existence while admitting that it

Computer Displays to Improve Reading Comprehension 89

might be weaker in above-threshold reading environments than in the threshold detection studies in which it was first investigated. In any case, frequent eye movements increase the total amount of visual information potentially lost through saccadic suppression.

A final potential problem for normal reading lies in the fact that, although the eye guidance system subjectively appears to be automatic, some cognitive capacity unavoidably must be devoted to the decisions of where and when to move the eyes. This demand for cognitive resources could diminish the capacity available for other processes related to the main task of comprehension. Thus, at least the potential exists for discovering methods of text presentation that eliminate the need for eye movements while also improving the efficiency of reading for comprehension. To the extent that such factors as conservative or inaccurate eye movements, saccadic suppression, and division of attention contribute disproportionately to certain types of reading disabilities, the potential increases that RSVP can contribute to the improvement in reading comprehension of less-able readers.

OPTIMIZING RSVP READING

Given that most studies comparing RSVP and page formats have found them to be nearly equivalent in readability, it is of interest to determine whether an optimal RSVP method can be found that might result in improved reading performance relative to pages. In a series of experiments we have sought to use eye movement data and theories of the cognitive processes involved in reading to guide our selection of the parameters of RSVP displays. These parameters include (1) the number of characters or words shown at one time (window size), (2) the amount of overlapping text between successive windows, (3) the amount of time that each successive window is shown (window duration), and (4) the degree and type of text structure used in window selection. In most of our research, each window includes whole words only, as window size is entered as a parameter in a text parsing program which selects successive segments that are as close as possible to the desired length (in number of character spaces) without truncating any words. We have also determined that the windows are more readable if they are left-justified, so that only the right-hand margin appears ragged in successive displays, and the text itself is more readable if blank windows are inserted between sentences.

We know that the field of useful vision in normal reading extends about 20 characters horizontally around the point of fixation, and most of this field extends to the right of fixation. However, only the leftmost, or fovea1,part of this region is seen clearly enough to identify words; the rightmost part of the field could be used for pre-analyzing words that will appear closer to the fovea in the following fixation or for guiding the saccadic movement to an appropriate place for that fixation (McConkie, 1979; McConkie and Rayner, 1975; Rayner, 1978). Most previous studies of RSVP have used successive windows of single words (e.g. Forster, 1970; Masson, 1983; Potter et al., 1980). This procedure has the dual limitation of under- utilizing the perceptual information processing capacity of the visual system and also of requiring a rapid succession of displays in order to achieve normal reading speeds (i.e. one word every 200 msec equals 300 words per minute, WPM).

In an experiment designed to determine the optimal window size for RSVP

90 J. F. Juola

displays (Cocklin, Ward, Chen and Juola, 1984, Exp. 11), the mean number of characters per window was varied from five to 20, and mean window duration varied from 76 to 992 msec to produce imposed reading rates ranging from 200 to 800 WPM. In most conditions, comprehension performance, as measured by ability to answer questions immediately after reading the paragraphs, was superior for windows of intermediate size. A best-fitting quadratic function fit to the comprehension by window size function had a peak between 12 and 13 character spaces. The quadratic trend accounted for a significant proportion of the variance in comprehension scores while failing to interact with text difficulty and reading rate. The peak performance for windows of about two to three words in length is an interesting corroboration of estimates of the functional visual field based on eye movement data, and it also points to the self-imposed limitation of most previous RSVP research that made use of one-word windows.

Other design parameters chosen to simulate normal reading have been less successful in improving performance with RSVP. For example, the inclusion of overlapping text segments on successive windows in RSVP, in order to mimic successive parafoveal and foveal views of text elements that occur in normal reading, resulted in lower levels of comprehension than those obtained for windows with no overlaps when progress through the text was controlled at the same overall rate (Cocklin et al., 1984, Exp. I). In another study, varying window duration in order to correspond to the expected processing time per word derived from Just and Carpenter’s (1980) regression model produced no advantage over a constant window duration condition (Ward and Juola, 1982). These results suggest that optimal RSVP conditions do not necessarily need to mimic eye movement data, and that, in fact, executing a series of fixations and saccades might not be the best way to read.

Besides examining the effects of window size, successive overlaps, and variations in window duration on the readability of RSVP text, we have tested the effects of window structure. In a third experiment reported in Cocklin et al. (1984), we contrasted comprehension scores for windows chosen to be as close as possible to a constant size of 13 characters with those selected to contain short ‘idea units’ that also averaged about 13 spaces in length. Idea units were defined by consensus among four independent judges who divided the paragraphs into short groups of words that belonged together to express a single idea. The results demonstrated a clear advantage for idea-unit segmentation, despite the fact that window length was four times as variable for idea units as for the arbitrary segment condition. Subsequent research has utilized a parsing program developed by Granaas (1985) that makes use of punctuation, a list of critical function words, and an upper bound on segment length (i.e. about 20 character spaces) to produce windows for RSVP display that have about a 90 per cent agreement with those selected by human judges.

