ArticlePDF Available

Design Improvements for Frequently Misrecognized Letters

Authors:

Abstract and Figures

To enhance typeface legibility we studied how to improve the design of individual letters. Three different fonts were created, each containing several variations of the most frequently misrecognized letters. These variations were tested both with distance and short exposure methodologies. Creating variations within a typeface avoided confounds that occur when letters from different typefaces are compared against each other. The studies found that some variations were more legible than others despite the letters within a font having similar size, weight, and personality. The results showed that narrow letters benefit from being widened, and that x-height characters benefit from using more of the ascending and descending area. These findings can be used to improve the design of future typefaces.
Content may be subject to copyright.
is is a contribution from Information Design Journal 18:2
© 2010. John Benjamins Publishing Company
is electronic file may not be altered in any way.
e author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of
offprints, for their personal use only.
Permission is granted by the publishers to post this file on a closed server which is accessible to members (students
and staff) only of the author’s/s’ institute, it is not permitted to post this PDF on the open internet.
For any other use of this material prior written permission should be obtained from the publishers or through the
Copyright Clearance Center (for USA: www.copyright.com).
Please contact rights@benjamins.nl or consult our website: www.benjamins.com
Tables of Contents, abstracts and guidelines are available at www.benjamins.com
John Benjamins Publishing Company
Information Design Journal (), 118–137
©  John Benjamins Publishing Company
do i: ./idj...bei

Sofie Beier and Kevin Larson
Design Improvements for Frequently Misrecognized Letters1
Keywords: legibility, visibility, typeface design, fonts,
lower-case letters, experimental study, typographic
research
To enhance typeface legibility we studied how to improve
the design of individual letters. Three different fonts
were created, each containing several variations of the
most frequently misrecognized letters. These variations
were tested both with distance and short exposure
methodologies. Creating variations within a typeface
avoided confounds that occur when letters from different
typefaces are compared against each other. The studies
found that some variations were more legible than others
despite the letters within a font having similar size, weight,
and personality. The results showed that narrow letters
benefit from being widened, and that x-height characters
benefit from using more of the ascending and descending
area. These findings can be used to improve the design of
future typefaces.
A common approach in experimental legibility studies
is to compare one font against another font. A valu-
able critique of this method is the issue of confounding
parameters between fonts, such as proportions, weight,
stroke, contrast and look. With so many parameters
varying in the test materials, it is difficult to identify the
variables that influence the findings.
To inform the choice of typeface for signage at
Heathrow Airport’s Terminal 5, Robert Waller (2007)
compared five different fonts: BAA Signs, Frutiger Bold,
Frutiger Roman, Vialog and Stempel Garamond Italic.
Waller’s study found, by measuring how long it would
take to recognize gradually enlarged words, that Frutiger
Bold is the most legible of the five and that Vialog is
less legible than either of the Frutiger variations or
BAA Signs. Waller speculates that the narrow width of
Vialog could be causing the font’s poor performance.
Unfortunately, it is very difficult to be certain why Vialog
performed poorer because it is different from the other
typefaces on several dimensions. BAA Signs is a serif
face while Vialog is a sans serif face. BAA Signs uses a
double-story g, while Vialog uses a single-story g. Both
BAA Signs and Frutiger Bold are heavier in weight than
Vialog. With all of these variables influencing the test
material, it is not easy to identify the exact reason why
Vialog was less legible than the others.
With a short-exposure method Fox, Chaparro, &
Merkle (2007) investigated the performance of the letters
e’ and o’ in 20 popular text typefaces. ey found that
they could measure performance differences between the
e’ in many typefaces and between the ‘o’ in many type-
faces, but that it was difficult to make claims about why
some letters performed better. With a regression analysis
they showed that when the letter ‘e’ had a higher crossbar
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

