Investigation into the Legibility of Tiresias Keyfont Version 1 and the Development of Tiresias Keyfont Version 2
by Antoinette Fennell
1. Introduction
Version 1 of Tiresias Keyfont (Tiresias Keyfont V1) was designed in 2003 for clear labelling of keys and controls. This report describes the results of an investigation into the legibility of Tiresias Keyfont V1 numerals and symbols and alternative character shapes. The outcome of the project has been the development of Tiresias Keyfont V2.
1.1. The Choice of Typeface for Legibility
Legibility is defined as "(of print or handwriting) clear enough to read; readable". The word is often used synonymously with the term readability, however a distinction can be made between the two. Legibility is used to describe the ability to recognise character and word shapes. Whereas readability is expanded to include the style of writing.
Investigations into typeface legibility often involve the test subject reading a continuous section of text (Mills and Weldon 1987; Tinker 1963; Zachrisson 1965). Variables such as speed of reading or the number of misspellings identified are recorded. However results can be potentially influenced by factors such as familiarity with the context of the text (i.e. similar to a teacher correcting untidy handwriting, the reader is familiar with the subject matter of the text, and so can make an educated guess as to what a word might be, even if the word is not visually clear). In a similar way, an understanding of how language is structured can also potentially influence the results (again, an educated guess can be made when a word is not visually legible).
When designing or assessing a typeface for labelling keys, the requirements are more specific. With respect to keypad labels, individual character recognition is of utmost importance.
Numeric keypads should have large clear open numerals, and the star (*) and hash (#) keys should be easily distinguishable. An alpha-numeric keypad has both letters and numerals on each key (such as a mobile phone keypad). The presence of both letters and numbers causes problems for legibility: the space for each character is more limited, subsequently limiting the maximum size that each character can be. Here the importance of individual character recognition is considerable.
The locations of numerals and symbols on keypads will not always be familiar to users. One typical problem is the inconsistent use of both the telephone (1 2 3 on top row) and calculator (1 2 3 on bottom row) layouts. The location of the star (*) and hash (#) keys can also be inconsistent between keypads.
Many public access terminals, such as information kiosks and public telephones with Internet access, use a full QWERTY keyboard. On these keyboards the numeric labels are usually located in a straight line at the top. When using these numeric keys, even the most experienced computer users often depend on reading the labels.
Furthermore, even if a consistent design was applied to every keypad, a user with an intellectual impairment may not necessarily remember the standard order or layout of numerals and symbols. The user may therefore rely solely on reading the labels.
All of these factors make it crucial for each numeral, letter and symbol to be as legible as possible.
1.2. The Choice of Typeface in the Keypad Design Process
When producing a keypad, manufacturers (based on a survey of UK-based manufacturers) tend to use a typeface chosen by their clients. And, similar to the selection of newspaper typefaces, a keypad typeface is often chosen simply for its familiarity.
Aesthetics can be another influencing factor, so that the keypad fits in with the design of the surrounding terminal. And a typeface might be chosen simply because it matches the existing style of the client product.
The client can dictate the overall appearance of the keypad they are ordering, including typeface, font size, and colour combination. Again, colours are often chosen to fit in with the existing style of the client company. Maximum legibility of a typeface is not always a priority. However, some manufacturers do make alternative recommendations if the client has made a less than preferable choice.
1.3. Current Choice of Typeface
A survey of eight UK-based keypad manufacturers and distributors revealed the following most commonly requested typefaces for labelling keys:
- Arial family (Arial, Arial Black etc.)
- Helvetica family (Helvetica, Helvetica Medium etc.)
- Courier or Courier New
- Times New Roman
Arial and Helvetica are often chosen as clear sans-serif typefaces, however they are a poor typeface for numerals. Looking at the numerals six, eight and nine:
Figure 1. The numerals 6, 8 and 9 in Arial, Arial Black and Helvetica
In all three typefaces the six and nine are very closed, reducing the level of white space. For a person with low vision, the three numerals are easy to confuse.
