Dudek (1997) surveyed 27 states to determine whether they had design standards to optimize VMS legibility distance. He found that 14% specified luminance contrast ratio between legend and sign background, 36% specified external and internal illumination, 46% character height, and 39% had specifications for character width, and spacing between characters and lines of text. Dudek cited the lack of national standards and the relative novelty of VMS as reasons for the low percentage of state standards. Marston (1993) wrote that because of knowledge gaps regarding recent product developments often a “state or local agency writes the specification so generally that any CMS project or technology is applicable.”
There are no validreasons, however, for the lack of state and national design standards to ensure VMS legibility. There have been hundreds of research projects conducted since the 1950’s that address related visibility issues in static sign legibility and dozens conducted since the 1970’s that directly address the legibility of VMS. There are numerous reports that have listed the critical attributes that influence VMS legibility, tested those attributes under day and night conditions with young and old observers, and provided recommendations regarding optimal, minimal, and acceptable design standards for various criteria. Many of these findings are based on principles of visual perception that were then field tested on actual VMS. Most of these findings are applicable regardless of VMS technology type (Garvey and Mace, 1996).
The following sections consist of a summary of those research results and recommendations as well as VMS requirements found in several specification, guideline, and standards publications (the latter are further detailed in Appendix B). Although most of the research did not address issues related to special user populations, an attempt was made to synthesize the general findings with those that did test VMS with subjects with vision impairments. Special emphasis is placed on how these recommendations and requirements relate to VMS legibility for individuals with vision impairments.
Note: The majority of VMS legibility research has been conducted on signs that use a matrix design and most of the recommendations are, therefore, couched in the terms related to this type of VMS. Also, VMS legibility is the result of the interaction of all the character variables discussed below; for example, while large letter heights might be desirable they will not be visible at appropriate distances if they are displayed at low contrast.
Letter height is perhaps the first sign characteristic manipulated when attempting to improve VMS legibility. This is because (if all other characteristics are appropriate) letter height has the greatest impact on the distance at which a sign can be read (Garvey and Mace, 1996). Unlike other key variables (e.g., contrast, luminance, and stroke width) legibility distance continues to improve with increases in letter height; there is no practical asymptote. There are, however real world limitations on sign size, and there is also research that reports letter heights that result in optimal reading speeds above which performance declines (Raasch and Rubin, 1993).
Although the Americans with Disabilities Act Accessibility Guidelines (ADAAG) requirements for sign design do not specify VMS, they are meant to apply to all public access signage. The ADAAG requirements state that for public transportation vehicles, “Characters on signs…shall have a…minimum character height (using an upper case “X”) of 1 inch for signs on the boarding side and a minimum character height of 2 inches for front headsigns.” The American Public Transit Association (APTA) states, “front destination signs shall have…a message display area of not less than 9.8 inches high…” and “side destination signs shall have…a display area of not less than 3.15 inches high…” (in Wourms, et al., 2001). This is consistent with the Public Service Vehicle Accessibility Regulations 2000 which require signs in the front of the vehicle to be no less than 8 inches in height and no less than 2.8 inches on the side of vehicles (in Wourms, et al., 2001).
The later requirements are in better agreement with empirical research findings than those of the ADAAG. Bentzen, et al. (1994) in a study where films of buses with reflective disk VMS displayed on the front and side found that 8 inch letters were read further away than 6 and 4 inch letters and that 6 in letters were legible at a greater distance than 4 inch letters for subjects with vision loss. Even with the 8 inch letters, though, on average the group with intermediate visual impairment (20/80 to 20/160) was only able to read the front mounted signs at a little over 20 feet. The eight-inch signs in this study had a stroke width of 1.2 inches resulting in a stroke width to height ratio of about 1:7.A comparison of these empirical results with the predictions in Table 1 show an overestimation in the analytical model, which calculates 40 to 82 feet of legibility distance for this group of subjects. Bentzen, et al.’s subjects with visual acuities between 20/200 and 20/400 only correctly identified the sign message under these conditions approximately 20% of the time.
ADAAG requires signs in public and commercial facilities to have “Minimum character height of 3 inches”” (ADAAG, 1994: Section 4.30 “Signage”). However, in studying VMS legibility Bentzen and Easton (1996) found that while 2 inch high character LED VMS resulted in 85% accuracy for low vision subjects, accuracy dropped to 66% for legally blind individuals at reading distances as close as three to thirty-three feet.
For freeways, Dudek (1997) recommended a character height of 18 inches and between 10 and 18 inches for non-freeway applications. Garvey and Mace (1996) found proportional improvement in legibility distance with increased letter height up to about 8 inches, above which there was some drop-off in improvements.
