Extent of the Problem
The World Health Report (1998) estimated that there are almost 45 million blind people worldwide. Of those, 43% lost their vision due to cataracts, 15% from glaucoma, 11% by trachoma, 6% were children under the age of five with vitamin-A deficiency, and 24% were caused by a combination of diabetic retinopathy, macular degeneration, optic neuropathy, and other causes. The World Health Organization (2002) reported that worldwide there are approximately 180 million persons with vision disabilities and 40 to 45 million blind (blind defined as “cannot walk about unaided”).
The US Census Bureau (1997) reported that 3.7% of U.S. citizens (7.7 million people) over 15 years of age “had difficulty seeing words/letters,” this increases to 12.1% for individuals 65 years of age and older. Klien (1991) reported that the leading causes of visual impairment in the U.S. were age related macular degeneration (24%), open-angle glaucoma (14%), cataract (13%), and diabetic retinopathy (8%). Based on National Center for Health Statistics data, Bentzen, Crandall, and Myers (1999) estimated that in the United States there are “6.6 million people unable to read printed signs at normal viewing distances.”
Although clinical diagnoses like those noted above can be useful, Legge and Rubin (1986) made the following critical observation: “It has been shown that visual criteria [e.g., presence or absence of central vision] are better than medical diagnoses [e.g., diabetic retinopathy] for predicting the success rates for the utilization of low-vision reading aids. We are concerned with the effects of visual abnormalities, whatever their cause, on reading performance. We have found that reading performance can be predicted from visual measures that transcend diagnostic categories.” Lovie-Kitchin, Bowers,and Woods (2000) took Legge and Rubin’s functional distinctions a step further including near and far visual acuity, and measures of scotoma size and position. While there are many ways to functionally describe vision loss, the Community Services for the Blind and Partially Sighted (2002) listed the following useful definitions:
Visual Impairment:Trouble seeing with one or both eyes even when wearing glasses or contacts.
Severe Visual Impairment:Inability to read ordinary newsprint even with the best correction (glasses or contact lenses).
Low Vision: Vision that cannot be further improved by corrective lenses or medical or surgical intervention, although low vision rehabilitation may help someone to use remaining sight more effectively.
Legal Blindness: A central visual acuity for distance of 20/200 or poorer in the better eye with correction, or a field of vision no greater than 20 degrees in widest diameter.
Functional Blindness: No useful vision; clinically measured light perception or less.
Impact of Vision Loss on Sign Legibility
Vision loss is a rather vague term that, as the definitions above attest, can be characterized in many ways. The critical relationship between visual impairment and sign legibility, however, is often portrayed as a loss in visual acuity (e.g., “A person is termed legally blind when their visual acuity…is 20/200 or worse after correction…” – Wourms, Cunningham, Self, and Johnson, 2001). While poor visual acuity is not the only functional deficit associated with visual impairment, it is clearly a significant factor in sign legibility.
Visual acuity can be broadly defined as the ability to discriminate fine detail, but what fine details are required for VMS legibility? When testing visual acuity with acuity optotypes, such as Snellen letters, Landolt C’s, and Lazy E’s, the critical detail is assumed to be stroke-width. The visual angle of the stroke-width at the test distance is considered to be the minimum angle of resolution (MAR) and is used to describe visual acuity. With Snellen letters, for example, the stroke-width is equal to 1.0 arcmin (1/60th of a degree) on the 20/20 line (logMAR = 0.0), and the letter height is 5.0 arcmin (1:5 stroke-width-to-height ratio).
A number of researchers have used MAR to predict sign legibility distance for a range of individual visual acuities (e.g., Howett, 1983; Colomb, Hubert, Carta, Bry, and Dore-Picard, 1991; and Wourms, et al. 2001). These efforts boil down to the application of a simple trigonometric calculation (A = arctan C/D). Where A isMAR, C is size of the critical detail, and D is the viewing distance. Table 1 contains legibility distances calculated in this fashion for letter heights ranging from one to eight inches, standard VMS stroke-width-to height ratios of 1:5 and 1:7, and acuities from 20/10 to 20/400.
This is a clean analytic tool based on principles of visual perception and trigonometry; however, while it is not surprising that sign legibility is correlated with visual acuity, it is not true to infer that the relationship is as straightforward as Table 1 implies. The reason for this is that sign reading is not the same as acuity chart reading. There are two main differences between acuity charts and signs. First, acuity charts test high contrast black symbols on white backgrounds while signs vary in contrast, are designed using both positive and negative contrast orientation, and use many different colors. Second, acuity charts test very special letters or symbols (i.e., optotypes) with specific characteristics, and signs use words, sentences, and phrases.