it was more likely to be misrecognized. is is an excel-
lent first step at investigating the issues that are impor-
tant in identifying factors that determine legibility, but it
might be because of the many differences between real-
world letters that no characteristics beyond the height of
the crossbar was identified for the letter ‘e’.
Waller and Fox et al. both study legibility by making
comparisons across a variety of different typefaces. is
kind of study has the advantage of studying real world
typefaces that have been optimized by type designers
for a particular purpose. ere are an infinite number
of ways to design any particular letter, but by examin-
ing existing typefaces we see a representative sample of
possible designs. e disadvantage of this kind of study is
that the designs that are studied differ on many dimen-
sions making it difficult to understand the source of the
observed difference.
Our plan in this project is to take a different approach
from Waller and Fox et al. Instead of studying exist-
ing typefaces that differ on many dimensions, we will
create different versions of letters in the same font in
order to reduce the number of characteristics that are
being changed between letters, and so making it easier
to understand the reasons for performance differences.
However, one downside of this method is that we are not
able to examine the full variety of designs that are seen in
the real world; another concern of looking only at varia-
tions within a single typeface is that the conclusions may
only apply to that typeface alone. However, we can be
more certain that the finding will broadly apply to letter
recognition by empirically testing three different fonts
under the same conditions.
Like Fox et al., we focus on recognizing isolated
letters because several researchers have demonstrated
that part of the reading process consists of a parallel
recognition of the letters in a given word (McClelland &
Johnston, 1977; Rayner & Pollatsek, 1989). More recent
work further suggests that out of the three mental opera-
tions: letter-by-letter, word-wholes, and sentence-context
recognition, the letter-by-letter operation is the strongest
(Pelli & Tillman 2007). To avoid the crowding phenom-
enon of interfering neighboring characters, the present
study has been based on a single-letter method, where
each individual character is exposed to participants and
not as part of a word.
Many typographers understand the reading process
and are similarly concerned with single-letter recogni-
tion. e renowned typographer Walter Tracy defined
legibility as being “clarity of single characters” (Tracy,
1986, p.31). Following this designation, issues such as
character differentiation, contrast, stroke angle, weight,
width, resolution, and hinting, can all be influencing
legibility. e variable under study in the present inves-
tigation is the differentiation of characters; the other
variables stay constant within each font. A notion oen
emphasized by typographers is that different reading
situations influence legibility in ways that are not always
the same. To study the performance of letter variations,
not only in relation to a specific situation but also on a
more general level, the present investigation contains
two test methods of threshold studies; one study is based
on the short-exposure method focusing on parafoveal
vision, the other focuses on recognition at a distance.
Figure 1. Typeface terminology.
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
Test Material
Studies of letter recognition tend to find similar error
patterns (Geyer, 1977; Bouma, 1971; Tinker, 1964). ere
are two main groups of characters with high error rates.
One is composed of the x-height characters of standard
width, built on a mixture of straight and curved lines
(e-c-a-s-n-u-o). e other group is composed of the
narrow letters with a single vertical stroke and no width
(i-j-l-t-f). ese two letter groups are the main subjects
of the present investigation.
e shapes of the skeleton variations under study
are inspired by the differentiation theory put forward by
Legros & Grant (1912). In a publication describing differ-
ent aspects of the printing process of the day, Legros &
Grant measured within a range of different fonts, the
amount of overlap of similar letter pairs placed on top of
each other (c-o-e, n-u, b-h, s-a, i-l). Fonts with the largest
amount of overlapping areas were defined as being less
legible than fonts of letter pairs with a smaller amount of
overlapping areas.
Familiar letter variations
e goal of all the letter variations is to create a greater
distinction between letters. With a few exceptions,
similar letter skeletons were tested on each of the three
fonts. e variations of the letter ‘i’ in the SpencerTest
and the OvinkTest faces focus on different levels of serifs.
e serifs emphasize the separation of the stem from the
dot, and are expected to have better legibility than the
versions without serifs. Serif faces need serifs on the ‘i’,
therefore there was no reason to test these variations in
the PykeTest. For similar reasons the tailless ‘u’ was not
tested in the PykeTest because it is aesthetically too out
of place in a Serif face.
A high level of differentiation between ‘n’ and ‘u’ is
expected to improve legibility. To study this hypothesis,
Figure 2. Skeleton variations of the three fonts SpencerTest,
PykeTest and OvinkTest. The fonts are all named after
twentieth-century legibility researchers.
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