Legibility Index (LI)
LI is a measure of the legibility distance of a sign as a function of letter height and is expressed in feet per inch of letter height (ft/in). Upchurch, et al. (1991) and Garvey and Mace (1996) found that LIs on the order of 35 ft/in would accommodate “average” older and younger observers. This means that for every inch of letter height, the subjects could read the VMS 35 feet away. However, Garvey and Mace found a significant reduction in daytime legibility with their poorest performing subjects, where LIs dropped to 22 ft/in for the 85th percentile younger subjects and 17 ft/in for the 85th percentile older groups. The situation is even worse for the truly visually impaired. Based on previous research by Muller-Munk (1986),Bentzen et al. (1994) estimated that LIs for the legally blind would more closely approximate 3 ft/in, which translates into ten-inch letter heights at 30 feet just to reach threshold.
The ADAAG requirements specify that “Characters on [transportation vehicle] signs…shall have a width-to-height ratio between 3:5 and 1:1.” Garvey and Mace (1996) found an improvement in VMS legibility distance with increases in width-to-height ratio (w:h) up to 1:1, or letters having equal width and height. Dudek (1991) reported that a 5x7 (w:h 5:7) matrix is slightly more legible than a 4x7 (w:h 4:7) matrix. Wourms, et al., (2001) found a recommendation that ranged from 3:5 to 4:5 (Woodson, 1981). Using subjects with vision impairments, Bentzen and Easton (1996) found slightly better performance with a 5x7 character LED VMS than a 6x7 character.
ADAAG specifications state, “Characters on [transportation vehicle] signs…shall have a… stroke width-to-height ratio between 1:5 and 1:10.” Wourms, et al., (2001) reported Saunders and McCormick’s (1993) recommendation, which ranged from 1:6 to 1:8. Stroke width for matrix VMS interacts directly with the width to height ratio discussed above. Although some form of double-stroke is possible on many models, typically VMS matrices have a stroke width equal to a single element. Therefore, for example, 5x7 VMS will typically have a stroke width to height ratio of 1:7.
Garvey and Mace (1996) studied VMS with red, white, and yellow elements and found no significant difference in VMS legibility for normal vision subjects. When Legge and Rubin (1986) tested non visually impaired subjects they also found no color effects when the letters were large, although they did find some small effects when the letters were near size threshold.
Legge and Rubin (1986) also tested subjects with vision impairments and found that the protanopic subjects showed “major reduction in sensitivity to red” while the deuteranopic subjects showed no change in legibility with color. Furthermore, subjects in their study who had photoreceptor degeneration performed better in blue or green than red with highest performance being achieved with the green letters. In studying the daytime legibility of LED VMS mounted within buses, Bentzen and Easton (1996) found a color effect for both subjects with vision impairments and those without under certain viewing conditions. When one-word messages were streaming at a fast rate (2.56 sec dwell), green signs were read more accurately than red signs,this finding was not evident with the slower streaming rate of 2.74 sec dwell (longer dwell means slower speed) or with static presentation. In addition to these objective results, they found a strong stated preference for the green VMS.
Legge and Rubin (1996) recommended green or gray letters stating, “these are the colors best suited for the design of reading displays for subjects with normal or low vision.” Marner (1991) reported that the U.K Royal National Institute for the Blind highly recommended yellow characters on a black background, while Schofield and Flute (1997) cited recent research suggesting people with visual impairments preferred white on deep navy blue (in Wourms, et al. 2001).
Garvey,Pietrucha, and Meeker (1997, 1998,2001) and Garvey, Zineddin,and Pietrucha (2001) have demonstrated that font can have a dramatic affect on standard highway sign legibility and, as mentioned previously, on large format letter legibility. Yager, et al. (1998) concluded that font can have an effect on reading speed when the letter heights and luminance contrast are close to threshold, they went on to state, “Until systematic comprehensive studies are done,choices of font characteristics for low vision reading will depend on uninformed biases and, perhaps, aesthetic considerations rather than optimization of performance.”
There is not, however, a great deal of flexibility in VMS font design as these signs are often restricted by a matrix format. Garvey and Mace (1996),Dudek (1991), and Bentzen and Easton (1996) all recommended fonts displayed using a 5x7 character matrix for VMS. Garvey and Mace found little variability in performance using different fonts within the 5x7 format. The Guidelines for Transit Facility Signing and Graphics (in Wourms, et al., 2001) recommend using a minimum of 7x9matrix and also state that a double stroke must be used. The latter recommendation, though, contradicts research findings (Garvey and Mace, 1996) of a 25% reduction in legibility distance with double fonts versus a single stroke font used within a 5x7 character matrix. The double font used in the Garvey and Mace research was, however, not a true double font as some of the letter strokes had only a single stroke width (for example the horizontal bars of the E were single stoke while the vertical bar was double). It is possible that a full double stroke font could provide better legibility, but many 5x7 character VMS are not capable of producing these.