Based on research with college student readers using a wide range of text difficulties and presentation rates, we have concluded that the optimal RSVP conditions are those which include windows averaging 12 to 13 character spaces in length that are selected to include short phrases or idea units. Presentation rate can be controlled by varying window duration, with 500 msec durations producing a reading rate of about 300 WPM. Comprehension can also be improved by inserting a blank window between sentences, presumably to enable the same kind of

Computer Displays to Improve Reading Comprehension 91

sentence wrap-up and integration processes that occur in normal reading (see Just and Carpenter, 1980; Masson, 1983).

INDIVIDUAL DIFFERENCES IN ABILITY AND READING RSVP TEXT

In a recent review of RSVP research? Potter (1983) remarked that the method has not been tested adequately with practised subjects or with readers younger and less skilled than college students. Further, although the method seems to hold promise as a technique for assisting readers with certain types of disabilities? little attention had been given to individual differences in ability and the relative efficacy of the RSVP format. Two studies will be reviewed in this section that address the concerns raised by Potter.

Chen (1986) reported a study in which college student readers were divided into high-span and low-span groups on the basis of Daneman and Carpenter’s (1984) procedure for measuring working memory capacity while reading. A number of paragraphs were presented for the same total amount of viewing time, either as intact pages that could be read directly from the screen with normal eye movements or in the RSVP format with windows averaging about 12 character spaces in length. Although the windows in the RSVP condition were about of optimal length, they were not linguistically parsed. Comprehension and memory of the texts were assessed by a sentence recognition task.

Chen’s results for the high-span group replicated earlier findings (e.g. Juola et al . , 1982) in showing no significant differences in reading comprehension between the RSVP and page formats. For the low-span group, however, there was a significant advantage for RSVP displays. This result is suggestive for the usefulness of RSVP displays in promoting the comprehension abilities of less-able readers, even among as select a group as university students.

A more recent study by Juola, Haugh, Trast, Ferraro and Liebhaber (1987) extends Chen’s (1986) results to a group of younger and less-skilled readers. The subjects were 19 students ranging in age from 12 to 15 years who were selected to include 10 readers of average ability and nine students who were decidedly below average in reading ability. The students were selected on the basis of standardized tests of reading comprehension administered in their schools.

The students participated in 5 hours of testing distributed over a week in which they read passages from a standard reading comprehension test appropriate to their chronological ages. On the first day the passages and comprehension questions were presented on typed sheets, and the subjects were allowed to proceed at their own pace. The day ended with the presentation of 10 paragraphs presented at 200 WPM in the RSVP format on a laboratory computer. The windows were approximately 13 character spaces in length, on average, and were parsed according to the procedures outlined by Granaas (1985). On days two through five all subjects read passages presented in the RSVP mode, and the second and fifth days also included some paragraphs shown as standard pages on the computer screen in order to compare comprehension performance in the two formats. The results are shown in Table 1.

In general, the data replicate Chen’s results, in that the relative readability of normal page and RSVP formats interacts with a reader’s ability level. Readers of

92 J. F. Juola

Table 1 . Mean reading rates (words per minute), comprehension levels (percentage correct answers to questions), and reading efficiency scores (rate x proportion correct) for average and below-average readers for texts read from printed pages, from computer displays using a page format, and from computer displays using the RSVP format (after 5 days of practice; see Juola et af., 1987)

Average readers Below-average readers

Printed page WPM 180 154 Percentage correct 74 60 Efficiency 133 92

Electronic page WPM Percentage correct Efficiency

RSVP WPM Percentage correct Efficiency

200 200 68 45

136 90

200 200

132 105 66 52.5

average ability demonstrated equivalent levels of reading efficiency (reading rate multiplied by proportion correct comprehension score) for typed pages, electronic pages, and RSVP. The poorer readers, however, demonstrated a distinct advantage for RSVP when given five days of practice with the novel format. Apparently there are some benefits to being paced through a text presented in small segments that reflect some degree of linguistic structure, at least for readers of less than average reading skill.

CONCLUSIONS

There are a variety of ways to present text on a computer screen such that it can be read rapidly and with good comprehension. Both electronic pages and the RSVP method result in readability levels that approximate those obtained for printed pages, whereas other text formats, such as those that present text that appears to move across the screen, generally result in lower levels of reading speed, comprehension, or both. Research at the University of Kansas for the past 5 years has centred on optimization of the RSVP method of displaying text. We have demonstrated that the optimal RSVP format consists of successive one- to five- word segments of text displayed at a common location on the screen. Com- prehension levels are highest when these segments are chosen to reflect, as much as is practically possible, syntactic structures in the text, and a pause or blank window is presented between sentences.