Looking at Courier, Courier New and Times New Roman, the six and nine are more open than Arial:
Figure 2. The numerals 6, 8 and 9 in Courier, Courier New and Times New Roman
However, these fonts almost represent the opposite end of the scale: very fine with a lot of white space. Although they do have more open numerals, these typefaces are not easy to read.
1.4. Tiresias Keyfont
Tiresias Keyfont V1 combined the positive attributes of existing typefaces, offering clear open numerals, with a relatively higher level of boldness.
Figure 3. The numerals 6, 8 and 9 in Tiresias Keyfont V1
The numeral shapes have now been further developed (Tiresias Keyfont V2), for example opening up the six and nine, thus maximising the distinction between the numerals six, eight and nine. The aim of this project was to compare the legibility of Tiresias Keyfont V1 numerals with a number of alternative shapes, thus leading to the development of Tiresias Keyfont V2.
1.5. Weight
According to Paterson and Tinker (1940), bolder typefaces are less legible when used for large sections of text. In the case of keypad labels, the amount of text used is minimal, therefore this particular finding is not likely to be relevant.
With respect to people who can be classified as having 'normal' eyesight, at the age of 60, approximately one third as much light reaches the retina as when they were 20 years of age (Gill 2001). Therefore the weight or boldness of a typeface can serve to reduce or increase legibility by influencing the amount of light that is reflected off a surface (be it a book or a keypad).
Keypads and buttons are usually fixed in position on a terminal or machine. Therefore they cannot usually be moved to compensate for poor lighting conditions or bright sunlight. The contrasting issues of glare (too bright) and low illumination (too dark) both serve to reduce the legibility of labels. Bolder typefaces reflect less light (Luckiesh and Moss 1942), providing a possible solution to the deleterious effects of glare. However legibility can be reduced again if the typeface is too bold (Paterson and Tinker 1940).
Three different levels of boldness are compared in this study.
1.6. Contrast
For on-screen text, visually impaired readers tend to prefer light text on a dark background (Silver et al. 1994). This is true for a range of different visual impairments including cataracts, macular degeneration and presbyopia. Examples of optimal light on dark colour combinations include white on black and white on blue.
The majority of printed text is presented as black text on a white background. With respect to black on white print, the ratio of white to black on a single page is high (estimated at 80% white, and 20% black: Rubin & Legge 1989). In the case of white text on a black background, the reverse is the case. White has high reflectance and consequently produces more glare. For that reason, it follows that having 80% black and only 20% white on a page will substantially reduce the amount of glare.
The contrast polarity of text (white on black or black on white letters) has little effect on normal reading, but it can affect low vision reading (Rubin & Legge 1989). Evidence suggests that, for certain eye conditions, black on white is beneficial (Legge et al. 1987). Legge et al. (1985) found that low-vision readers with cloudy ocular media (e.g. cataract) read up to 50% faster with white on black text.
Two contrasting colour combinations are investigated in this study: Black text on a White background and White text on a Black background.
1.7. Hypotheses Tested
To determine the legibility of Tiresias Keyfont V1, numerals of the original font are being tested against alternate numeral shapes. Null hypothesis: The current design of Tiresias Keyfont V1 numerals and the alternate numeral shapes are of equal legibility.
1.7.2. Hypothesis 2
To maximise the legibility of Tiresias Keyfont V2, three alternate weights (levels of boldness) are being tested. Null hypothesis: the original weight of Tiresias Keyfont V1 is as equally legible as bolder alternates.
1.7.3. Hypothesis 3
Alternate shapes of the symbols Hash (#) and Star (*) are being tested. Null hypothesis: the original shapes of Tiresias Keyfont Hash and Star symbols are of equal legibility as the alternate shapes.
1.7.4. Hypothesis 4
Each of the tests in Hypotheses 1 to 3 above is repeated using Black text on a White background and White text on a Black background. Null hypothesis: Tiresias Keyfont V1 and the alternate shapes tested in this study are equally legible when presented as Black text on a White background and White text on a Black background.
2. Methodology
2.1. Test Subjects
A pilot test was carried out to ensure that the tasks were clearly defined, and to calculate the duration of the evaluation.
