COMPUTERIZED PRESENTATION OF TEXT
FOR THE VISUALLY HANDICAPPED
NANCY L. WILLIAMSON, PAUL MUTER, and RICHARD S. KRUK
Department of Psychology
University of Toronto
A method of text presentation which has generated a considerable amount of research with normally sighted readers may be of benefit to people suffering from a range of visual impairments. Rapid serial visual presentation, or RSVP, involves the brief display of consecutive words at a fixed location. Because this technique eliminates the need for eye movements, readers who experience difficulties executing these movements would presumably benefit from this type of display. Readers with retinitis pigmentosa, glaucoma, or other medical conditions which limit peripheral vision represent such a population, since peripheral vision is thought to be critical in guiding readers' eye movements (O'Regan, 1981). In this chapter, we will first review the RSVP technique. Then, after describing retinitis pigmentosa, we will report preliminary data which suggest that RSVP may prove beneficial in assisting readers with this condition read more quickly without a concomitant loss of comprehension.
2 DESCRIPTION OF PRESENTATION TECHNIQUE
Rapid serial visual presentation (RSVP) entails the brief presentation of consecutive words, or groups of words, at a common location. Typically, the rate of presentation per word is set by the experimenter, and each word is shown for an equivalent amount of time. In most of the research with RSVP (Bouma & devoogd, 1974; Cocklin, Ward, Chen, & Juola, 1984; Forster, 1970; Juola, Ward, & McNamara, 1982; Masson, 1983; Potter, Kroll, & Harris, 1980; Williamson & Muter, 1985; see Potter, 1984 for a review), the display duration of each word is very brief, as implied by the word "rapid" in the acronym. However, there is nothing about the presentation technique that prohibits the use of slower rates. Regardless of the rate of presentation, the functional relevance of RSVP lies in the fact that the reader is not required to make eye movements. Instead, the words are brought to a single fixation point.
There are several aspects of this presentation technique that are essential in displaying the words most effectively. One critical feature of RSVP is the brief presentation of a blank display window at the end of each sentence. Masson (1983) found no difference in comprehension between page- format presentation and RSVP with sentence-end pauses of 500 ms. However, performance was higher with the page format display when no sentence-end pauses were included in RSVP. Pauses of 1000 ms did not result in any further benefit.
The rate of presentation per word is another factor that influences the readability of RSVP displays. As would be expected, it has been shown that as the presentation rate increases, RSVP reading comprehension steadily declines (Juola et al., 1982; Masson, 1983).
In multiple-word RSVP, the alignment of the words is important. Displays in which English words were vertically aligned, one above the other, yielded worse comprehension performance relative to one-word RSVP, whereas horizontal alignment of multiple-word displays led to comprehension performance equivalent to that obtained with one-word displays (Williamson & Muter, 1985).
The presentation of idea units rather than the strict presentation of 15 characters (on average, three words) in each window has been found to benefit RSVP performance (Cocklin et al., 1984). However, the practical value of this finding is small at present since the availability of text parsers to determine such specific boundaries is limited.
Conversely, there are other factors which seem to have little influence on the effectiveness of RSVP displays. For instance, presenting each word for a duration corresponding to the average length of time spent looking at the words under normal reading conditions was not found to affect RSVP comprehension performance. Equal presentation durations were found to be just as effective as these variable durations (Ward & Juola, 1982).
In addition, RSVP reading has been shown to be possible using a wide variety of display apparatus, ranging from CRT screens (Cocklin et al., 1984; Juola et al., 1982; Masson, 1983; Potter et al., 1980; Williamson & Muter, 1985) and projection screens (Forster, 1970) to cylindrical rollers moving a page of text, with a window of a fixed size being displayed (Bouma & devoogd, 1974).
The role of the size of the RSVP display is, at best, unclear. Cocklin et al. (1984) showed that the optimal reading size for horizontal RSVP displays was 12 characters, a size considered to be on average, between two and three words. This result was conceptually appealing since the visual span of apprehension has been found to be approximately 12 characters (Rayner, 1983). Contrary to this research, Williamson and Muter (1985) found no difference in performance among consistent one-word, two-word and three-word RSVP displays. Since Cocklin et al. assert that this optimal display size is independent of presentation rate or task difficulty, these two factors cannot be used to account for the apparent discrepancy between experiments. In the Cocklin et al. study, only complete words were shown in each consecutive RSVP display, resulting in a variable number of words per display. As Potter (1984) points out, this method introduces an element of chance since "the change of a single letter in the preceding text could double a word's effective viewing time by causing it to be presented alone instead of with another word." (p.92). In the Williamson and Muter study, a consistent number of words was maintained within each display for each of the different RSVP conditions.