u2 has no tail and the bowl of versions n2 and u3 detach-
es closer to the middle of the stem than does versions n1
and u1. In doing so it is expected that focus will be direct-
ed towards the areas where the letters are most different
from each other. A similar diagonal stroke is represented
in the crossbar of version e2. is is expected to improve
the recognition rate by opening up the counter.
e Law of Closure described by the German school
of Gestalt psychologists, suggests that our perceptual
system tends to complete incomplete shapes by filling out
gaps. Following this hypothesis it would be expected that
the smaller the aperture in ‘c’ and ‘e’ the larger the risk
that the eye will close the gap and mistake these letters
for ‘o. e hypothesis is further studied in open and
closed apertures of the letter ‘s’ in the OvinkTest. Follow-
ing the same idea, the familiar two-storey ‘a’ has versions
with open apertures, and versions with more closed aper-
tures. It would be expected that closed apertures result
in terminals optically joining the bowl and then lower
legibility.
e one-storey ‘a’ was tested in the SpencerTest and
the OvinkTest. Due to the dominating x-height round
shape, this version would be expected to show a low level
of legibility and a high number of misreadings for the
lowercase ‘o. Yet, the single-storey ‘a’ has a skeleton that
more closely resembles a handwritten ‘a’, and is therefore,
possibly, more familiar.
e narrow letters (l-f-t-j-i) cover a small horizontal
area. It would be expected that if spread over a larger
area the legibility of the characters will improve. Letters
of this group all have wide and narrow versions tested.
Unfamiliar letter variations
Some of the tested variations were more unusual than
others; these more unfamiliar versions can be divided
into two main groups. One approach explores the possi-
bility of extending the height of the character; the other
the possibility of adding uppercase character shapes to
the lowercase alphabet.
Many lowercase letters use neither the ascending nor
descending space. e approach investigates the inclu-
sion of the ascending and descending areas of some of
the letters that do not usually make use of this space.
Normally being x-height characters, the a4 and s3 move
above and below this area. We know that larger sizes are
more easily perceived than smaller sizes at a distance,
so by extending the ‘a’ into the ascending area and the ‘s’
into the descending area, it would be expected that the
otherwise highly compact inner spaces of the characters
open up and become more distinctive.
During the evolution of the lowercase alphabet, the
early uncial pen hands mixed present-day upper and
lowercase alphabets. Inspired by this tradition, the letter
variations n3, e5 and t3 are uppercase letters reduced
to x-height characters – the hypothesis goes that these
already recognized letterforms could replace the exist-
ing lowercase versions, and still function in combination
with the uppercase alphabet.
Methods of short exposure study
e first study applied a method of short-time exposure
of a single character in the parafoveal view. e findings
relate to situations of reading running text.
Figure 3. Gestalt psychology’s Law of Closure.
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
Participants
ere were a total of 41 participants in this study. Not
all participants saw all three typefaces. 15 only saw the
SpencerTest, 2 only saw the OvinkTest, 18 saw the Ovink-
Test and the PykeTest, 3 saw the SpencerTest and the
OvinkTest, and 3 saw the SpencerTest and the PykeTest.
e SpencerTest and the PykeTest were each exposed to
21 participants, where the OvinkTest was exposed to 23
participants. Most of the participants were compensated
with a gratuity of Microso soware or hardware. Some
early participants received no compensation.
e participants included 26 students with art and
design backgrounds from the Royal College of Art,
London, and 15 students from Imperial College, London.
eir ages ranged from 19 to 34, with an average age of 25.7
years. e participants came from a variety of backgrounds
(British, French, Brazilian, Danish, Canadian, Swedish,
Norwegian, Spanish, Slovenian, Polish), and all self-report-
ed either normal or corrected-to-normal visual acuity.
Because the mean number of errors made by participants
from the two schools was not reliably different, the data
from the two groups will be reported combined.
Material
e test material was created in Macromedia Flash MX
and shown on a 15-inch MacBook Pro laptop with a screen
resolution of 1440 x 900 pixels set to maximum bright-
ness. e three fonts (SpencerTest, OvinkTest and PykeT-
est) were all presented with anti-aliasing at the vertical size
of 45 pixels (an Em-square of about 1 cm). Since this is not
a study of comparison between fonts, the three faces are
not adjusted according to x-height. To minimize eyestrain
caused by the background light of the screen, the back-
ground color was a shaded white (#E6E6DD) with the
presented letters in black (#000000). e ambient room
light was typical for an office environment.
Procedures
Both the foveal and the parafoveal areas are important
for continuous reading (Rayner, 1978; Rayner, McConkie
& Ehrlich, 1978), yet the short-exposure method applied
in the present study did not detect any errors of identi-
fication when test material was placed in the foveal, and
so the focus was on the parafoveal alone. Test materials
were therefore located 2 cm to the right of the fixation
point where participants placed their focus. eir eyes
were placed at a distance of 50 cm from the screen.
Each character variation within a typeface was presented
3 times per participant. To maintain an approximately
equal appearance between the 26 letters of the alpha-
bet, the 15 characters of the English alphabet that were
not under investigation were each exposed 5 times – all
occurring in the same random order.
e instruction was to focus on a red dot on the
screen and then press the space key to trigger an expo-
sure of a single character, which participants were asked
to name. Each letter was exposed for a period of about
43 milliseconds. To ease the participant into the test, a
selection of the characters not under investigation were
presented as the first 5 exposures. A mask (exposed for
43 milliseconds) of randomly placed black dots followed
directly aer each letter exposure: this removes the
aerimage on the retina and controls the timeframe in
which the image in reality would appear on the retina.
Figure 4. The test character (left) and the after image (right),
both with a visible baseline and a dot to focus on.
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