Garvey and Mace (1996) recommended nighttime luminance of30 cd/m2 and 1000 cd/m2 for bright daytime viewing. They found, however, that as visual acuity worsened, more light was needed to achieve equivalent performance. Dudek’s (1991) nighttime luminance recommendation was from 30to 230 cd/m2. The European highway community has been attempting to derive standard optical test methods for VMS for the past 15 years, but have been slowed down by, among other factors, rapidly changing technology (Grahame Cheek, European Standards body (CEN), March 8, 2002: personal communication). There currently are no photometric standards to specify what aspect of the sign should be measured (for a discussion on the issues, see Garvey and Mace, 1996; or Lewis,2000).
Combining the results of six research efforts on static traffic sign legibility, Sivak and Olson (1985) derived a recommended contrast ratio of 12:1 for positive contrast signs. Staplin, et al. (1997) expanded this to between 4:1 and 50:1. On VMS, Colomb and Hubert (1991) found improvements in daytime legibility to level off between 8 and 20 percent contrast. Stainforth and Kniveton (1996) reported that generally accepted luminance contrast ratio for VMS is 10:1. Dudek (1991) stated that for VMS, contrast ratio between 8:1 and 12:1 should be used for light emitting technologies and 40% daytime and 50% nighttime contrast for light reflecting technologies. The “Passenger Information Services: A guidebook for Transit Systems” recommends 70% contrast for all signs (Wourms et al. (2001). At extremely low contrast (less than 10%) Legge, Ahn, Klitz, and Luebker (1997) found reduction in reading speed.
Letter, Word, and Line Spacing
“Characters on [transportation vehicle] signs…shall have…“wide” spacing (generally, the space between letters shall be 1/16 the height of upper case letters)” (ADAAG, 1991). This is an extremely narrow spacing, which is not supported by experimental research with or without visually impaired subjects. Garvey and Mace (1996) tested inter-letter and inter-word spacing in computer simulated matrix VMS words and found that the “single element” inter-letter spacing (1/7 letter height) produced the poorest results. They recommended a minimum spacing of 3/7th letter height, or almost seven-times that required by the ADAAG. Dudek (1991), in summarizing European VMS standards, wrote that the desirable inter-character spacing is 2/7th letter height and line spacing is 4/7th letter height. Mace and Garvey (1996) found an inter-line spacing of 75% to be best for three-line signs, but noted that this was probably excessive for signs displaying only two lines. Woodson (1993), reported that inter character spacing should be between 25 and 50 percent of character height and inter word spacing should be from 75 to 100 percent of letter height (in Wourms,et al. 2001).
The ADAAG requirements read, “Lower case characters are permitted” (ADAAG, 1994). Research by Garvey, et al. (1997, 1998) support this statement by demonstrating that for highway signs lower case words are more legible (by 12 to 15%) than uppercase for word recognition. They also found that upper case and lower case words perform equally well for word legibility, where individual letter reading is required. The “Passenger Information Services: A guidebook for transit systems” stated that uppercase letters should be used for destinations and other short messages, and mixed case should be used for “long legends and instructions,” and the Public Service Vehicle Accessibility Regulations (2000) state, “Destination information shall not be written in capital letters only” and that “the use of both upper and lower case text helps ensure that words that are not completely clear and legible to people with a degree of vision impairment or learning disability, are still identifiable through shape recognition of the word.” (in Wourms, et al., 2001). Dudek (1991) made an important practical argument, however, when he pointed out that for matrix signs, a 7x9 character matrix is required to produce reasonable lower case fonts, and because of this he recommended using upper case letters.
Contrast Orientation (or Polarity)
Positive contrast signs have light letters on dark backgrounds and negative contrast signs have dark letters on light backgrounds. This terminology results from using the contrast ratio Lt-Lb/Lb where Lt is the luminance of the “target” (or in the case of signs, the letters) and Lb is the background luminance. While the ADAAG sign requirements state “Characters on [transportation vehicle] signs…shall…contrast with the background, either dark-on-light or light-on-dark” (ADAAG, 1991), Garvey and Mace (1996) reported a 29% improvement in nighttime VMS legibility distance with positive versus negative contrast messages. Positive contrast is also recommended by Iannuzziello (2001) for general transit signage.