It is somewhat surprising that our research has found no significant differences in comprehension levels for skilled readers between an optimal RSVP format and electronic pages. Previous research using what we consider to be non-optimal presentation schemes has indicated that at least some aspects of reading

Computer Displays to Improve Reading Comprehension 93

comprehension are suprior for RSVP relative to pages (e.g. Bouma and devoogd, 1974; Potter et al., 1980; Raygor, 1974). Recent studies by Chen (1986) and by Juola et al. (1987), using improved RSVP formats, have demonstrated that skilled readers are equally capable of reading page and RSVP displays. However, less- skilled readers at both college and middle-school levels show a comprehension advantage for RSVP.

There are several reasons for expecting an RSVP advantage over pages, especially for less-skilled readers. These include the substitution of a regular and complete sequence through the text for any irregular and inefficient sampling that results from the distribution of eye movements over a normal page (see Kliegl and Olson, 1986). Second, it is known that dividing a normal text into subunits that reflect phrase structures can improve reading comprehension for less-skilled readers, whereas skilled adult readers are typically indifferent to such parsing aids (see Carver, 1970; Martinez, Ghatala and Bell, 1980; O’Shea and Sindelar, 1983). Thus, less-skilled readers could benefit from RSVP due to improvements in both peripheral processes involved in sampling visual information from the text and central processes involved in parsing and integrating higher-order text units.

If the advantage for RSVP is most clearly demonstrated for younger and less- skilled readers, then it follows that the method could result in improved methods of instruction for beginning readers. The presentation of short text segments that can be read without eye movements would automatically direct attention to the appropriate text unit in its proper order, and eliminate the need to reserve attentional capacity for decisions about where and when next to move the eyes. Parsing text into meaningful subunits would also emphasize relations among words that form the basis of sentence comprehension. Finally, reading instruction begins with the affirmation that the visual code represents the speech code at several levels of analysis. Oral reading in groups, and following along a line of text while it is being read by another person, is at the heart of many beginning reading lessons. Presentation of audio versions of text segments and their RSVP counterparts insures a simultaneity of occurrence that is impossible to guarantee using other procedures, thus maximizing the benefit of cross-modal matching of linguistic inputs in reading instruction. The beginning reader, and those in need of remedial training, could benefit from RSVP by exploiting its inherent readability as well as its ease in matching the sequential aspects of spoken language.

ACKNOWLEDGEMENTS

Preparation of this paper was assisted by a grant from the General Research Fund of the University of Kansas. Part of the work was done while the author was a visiting research fellow at the Institute for Perception Research (IPO) in Eindhoven, The Netherlands, and the recipient of a Fulbright Senior Research Award. Helpful comments from Don Bouwhuis, Sonia Ann Juoia, and two anonymous reviewers on an earlier version of this manuscript are gratefully acknowledged.

94 J . F. Juolu

NOTE

1. It is perhaps an interesting conjecture that many of the studies that have found predictable variations in fixation pauses, saccadic extents, and regressive eye movements have used texts designed to promote such variants from steady, progressive reading. The literature is rife with short passages littered with infrequent technical terms, ambiguous words in garden-path contexts, and vague anaphoric referents. Undoubtedly there exist prose forms that would be difficult to read in the RSVP format, but such generalizations should probably not be made so freely to normal well-written, unambiguous text.

REFERENCES

Bouma, H. and deVoogd, A. H. (1974). On the control of eye saccades in reading. Vision Research, 14, 273-284.

Breitmeyer, B. G. (1983). Sensory masking, persistence, and enhancement in visual exploration and reading. In K. Rayner (Ed.), Eye movements in reading. New York: Academic Press.

Carver, R. P. (1970). Effect of a ‘chunked’ typography on reading rate and comprehension. Journal of Applied Psychology, 54, 288-296.

Chen, H.-C. (1986). Effects of reading span and textual coherence on rapid-sequential reading. Memory and Cognition, 14, 202-208.

Cocklin, T. G., Ward, N. J., Chen, H.-C. and Juola, J. F. (1984). Factors influencing readability of rapidly presented text segments. Memory and Cognition, 12, 431-442.

Crowder, R. G. (1982). The psychology of reading: an introduction. New York: Oxford University Press.

Daneman, M. and Carpenter, P. A. (1983). Individual differences in working memory and reading. Journal of Verbal Learning and Verbal Behavior, 19, 450-466.

Forster, K. I. (1970). Visual perception of rapidly presented word sequences of varying complexity. Perception and Psychophysics, 8, 215-221.

Gilbert, L. C. (1959). Speed of processing verbal stimuli and its relation to reading. Journal of Educational Psychology, 55, 8-14.

Granaas, M. M. (1985). Simple, applied text parsing. Behavior Research Methods, Instruments, and Computers, 17, 209-216.