Volunteers were recruited from the RNIB Scientific Research Unit database and Merton Vision, a London-based organisation for blind and partially sighted people. Additional volunteers were acquired through the RNIB internal mailing system.
A total of 21 test subjects took part in the Tiresias Keyfont Eyechart study. Of these, eight considered themselves to have good vision and 13 had poor vision, such that they could be registered blind or partially sighted.
The test included subjects with a wide variety of eye conditions. Visual acuity was not tested during the course of this study; information on eye condition was provided by the subjects themselves. The reported eye conditions included macular degeneration (n = 5), cataract (n = 3), glaucoma (n = 1), optic atrophy (n = 1), diabetic retinopathy (n = 1) and tunnel vision (n = 1). One test subject did not know the cause of her visual impairment.
A separate sample of 54 visually impaired test subjects took part in the symbol testing (see Section 2.4. below). These subjects were recruited from the RNIB Scientific Research Unit database. The reported eye conditions in this group included retinitis pigmentosa (n = 13), macular degeneration (n = 7), nystagmus (n = 5), vision impairment resulting from injury or illness (n = 5), glaucoma (n = 4), albinism (n = 4), hereditary (n = 2), cataract (n = 1), optic atrophy (n = 1), cone dystrophy (n = 1), neuropathy (n = 1), myopia (n = 1), more than one eye condition (n = 8). One test subject did not know the cause of her visual impairment. Again, this information is based on information provided by the subjects themselves.
2.2. The Tiresias Keyfont Numerals
The numeral shapes were altered using Macromedia Fontographer version 4.1.
The numerals one (1), four (4), six (6) and nine (9) were altered. The shape variations can be seen by following the link to Table 1 below.
Table 1: The shape variations tested for the development of Tiresias Keyfont V2 numerals.
An adaptation of the Snellen eye chart (van derWeijer 1862) was produced, comprising of original Tiresias Keyfont numerals, rather than letters. With four numerals in each line, the following font sizes were used: 152, 130, 108, 87, 65, 43, 33, 21, 15 and 9 point. The test subject was asked to call out the numbers on the chart (referred to as Eye Chart 1), until a size at which the subject was unable to read was reached.
2.3.2. Numeral Shapes
The subject was then asked to read out a list of 60 randomly arranged numbers (referred to as Eye Chart 2) at the smallest size (s)he was capable of reading. The subject was asked to call out each number, and also to mention if it resembled any other number (example of test subject comment: "I'm pretty sure it's a six but it could be an eight"). Comments and mistakes were recorded.
The 60 numerals comprised of the numbers 0 to 9. They included four different shapes of the number one, and three different shapes of the numbers four, six and nine. Every other number occurred as only one shape. There were, therefore, a total of 19 different numerals (see Fig. 1). Sixteen of these were each repeated three times (16x3=48), and three of these were each repeated four times (3x4=12), adding up to a total of 60. The order was randomly selected using the Microsoft Excel "RANDBETWEEN(bottom,top)" function. Each of the 19 numerals occurred at least once in the first twenty, second twenty and third twenty numerals on Eye Chart 2. Therefore each numeral occurred in the top middle and bottom of the eye chart.
The repetition of each number (with either three or four occurrences of each) reduced the likelihood of a bias in the results due to the location of the numbers on the chart.
As the subject read out the numbers, mistakes and comments were recorded. The subject was not prompted as to whether the numbers were read out correctly or incorrectly until the test was completed. Also, the subject was not told that there were different shapes of the numbers one, four, six and nine in the eye charts until the test was completed.
2.3.3. Numeral Weights
A test similar to Eye Chart 2 was repeated at three different weights (the original Tiresias Keyfont V1 weight, V1 plus 40 em units, and V1 plus 80 em units). An "em unit" is a Macromedia Fontographer-specific measurement. It is a relative measurement associated with the area covered by a numeral, letter or symbol. Weight increase was controlled so that the vertical and horizontal sizes remained the same. Test subjects were asked to read out a set of twenty numbers, containing at least one occurrence of each of the 19 numeric shapes. The weight chart was referred to as Eye Chart 3.