Obviously, a direct comparison of the role of number of characters versus number of words per RSVP display is needed. However, if a consistent number of words per window is to be maintained throughout the RSVP reading, then whether the size of the display is one, two, or three words is apparently irrelevant. All lead to equivalent comprehension performance for a given presentation rate. For readers with little peripheral vision, a window size of one word would seem suitable. A larger window would result in more information appearing outside the foveal area, a region unavailable to such readers.
3 CURRENT EVIDENCE REGARDING RSVP VERSUS PAGE-FORMAT READING
There has been a considerable amount of research carried out comparing comprehension performance with RSVP displays to performance with traditional page-format displays presented for the same amount of time (Cocklin et al., 1984; Forster, 1970; Juola et al., 1982; Masson, 1983; Potter et al., 1980; Williamson & Muter, 1985). The majority of these studies have found that readers experience no deficit in comprehension as a result of reading via an RSVP display (Cocklin et al., 1984; Juola et al., 1982; Masson (Experiment 4), 1983; Potter et al., 1980). Given practice with both of the display techniques, readers still reach comparable comprehension levels under the two types of displays (Cocklin et al., 1984). In most of these studies, however, type of display was confounded with presence of pacing. The RSVP technique typically includes computer pacing, whereas in page-format reading, no such externally imposed pacing is present. Perhaps there is some advantage gained with a source of external pacing that compensates for any loss experienced with other features of the RSVP technique. However, Williamson and Muter (1985) found that even when a source of external pacing was provided in both a page-format and RSVP condition, subjects' comprehension of text presented via an RSVP display was at least equivalent to comprehension of traditionally displayed text.
Some qualification to the finding of RSVP and page-format comprehension equivalence is warranted. Cocklin et al. (1984), Juola et al. (1982), Masson (Experiment 4, 1983), Potter et al. (1980), and Williamson and Muter (1985), found no impairment when comprehension was assessed by performance in answering questions. However, Potter et al. found that comprehension as measured by verbatim recall was worse with RSVP as compared to page-format display. Thus, the results of these RSVP studies appear to be somewhat task-dependent, but as long as the reader is not trying to memorize text, the RSVP technique leads to comprehension equivalent to that obtained with page-format display.
It is thought that the effectiveness of the RSVP display lies in the fact that, compared to page-format reading, readers do not have to use valuable processing capacity and time in planning when and where to fixate next (Cocklin et al., 1984). If the Rayner, Slowiaczek, Clifton, and Bertera (1983) finding is correct, that the planning and execution of an eye movement from one location to another takes, on average, 175-200 ms, then elimination of this process in reading using an RSVP display may result in much more time being available for processing of the words themselves. This advantage may offset any disadvantage entailed in being able to see only one word at a time at a fixed rate.
4 RELEVANCE TO VISUALLY IMPAIRED READERS
In normally sighted readers, eye movements from one word to the next during reading appear to be controlled by the length of the next word or two to the right of the word currently being fixated (O'Regan, 1981). Parafoveal information was found to be useful in guiding eye movements from one location to the next in an examination of the effects of restricting foveal and parafoveal information in reading (Rayner & Bertera, 1979). If this information is unavailable, as it is to some visually impaired individuals, then presumably inefficient eye movements occur, the reader loses his or her location on the page, and an unusually large number of regressions and disruptions occur. Since RSVP eliminates the need for readers to move their eyes, it would seem that readers to whom parafoveal information is unavailable or limited would benefit from this presentation technique.
5 RETINITIS PIGMENTOSA
The progressive loss of peripheral vision is one of the primary symptoms of retinitis pigmentosa, a slow bilateral degenerative disorder of the retina. This disease is generally genetically determined. Its onset usually begins in childhood, often resulting in total blindness by middle or old age. The degenerative effect begins near the equator of the eye and gradually spreads both anteriorly and posteriorly, primarily affecting the rods.
Night blindness is the first detectable symptom of retinitis pigmentosa. Dark adaptation ability is attenuated because of the degeneration of the rods, which are primarily responsible for vision under low illumination.
Later, a progressive constriction of the visual field produces a constantly decreasing tube of vision, known as tunnel vision. The concentric contraction of the field of vision is especially marked under reduced illumination. The degenerative process extends outward toward the peripheral areas, and more slowly inward toward the foveal area in the later course of the disease. Thus, visual acuity is preserved until later stages of the disease. Fifty percent of all patients have an acuity better than 20/50, but by age 50 at least half of patients have a visual acuity of 20/200 or worse (Marmor, 1980). However, rate of loss of central vision, and acuity, varies according to the severity of the disorder (Marmor, 1980).