Letters used on signs are different than the stimuli used on acuity test charts and, because of this, sign legibility distances cannot be accurately predicted from measures of visual acuity. Garvey, Zineddin, and Pietrucha (2001) found that stroke-width resolution alone does not determine letter acuity, even with fairly simple letterforms and high visual functioning observers (i.e., acuity better than or equal to 20/40). These researchers replaced the letters on a standard Snellen chart with letters displayed in thirteen different fonts. They found that to be read at the same distance, letters in some fonts had to be twice the height of letters in other fonts. Mean acuity for the Snellen letters was 20/16 and acuity for the other fonts ranged from 20/12 to 20/30. Furthermore, the differences were a function of a complex combination of letter attributes and not stroke width alone. This study illustrates the lack of generalizability of measured acuity to other test or sign fonts. To complicate matters further, in a follow-up study, Zineddin (2002) found that acuity can be influenced by familiarity with the test stimuli, and that subjects’ acuity for a given font could actually improve with practice.
Another reason for the lack of correspondence between MAR and sign reading is that signs use words, sentences, and phrases, and not merely strings of letters. Word legibility introduces cognitive factors quantitatively and qualitatively different from those posed by letters. Sentence reading takes this a step further as mentioned by Legge, Ahn, Klitz, and Luebker (1997) who stated that reading speed for words in sentences could be faster than for single words because of the “predictability of the words in sentences.” Fine, Peli, and Reeves (1997) suggested that this was due in part to the additional information provided by syntactic and semantic sentence content. Because of these cognitive components, sign message recognition does not require the ability to discriminate all content elements (e.g., every stroke of a letter or even all the letters in a word, or words in a sentence or phrase) for correct message identification to occur (Proffitt, Wade, and Lynn, 1998). Familiar word recognition has been shown to be based more on global features such as overall shape or “footprint” (Garvey, Meeker, and Pietrucha, 1998) than on letter characteristics. As a result, at least for normally sighted individuals, sign legibility distances are longer than would be predicted by either visual acuity or sign characteristics alone (Kuhn, Garvey, and Pietrucha, 1998). This is known as the word superiority effect (for a review see Zineddin, 2001). The extent of this effect on VMS reading for individuals with vision impairments has not been evaluated.
Table 2.Legibility distance (in feet) for various letter heights, stroke-width-to-height ratios, and visual acuities.
|Visual Acuity||MAR (arcmin)||Stroke-Width-to-Height Ratio 1:5|
|Letter Height (in)|
|Visual Acuity||MAR (arcmin)||Stroke-Width-to-Height Ratio 1:7\|
|Letter Height (in)|
While sign legibility has been the major focus of VMS readability research, it is not the only measure of sign performance. Reading speed also has a dramatic impact on the effectiveness of VMS, which are designed to present information in a sequential format. Research evaluating “optimum acuity reserve” (the ratio between threshold acuity and optimal print size) has demonstrated that the best possible reading speeds result from print size that may be as much as four times size threshold (Bowers and Reid, 1997; Yager, Aquilante, and Plass, 1998; Lovie-Kitchin, et al. 2000). In fact, Yager et al. (1998) reported that reading speed at size threshold is 0.0 words per minute (wpm) increasing linearly with log letter height (average reading speeds is approximately 250 wpm). Calculations such as those shown in Table 1 however, only provide letter height thresholds and do not take reading time into consideration. Based on text reading research, the letter heights in Table 1 (even if they accurately predict sign reading) would need to be increased by a factor of four to optimize reading performance.
Other Visual Impairments
While central vision is, of course, critical to reading ability, deficits in peripheral vision and the presence and position of localized scotoma also impact an individual’s ability to read VMS (Raasch and Rubin, 1993).Scotomata are “small area[s] of abnormally less sensitive or absent vision in the visual field, surrounded by normal sight…islands of total visual loss in other parts of the field are referred to as absolute scotomata.A relative scotoma is a spot where the vision is decreased but still present” (Concise Medical Dictionary, 1998). Research discussed below in the section on streaming text demonstrates that these impairments can result in a reduction in reading speed.
While an analytic solution to the problem of attaining appropriate letter height for various acuity populations (e.g., Table 1) is enticing, to truly determine VMS readability for the population of individuals with vision impairments it is necessary to evaluate real-world sign reading by individuals with visual deficits.