Table 2: The three weights tested for the development of Tiresias Keyfont V2 numerals
2.4. Symbols
Four alternate shapes for the symbols Star (*) and Hash (#) were also investigated (see Tables 3 and 4). These symbols could not be included in the eye charts, as the pilot test revealed that they caused confusion when they were included with the numerals.
Test subjects were shown a set of flash cards with a variety of icons. The results revealed that the Star and Hash icons were correctly identified 100% of the time (each test subject was not necessarily familiar with the term 'Hash', but the symbol was recognised as that on a telephone, or else was accurately described: e.g. "four intersecting lines").
It was finally decided to ask test subjects, by means of a questionnaire, which Star and Hash shape they found the clearest and easiest to read in a questionnaire. This test was performed by a larger sample of volunteers (n = 54).
Table 3: The shape variations tested for the development of Tiresias Keyfont V2 symbol 'Hash'
Table 4: The shape variations tested for the development of Tiresias Keyfont V2 symbol 'Star'
2.5. Contrast
All of the testing was repeated using black text on a white background and white text on a black background. The order of testing was alternated between subjects, 50% of subjects starting on Black on White, 50% starting on White on Black. Subjects were given time to allow their eyes to adjust.
2.6. Light Level
In order to maximise the number of test subjects recruited, it was necessary to carry out the testing in a number of different locations, therefore the lighting level could not be controlled between subjects. A measurement of ambient lighting was recorded, using a Tenmars TM-201 Lux/FC Digital Light Meter, at the time of each test. The testing procedure was short enough that the light level did not vary during the course of the test.
Statistical analysis revealed that light level was not a biasing factor in the observed results.
2.7. Limitations to the Study
Due to the nature of the test, subjects with a variety of eye conditions took part. This meant that the smallest font size at which each test subject could read varied quite considerably. Similarly, the preferred distance at which each subject could read was not controlled. Since there is a large variation in the distances at which general users operate a keypad, this was not deemed a concern. Therefore, as long as the distance was consistent through the testing procedure itself, subjects were told to stand at a distance at which they were most comfortable.
The amount of time taken to read the numbers was not recorded. This project focussed on accuracy and not speed of reading.
2.8. Statistical Analysis
For each test subject, the numbers read from Eye Charts 2 and 3 were marked as either 1 (correctly read) or 0 (incorrectly read). If the test subject gave more than one answer for a number (i.e. "it looks like a six or an eight") it was marked as 0. This was done even if the subject eventually recognised the number correctly.
Each numeral was tested between three and four times with each test subject. To rule out the biasing effects of pseudo-replication, an average for each numeral for each test subject was calculated. Statistical analysis of the averages revealed the same patterns of significance as the pooled results. Therefore it can be concluded that pseudo-replication was not a biasing factor in the analyses.
The data were analysed using SPSS Version 14.1. None of the tested variables had a normal distribution, so non-parametric analyses were used throughout.
2.8.1 Numeral Shapes
Within each number (i.e. one, four, six or nine), the success rates of different shapes were compared using the Cochrane Q Test. Tests were repeated including and excluding results that were 100% correct. The Q values and the significance values of both sets of test results were equivalent therefore (for simplicity) 100% correct test subjects were excluded from the analysis.
2.8.2 Numeral Weights
Weight comparisons were carried out using One-Way ANOVA with Games-Howell post-hoc.
2.8.3. Symbols
The Hash and Star results were analysed using Cochrane Q Test and McNemar Test.
2.8.4 Contrast
McNemar Test was used for pairwise comparison of Black text on a White background versus White text on a Black background.
3. Results
3.1. Test subjects with 'good vision'
All test subjects who classified themselves as having 'good vision' (n = 8) were capable of reading the smallest font size (9 point). All scored 100% in the eye chart testing. From this it can be concluded that all Tiresias Keyfont variations tested in this project can be comfortably read by people with good vision.
3.2. Numeral Shapes
3.2.1. Black Text on White Background
The results of the investigation into alternate numeric shapes presented as Black text on a White background are summarised in Table 5.