In comparison with normally sighted individuals whose visual fields extend through 180 degrees, the restricted visual field of a retinitis pigmentosa patient, in later stages, extends about two to three degrees around the fixation point (Marmor, 1980). For further details on retinitis pigmentosa, see Duke-Elder (1970), Merin and Auerback (1980), and Trevor-Roper and Curran (1984).
6 READING AIDS
People with retinitis pigmentosa have good reading vision until the condition reaches an advanced stage. However, with the loss of peripheral information to help guide their eye movements, people with retinitis pigmentosa would be expected to read continuous text with less than optimal efficiency. This is the general observation of clinicians working with these patients. Patients frequently lose track of their place on the page as they read.
A number of reading aids are available for people with retinitis pigmentosa. Magnifiers are helpful in mild cases. The patient is taught to focus on a single fixation point and move the page rather than the eyes while reading continuous text. Other aids include closed circuit television, which is a form of text magnification (Harley & Lawrence, 1977), and reversed telescopes, which increase the number of text characters seen in the patient's limited field of vision. The reading aids described here involve the reader manually manipulating the text page to bring the words into the reader's visual field. RSVP, on the other hand, performs this manipulation automatically.
7 PRELIMINARY INVESTIGATION
A female retinitis pigmentosa patient volunteered to participate in an exploratory experiment. Since she has 10 degrees of vision, she may be classified as suffering from a moderately severe form of the disease. She reported that she read frequently, both at work and for pleasure.
The experiment involved reading 20 self-contained paragraphs, with an average length of 124 words. Ten short-answer questions had been developed for each of these paragraphs and these questions had been found to be a sensitive measure of reading comprehension (Williamson & Muter, 1985). The paragraphs were randomly assigned to one of ten pairs and the method of presentation (RSVP versus page-format) was randomized within each pair of paragraphs.
One-word RSVP was implemented on an Apple Macintosh, which has a white background with black letters. A 14-point font size was used. In the computer-paced RSVP condition, each word was shown for 300 ms, and a pause of 500 ms took place at the end of each sentence. In the page condition, each paragraph was printed out on separate 21.5x28.0 cm sheets of paper in letters of the same font and size as those displayed on the screen. Each paragraph was available to the subject for the same total time as in the RSVP condition, namely: 300 x (number of words) + 500 x (number of sentences) ms. Because the total display duration of each passage was based on the number of sentences in that passage, as well as the number of words, the number of words presented per minute (including pauses) was not the same for all passages because the number of sentences was not uniform. The mean words per minute was 202.4.
Two practice paragraphs in each presentation mode, followed by five test paragraphs in each mode were read under the externally- paced conditions described above. Six further paragraphs (three in RSVP, three in page-format) were self-paced by the subject. In the page condition, the subject initiated the timing of her reading and then indicated when she was finished by stopping the timer. Two keys on the keyboard were identified in the self-paced RSVP condition. By pushing one of the buttons, the reader could reduce the display duration of each word by ten percent, thus increasing her reading speed. Pressing the other key caused the display duration of each word to be increased by ten percent, decreasing reading speed. The initial presentation rate for the self-paced RSVP was the same as that used in the externally-paced display: 300 ms per word. As usual, there was a 500 ms pause at the end of each sentence.
In all conditions, the subject initiated the presentation of the text. After reading a passage, ten comprehension questions were administered. These were read aloud, one at a time, to the subject, who recorded her responses on paper. If the subject had not responded to a question within 10 seconds, the next question was read.
When the reading was externally-paced at approximately 200 words per minute, the average comprehension score was 8 out of 10 with the page-format display, and 7 out of 10 with the RSVP display. Because of the limited number of observations, no statistical analyses were performed on the data. However, comprehension levels in these two display formats seemed to be comparable.
The reading rate for the six paragraphs in the self-paced block of the experiment along with their corresponding comprehension scores are shown below in Table 1. In the RSVP condition, the subject decreased the rate of presentation twice while reading the first paragraph. The self-paced RSVP rate of presentation was not altered in subsequent paragraphs. It appears that RSVP allowed the participant to read the paragraph more quickly while still comprehending at the same level as that obtained with the slower page-format. Certainly, with such a small number of observations, no strong conclusions may be drawn from this data. However, these preliminary results are promising.
The participant's comments indicated that she preferred the page-format display to RSVP. She felt that RSVP was more stressful because it did not allow her to return to an earlier word if it had been missed. The ability to make regressions existed in the page-format display. This problem could be overcome within the RSVP paradigm by providing a means of accessing text which had already been presented. In addition, extended practice with the display may allow the reader to become more comfortable with the RSVP technique in general.