Granaas, M. M., McKay, T. D., Laham, R. D., Hurt, L. D. and Juola, J . F. (1984). Reading moving text on a CRT screen. Human Factors, 26, 97-104.

Hochberg, J. (1976). Toward a speech-plan eye-movement model of reading. In R. A. Monty and J. W. Senders (Eds), Eye movements and psychological processes. Hillsdale, NJ: Erlbaum.

Juola, J. F., Ward, N. J. and McNamara, T. (1982). Visual search and reading of rapid, serial presentations of letter strings, words, and text. Journal of Experimental Psychology: General, 111, 208-227.

Juola, J. F., Haugh, D., Trast, S . , Ferraro, F. R. and Liebhaber, M. (1987). Reading with and without eye movements. In J. K. O’Regan and A. Levy-Schoen (Eds), Eye movements: from physiology to cognition. Amsterdam: Elsevier.

Just, M. A. and Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological Review, 87, 329-354.

Kliegl, R. and Olson, R. K. (1986). Individual differences in disabled and normal readers’ eye movements. Paper presented at the International Conference on Cognitive Approaches to Reading, Leicester, England.

Kolers, P. A. (1983). Locations and context in eye movements. In K. Rayner (Ed.), Eye movements in reading. New York: Academic Press.

Kolers, P. A., Duchnicky, R. L. and Ferguson, D. C. (1981). Eye movement measurement of readability of CRT displays. Human Factors, 23, 517-527.

Computer Displays to Improve Reading Comprehension 95

Kruk, R. S. and Muter, P. (1984). Reading of continuous text on video screens. Human Factors, 26, 339-345.

Levy-Schoen, A. and O’Regan, K. (1979). The control of eye movements in reading. In P. A. Kolers, M. E. Wrolstad and H. Bouma (Eds), Processing of visible language, Vol. 1. New York: Plenum.

Martinez, P., Ghatala, E. S. and Bell, J. A. (1980). Size of processing unit during reading and retention of prose by good and poor readers. Journal of Reading Behavior, 9,89-95.

Masson, M. E. J. (1983). Conceptual processing of text during skimming and rapid sequential reading. Memory and Cognition, 11, 262-274.

McConkie, G. W. (1979). On the role and control of eye movements in reading. In P. A. Kolers, M. E. Wrolstad and H. Bouma (Eds), Processing of visible language, Vol. 1. New York: Plenum.

McConkie, G. W. and Rayner, K. (1975). The span of the effective stimulus during a fixation in reading. Perception and Psychophysics, 17, 57&586.

Muter, P., Latremouille, S. A., Treurniet, W. C. and Beam, P. (1982). Extended reading of continuous text on television screens. Human Factors, 24, 501-508.

Oleron, G. and Tardieu, H. (1978). Influence of scrolling upon the recall of texts. In M. M. Gruneberg, P. E. Morris and R. N. Sykes (Eds), Practical aspects of memory. London: Academic Press.

O’Shea, L. J . and Sindelar, P. I . (1983). The effects of segmenting written discourse on the reading comprehension of low- and high-performance readers. Reading Research Quarterly, 18, 458-465.

Potter, M. C., Kroll, J. F. and Harris, C. (1980). Comprehension and memory in rapid- sequential reading. In R. Nickerson (Ed.), Attention and performance, VZZZ. Hillsdale, NJ: Erlbaum.

Raygor, R. (1974). An investigation of the relationship between eye movements and comprehension in fluent reading. Unpublished doctoral dissertation, University of Minnesota, Minneapolis.

Rayner, K. (1978). Eye movements in reading and information processing. Psychological Bulletin, 85, 618-660.

Rayner, K. and Duffy, S. A. (1987). Eye movements and lexical ambiguity. In J. K. O’Regan and A. Levy-Schoen (Eds), Eye movements: from physiology to cognition. Amsterdam: Elsevier.

Rayner, K., Inhoff, A. W., Morrison, R. E., Slowiaczek, M. L. andBertera, J. H. (1981). Masking of foveal and parafoveal vision during eye fixations in reading. Journal of Experimental Psychology: Human Perception and Performance, 7 , 167-179.

Sekey, A. and Tietz, J. (1982). Text display by ‘saccadic scrolling’. Visible Language, 16,62- 76.

Volkmann, F. C. (1976). Saccadic suppression: a brief review. In R. A. Monty and J. W. Senders (Eds), Eye movements and psychological processes. Hillsdale, NJ: Erlbaum.

Ward, N. J. and Juola, J. F. (1982). Reading with and without eye movements: a reply to Just, Carpenter, and Woolley. Journal of Experimental Psychology: General, 111, 239- 241.

Wolverton, G. S. and Zola, D. (1983). The temporal characteristics of visual information extraction during reading. In K. Rayner (Ed.), Eye movements in reading. New York: Academic Press.