Table 5: Descriptive statistics - Black Text on White Background
Table 6: Cochrane Test results for Black Text on White Background
|
One |
Foura |
Six |
Nine |
Q |
3.8 |
- |
3.6 |
13.6 |
N |
33b |
33b |
33b |
33b |
df |
3 |
- |
2 |
2 |
p-value |
0.290 |
- |
0.168 |
0.001 |
a100% were read correctly, therefore analysis could not be performed
bPlease Note: N = 33 data elements is equivalent to N = 11 test subjects
A significant difference was detected between the three shape variations of the number Nine (Cochrane Test: Q = 13.6, df = 2, p < 0.01).
Closer analysis (see Table 7) revealed that this significant difference specifically referred to the '9-1 vs. 9-2' relationship, with 9-2 showing a significantly higher success rate.
There was no significant difference between the shape variations for the numbers one, four and six.
It should be noted, however, that the shape for the number six (6-2) that corresponds to the most legible nine (9-2; see Table 7) did have a higher success rate than the other two shape variations.
Table 7: McNemar Test 2-tailed p-values for the 'Nine' numerals
|
9-1 |
9-2 |
9-3 |
9-1 |
- |
|
|
9-2 |
0.001* |
- |
|
9-3 |
0.096 |
0.063 |
- |
*Significant at p < 0.01
3.2.2. White Text on Black Background
The results of the investigation into alternate numeric shapes presented as White text on a Black background are summarised in Table 8.
Table 8: Descriptive statistics - White Text on Black Background
There were no significant differences between any of the shape variations when presented as White Text on a Black Background (see Table 9).
Table 9: Cochrane Test results for White Text on Black Background
|
One |
Four |
Six |
Nine |
Q |
3.0 |
2.0 |
2.0 |
0.0 |
N |
36a |
36a |
36a |
36a |
df |
3 |
2 |
2 |
2 |
p-value |
0.392 |
0.368 |
0.368 |
1.000 |
aPlease Note: N = 36 data elements is equivalent to N = 12 test subjects
3.3. Numeral Weights
3.3.1 Black text on White Background
For the numeral 6-1 (the original Tiresias Keyfont V1 numeral 'six'), Weight 1 (the original weight) had a significantly lower success rate than the bolder Weights 2 and 3 (Cochrane Test: Q = 8, df = 2, p <0.05). This was the only significant result within this category.
3.3.2 White text on Black Background
There was no significant difference between Weight 1, Weight 2 and Weight 3 (see Table 2 for description) for White text on a Black Background (Cochrane Test: n.s.).
3.4. Symbols
Table 10: The descriptive statistics for the Tiresias Keyfont V2 symbol 'Hash' for Black text on White background
H2 was chosen the highest number of times (see Table 10). The number of times H2 was chosen was significantly higher than that for H3 (McNemar Test: p < 0.05) and H4 (McNemar Test: p < 0.01).
Table 11: The results of preferences for the Tiresias Keyfont V2 symbol 'Hash' for White text on Black background
H6 was chosen the highest number of times (see Table 11). The number of times H6 was chosen was significantly higher than that for H5 (McNemar Test: p < 0.01) and H8 (McNemar Test: p < 0.001). It should be noted that H2 (see Table 10) and H6 (see Table 11) are equivalent.
Table 12: The descriptive statistics for the Tiresias Keyfont V2 symbol 'Star' for Black text on White background
S2 was chosen the highest number of times (see Table 12). The number of times S2 was chosen was significantly higher than that for S1 (McNemar Test: p < 0.001), S3 (McNemar Test: p < 0.05) and S4 (McNemar Test: p < 0.001).
Table 13: The results of preferences for the Tiresias Keyfont V2 symbol 'Star' for White text on Black background
S6 was chosen the highest number of times (see Table 13). The number of times S6 was chosen was significantly higher than that for S5, S7 and S8 (McNemar Test: all p < 0.001). It should be noted that S2 (see Table 12) and S6 (see Table 13) are equivalent.
3.5. Contrast
The following results are a comparison of Black text on a White background versus White text on a Black background.