Reading rates and comprehension scores with self-paced reading
Trial Reading Rate Comprehension
16............169.5........7/10 18............164.7........8/10 20............169.1........8/10 Mean..........167.8........8/10
Trial Reading Rate Comprehension
15............134.7........6/10 17............156.5........7/10 19............132.0.......10/10 Mean..........141.1........8/10
Although the evidence presented here is limited, it is encouraging for the application of RSVP to readers with visual impairments. A retinitis pigmentosa reader was able to increase her reading speed by almost 30 words per minute while experiencing no loss in comprehension. This enhancement of reading performance required no special equipment other than a microcomputer. Perhaps the most promising feature of the present demonstration stems from the fact that the positive effects of RSVP over page-format display were obtained with a reader who had approximately seven to eight more degrees of peripheral vision than individuals in advanced stages of the disease. With less peripheral vision, one would expect to find a greater degree of reading difficulty and thus, a slower page-format reading speed than was observed with the present subject. The benefit obtained with the RSVP display would be even more substantial. Further research is in progress, in the hope of firmly establishing the technique of rapid serial visual presentation as an important aid to readers with little or no peripheral vision.
This work was supported by Research Grant UD 149 from the Natural Sciences and Engineering Research Council of Canada to the second author. We thank Pat O'Seaghdha and Kim Wrigley for helpful comments. We also thank Cathleen Drews and Dr. Barry Skarf for their comments on the clinical aspects of retinitis pigmentosa.
Bouma, H., & devoogd, A.H. (1974). On the control of eye saccades in reading. Vision Research, 14, 273-284.
Cocklin, T.G., Ward, N.J., Chen, H.-C., & Juola, J.F. (1984). Factors influencing readability of rapidly presented text segments. Memory & Cognition, 12(5), 431-442.
Duke-Elder, Sir Stewart (1970). Parsons' Diseases of the Eye (15th ed.). London: J. & A. Churchill.
Forster, K.I. (1970). Visual perception of rapidly presented word sequences of varying complexity. Perception & Psychophysics, 8, 215-221.
Harley, R.K., & Lawrence, G.A. (1977). Visual Impairment in the Schools. Springfield, ILL: Charles C. Thomas.
Juola, J.F., Ward, N.J., & McNamara, T. (1982). Visual search and reading of rapid serial presentations of letter strings, words, and text. Journal of Experimental Psychology General, 111(2), 208-227.
Marmor, M.F. (1980). Visual loss in retinitis pigmentosa. American Journal of Ophthalmology, 89, 692-698.
Masson, M.E.J. (1983). Conceptual processing of text during skimming and rapid sequential reading. Memory & Cognition, 11(3), 262-274.
Merin, S., & Auerbach, E. (1976). Retinitis pigmentosa. Survey of Ophthalmology, 20(5), 303-346.
O'Regan, K. (1981). The "convenient viewing position" hypothesis. In D.F. Fisher, R.A. Monty, & J.W. Senders (Eds.), Eye Movements, Cognition & Visual Perception (pp. 289-298). Hillsdale, NJ: Erlbaum.
Potter, M.C. (1984). Rapid serial visual presentation (RSVP): A method for studying language processing. In D.E. Kieras & M.A. Just (Eds.), New methods in reading comprehension research (pp. 91-118). Hillsdale, NJ: Erlbaum.
Potter, M.C., Kroll, J.F., & Harris, C. (1980). Comprehension and memory in rapid sequential reading. In R.S. Nickerson (Ed.), Attention and Performance VIII (pp. 395-418). Hillsdale, NJ: Erlbaum.
Rayner, K. (1983). The perceptual span and eye movement control during reading. In K. Rayner (Ed.), Eye movements in reading: Perceptual and Language Processes (pp. 97-120). New York: Academic.
Rayner, K., & Bertera, J.H. (1979). Reading without a fovea. Science, 206, 468-469.
Rayner, K., Slowiaczek, M.L., Clifton, C. Jr., & Bertera, J.H. (1983). Latency of sequential eye movements: Implications for reading. Journal of Experimental Psychology: Human Perception and Performance, 9 (6), 912-922.
Trevor-Roper, P.D., & Curran, P. (1984). The Eye and Its Disorders. Oxford: Blackwell Scientific.
Ward, N.J., & Juola, J.F. (1982). Reading with and without eye movements: Reply to Just, Carpenter, and Woolley. Journal of Experimental Psychology: General, 111(2), 239-241.
Williamson, N.L., & Muter, P. (1985). Rapid serial visual presentation and reading: The role of practice, external pacing, and multiple-word windows. Unpublished manuscript.