Peli and Fine (1996) found that individuals with vision impairments (median acuity of 20/100) need a larger “window” than their normal vision counterparts to read streaming text. While non-visually impaired subjects in their study performed well with a window of about four to five letters, those with vision impairments required six to seven letters for optimal performance.
VMS often present more information than will fit on a single display. The messages must therefore be displayed in a dynamic format, either by paging or by some form of scrolling or streaming. Paging means that the information is static, but a number of pages of information are shown sequentially to convey the entire message. Scrolling typically denotes that the text is moving down the sign from the top to the bottom. Streaming refers to text that moves across the sign from the right to the left. Streaming is the method used most frequently with single line message boards.
Although the capability of providing a large amount of information in a small space is part of what makes VMS such useful tools, the necessity of paging, scrolling, and streaming also presents some unique challenges when attempting to accommodate readers with vision impairments. Perhaps one of the most important considerations in this regard is reading speed. Legge, et al., (1997), stated that people with visual impairments read more slowly than their sighted counterparts even with visual aids such as magnification and high contrast materials. They stated that the slower reading is due to reduced visual span (number of letters legible in a fixation), which results in smaller eye movements and longer fixation duration. Unfortunately, researchers have yet to reach consensus on how best to display dynamic information to individuals with vision impairments.
Paging or Streaming
In evaluating LED next-stop VMS on buses, Bentzen and Easton (1996) found that, “static messages were clearly superior to streaming messages.” However, some messages are too large to fit on a single display.Bentzen and Easton (1996) found that when this was the case streaming messages outperformed paging messages. However, Kang and Muter (1989) reported earlier research (Sekey and Tietz, 1982) that found reading speed to be slower for constant scrolling (i.e., Times Square) than either “saccadic scrolling or page mode.” Their ownresearch, supports that of Bentzen and Easton (1996) and they concluded that scrolling works as well as static techniques and is “preferred by readers.” With regard to readers with vision impairments, however, they went on to quote Williamson, et al.’s (1986) suggestion that static presentation may help those with peripheral field loss because it “paces readers and prevents [eye movement] regressions.” Fine, et al. (1997) disputed this in reporting that subjects with vision impairments did not show any improvement with static presentations (i.e., rapid serial visual presentation or RSVP) over dynamic or scrolling text. They stated that unlike people with normal vision, those with central field loss (CFL) make eye movements during static presentations, which slows reading speed. In 1996, Fine and Peli took this a step further when they suggested that individuals with CFL might actually benefit from dynamic text, because tracking the text might help to stabilize their eye movements. As recently as 2001, however Wourms, et al. wrote, “Scrolling information is very difficult for a person who is visually impaired to read, text should be displayed in a fixed manner if possible.”
Display Time and Speed
How long a static message should be displayed and how fast a dynamic message should stream across the sign is mainly a function of the target audience’s reading rate. Proffitt, et al. (1998) stated, “The average adult reads about 250 words per minute (wpm) during normal reading.” Kang and Muter (1989) put the rate at 280 wpm for college students. On the other hand, Lovie-Kitchin, et al. (2000) reported that the mean silent “rauding” rate (reading for comprehension) for a group of subjects with macular degeneration was 105 wpm while the oral rate was 79 wpm. Krischer and Meisser (1983) however, stated that it was impossible to determine the reading speed of individuals with vision impairments as a group because of the heterogeneity of the population and the fact that reading speeds are highly dependent on the functional type of impairment. Their own research provided evidence for this when it revealed that central scotoma has “by far” the largest effect on reading speed, and that individuals with peripheral field loss had relatively fast reading rates.
Perhaps mainly because of the reasons mentioned by Kricher and Meisser (1983), there are numerous and varied recommendations regarding both VMS message duration and speed. It has been recommended that, “a line of text should be displayed for at least ten seconds, preferably a little longer.”(ECMT, 1999). Dudek, 1991 recommended a minimum exposure time of “one second per short word… or two seconds per unit of information” for use with unfamiliar drivers. Harris and Whitney (1993) wrote that if scrolling is used, information should be left on the screen for at least twice the normal reading time. Barham, Oxleuy, and Shaw (1994) found a fixed time of about 10 seconds to avoid confusion and recommended message display from 10 to 20 seconds (in Wourms, et al., 2001). Joffee (1995) recommended a display time of 1.6 seconds when a VMS must display multiple pages of information. Finally, when Bentzen and Easton evaluated two streaming rates (i.e., 2.75 and 2.56 sec dwell times) they found that, while there was no main effect, the faster rate interacted negatively with both color and sign height. Furthermore, in their focus group study these researchers found that a dwell time of 3.5 sec was so slow as to appear to flicker and a dwell time of 1.5 sec was too fast for subjects to consistently read the message.