Table 14: McNemar Test results for Black Text on White Background versus White Text on Black Background. The Mean values represent the averages of the 1/0 scores. A Mean value closer to 1 indicates a higher success rate.
For the numeral shapes 1-3, 2, 4-1, 4-2 and 4-3, Black text on White Background had a significantly higher success rate than White text on a Black background (McNemar Test: all p < 0.05). In contrast, for the numeral shape 9-1, White text on a Black background had a significantly higher success rate than Black text on White Background (McNemar Test: p < 0.05).
Eight of the 11 test subjects showed a clear preference for either Black on White or White on Black. In each case the colour combination was preferred by four of the eight subjects. The remainder of the test subjects had no preference. However it should be noted that the four subjects who preferred Black on White were able to read White on Black. Whereas the four test subjects who preferred White on Black either made a negative comment or failed to read this category of eye chart at all. In all cases the contrast strained their eyes.
4. Discussion
This study has demonstrated that a typeface with open numerals is preferable for readers with low vision. With respect to a more closed typeface, the numbers six, eight and nine in particular can be easy to confuse.
There are many keypads that are used by the general public (e.g. ATMs, telephones, Chip and PIN keypads). And a typeface that is more legible for people with a visual impairment may not necessarily be more legible for people with good vision. A chosen typeface should be comfortably legible by users with good vision. It does not necessarily have to be more legible, but it cannot be less so. All of the test subjects with good vision who volunteered to take part in this project were able to read all of the shape variations successfully. Therefore it can be concluded that Tiresias Keyfont V2 will be legible by people with good vision, irrespective of which shape variation was chosen.
4.1. Numeral Shapes
The eye chart testing carried out in the project revealed that the original shapes of the numbers six and nine (i.e. those in Tiresias Keyfont V1) were less legible than new alternative shapes. A new, more open shape was tested for both numbers and had a higher success rate, with respect to legibility. For these numerals the Null Hypothesis can be rejected.
Alternative shapes of the numbers one and four were of equal legibility, when compared to the original shapes. As a result, it is not necessary to change the current design (i.e. those of Tiresias Keyfont V1) for these numbers. For these numerals the Null Hypothesis can be accepted.
4.2. Numeral Weights
The only significant result in the weights testing was in the case of the Numeral 6-1 (i.e. the original Tiresias Keyfont V1 shape of the numeral Six). Since this shape will not be included in Tiresias Keyfont V2 (see Section 4.1. above), it can be assumed that the accepted Tiresias Keyfont V2 numerals are equally legible at the weights tested in this study. Therefore, with respect to the Tiresias Keyfont V2 numerals, the Null Hypothesis can be accepted.
4.3. Symbols
A significant difference was detected in preference for alternate Hash and Star shapes. Therefore, the Null Hypothesis (that the original shapes of Tiresias Keyfont Hash and Star symbols and the alternate shapes are equally legible) can be rejected.
The Hash symbol H2 and the Star symbol S2 were chosen as the preferred shapes for Black text on a White background. The Hash symbol H6 and the Star symbol S6 were chosen as the preferred shapes for White text on a Black background.
H2 (#) and H6 ( # ) are the same symbol shape, with the Black/White colour combination reversed. Similarly S2 (*) and S6 ( * ) represent the same symbol shape.
The new Hash symbol is the same weight as the original Tiresias Keyfont V1 Hash symbol, with a small increase in the spacing between the four intersecting lines.
The new Star symbol is the same weight as the original Tiresias Keyfont, with the arms of the star elongated.
4.4. Contrast
The results of this part of the investigation were, in a way, conflicting. On the one hand, comparison of the Black text on a White background versus White text on a Black background revealed that for 15 of the 19 numeral shapes, Black on White was more legible than White on Black. This result was only statistically significant for five of these numeral shapes (1-3, 2, 4-1, 4-2 and 4-3; see Table 9). These results indicate that the Black on White colour option is more legible.
On the other hand, four of the 19 numeral shapes had a higher success rate for White on Black, with only one of these results being statistically significant (9-1; see Table 9). However the four numerals in question were 6-1, 6-3, 9-1 and 9-3 i.e. the numerals that have not been accepted for Tiresias Keyfont V2. The results in Table 9 suggest that there is more consistency in the legibility of characters presented in the form of White text on a Black background.
On the basis of these results, the Null hypothesis (Tiresias Keyfont V1 and the alternate shapes tested in this study are equally legible when presented as Black text on a White background and White text on a Black background) can be rejected. However further research should be carried out to investigate the Black on White versus White on Black options.
4.5. Final Conclusions
The results of this study emphasise the requirement for open numerals when choosing a typeface for labelling keys.
This investigation into the legibility of Tiresias Keyfont V1 has resulted in the development of Tiresias Keyfont V2. The results suggest that an increase in the weight of Tiresias Keyfont V2 does not increase or decrease legibility.
In general, Black text on a White background was more legible than White text on a Black background for Tiresias Keyfont V1 and V2. However, there appears to be more consistency in the legibility of characters presented in the form of White text on a Black background. These conflicting results indicate that further research is warranted.
4.6. Further Work
This study investigated the legibility of Tiresias Keyfont characters when printed on paper. Ideally, a typeface for labelling keypads should be tested on the medium to which they shall be applied. Accordingly the legibility of Tiresias Keyfont V2 will be tested on keypads at a later date.
Similarly the legibility of white text on black keys versus black text on white keys should be tested.
In both cases, the ability of users to complete specific tasks (in this case the ability to key in a sequence) can be tested, using speed and accuracy of data input as a measure.
5. Bibliography
Gill, J. (2001). Keeping Step: Scientific and Technological Research for Visually Impaired People. RNIB. UK.
Legge, G.E., Rubin, G.S. and Luebker, A. (1987). Psychophysics of reading. V. The role of contrast in normal vision. Vision Research, 27:1165-1177.
Legge, G.E., Rubin, G.S., Pelli, D.G. and Schleske, M.M. (1985) Psychophysics of reading. II. Low vision. Vision Research, 25: 253-266.
Luckiesh, M. and Moss, F.K. (1942). Reading as a Visual Task. Van Nostrand Company Inc. USA.
Mills, C.B. and Weldon, L.J. (1987). Reading Text from Computer Screens. ACM Computing Surveys, 19: 329-358.
Paterson, D.G. and Tinker, M.A. (1940). How to Make Type Readable: A Manual for Typographers, Printers and Advertisers. Harper and Brothers Publishers. UK.
Rubin, G. S. and Legge, G. E. (1989). Psychophysics of reading. vi---the role of contrast in low vision. Vision Research, 29(1):79--91.
Silver J.H., Gill, J.M. and Wolffsohn, J.S.W. (1994). Text Display Preferences on Self-Service Terminals by Visually Disabled People. http://www.tiresias.org/reports/atm.htm
Tinker, M.A. (1963). Legibility of Print. Iowa State University Press. USA.
Snellen H. Letterproeven tot Bepaling der Gezigtsscherpte (PW van der Weijer 1862). Cited in Bennett A.G. (1965). Ophthalmic test types. British Journal of Physiological Optics, 22: 238-71.
Zachrisson, B. (1965). Studies in the Legibility of Printed Text. Almquist and Wiksell. Sweden.
6. Acknowledgements
The author would firstly like to extend her gratitude towards the anonymous volunteers who took part in the focus groups and/or the testing phases of the project.
Thank you to Merton Vision, the Pocklington Resource Centre, Balham and Beyond Sight Haringey: The Winkfield Centre.
Keypad manufacturers who helped with the survey:
- Storm Interface
- Calman Technology
- Diamond Elec
- EECO Switch
- MMI
- Radiatron/ACAL
- Relec Electronics Ltd
- Tabcrest Graphics Ltd
- Touch Panel Products
7. Download Tiresias Keyfont for Free
To obtain a free copy of Tiresias Keyfont V2 for a PC go to the free download page.
Appendix 1 - Sample of Eye Chart 2 (MS Word Document; 157KB)
Last updated: 20.11.2009 © Copyright reserved Website design: Digital Accessibility Team