Participants were informed that they would be presented
uppercase and lowercase letters; they were not asked to
hurry their response, as their responses were not timed.
Methods of distance study
e second study used a distance threshold method to
study the legibility of the same font letter variants. e
findings relate to typefaces presented on signs viewed at
a distance.
Participants
ere were 41 participants in this study, though 7 were
disqualified because they did not meet the minimum
visual acuity requirement of being able to recog-
nize stimuli at a distance of 4.5 meters. is le 34
participants. All three fonts were each exposed to 20
participants: 6 participants saw the SpencerTest and
the Pyketest, 6 participants saw the TinketTest and the
OvinkTest, 14 participants saw the OvinkTest and the
Pyketest, and 8 only saw the SpencerTest.
e participants were compensated with a gratuity of
Microso soware or hardware.
Material
e fonts, computer, and environment were identical in
the two studies.
Procedures
In this investigation, the laptop was placed on a podium
at the eye-level height of a standing person of about 175
cm. e angle of the screen was adjusted to fit the given
height for each person.
e first presented character was the letter ‘d’. As
identified by Tinker (1964) this character is one of the
most easily recognized letters. e purpose of this first
exposure was to locate the individual vision threshold.
e participant was placed at a distance of 10 meters
from the screen, and asked to move slowly forward
until the presented letter was at the threshold of being
identifiable; this was the distance – varying from 4.5-9
meters (with an average of 6 meters) from the screen – at
which the individual participant was tested. From this
distance, participants were asked to name each of the
letter stimuli. A new letter was presented on screen aer
each participant response. Participants were not asked to
hurry, and were permitted to take as many breaks as they
felt necessary.
is method is different from the one applied in
other recent distance studies, such as those by Sheedy
and colleagues (2005) and the studies of the Clearview
typefaces (Garvey, Pietrucha & Meeker, 1997), where the
maximum distance is measured for each letter, and the
distance itself becomes the data rather than the accuracy
from a particular distance. However, the Sheedy and
Clearview method does not identify which other letters
the character tested is most likely to be misread for – a
parameter that is measurable with the present method.
Results & Discussion
In the exposure study, each letter variation for the
SpencerTest and the PykeTest was presented a total of
63 times, and for the OvinkTest 69 times. In the distance
study, each letter variation was presented a total of 60
times. If the participant correctly identified the presented
letter, the trial was counted as correct. e inferential
statistics of a chi-square distribution were conducted on
the raw totals of correct and incorrect observations. Tests
were only conducted between variants within a font, as
it was not a goal of this investigation to compare differ-
ences between the fonts.
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
0
10
20
30
40
50
60
errors
errors
1234 1234 12 4 1234 1234 12 4
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Short exposure
0
10
20
30
40
50
60
Distance
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure Short Exposure Short
Exposure Distance Distance Distance
a errors     
a errors     ***  
a errors  *  * † -  ** ††  ** †† -
a errors      
Chi-square χ²()=.,
p=.
χ²()=.,
p=.
χ²()=.,
p>.
χ²()=.,
p=.
χ²()=.,
p=.
χ²()=.,
p>.
Statistically
Reliable
yes Yes no Yes yes no
Figure 5. Letter ‘a’
* Post-hoc tests showed reliably more errors than each of the other versions.
** Post-hoc tests showed reliably more errors than each of the other versions.
*** Post-hoc tests showed reliably more errors than versions  and .
† A high frequency of misreadings for the letter ‘q’ ().
†† A high frequency of misreadings for the letter ‘o’. SpencerTest a (), OvinkTest a ().
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

e performance of the one-storey a3 was generally
bad, with recurrent misreadings for letters ‘q’ and ‘o’. Does
this finding suggest that a one-storey ‘a’ should never be
used? In relation to the inexperienced reader it does ap-
pear to have a purpose. Recognition is a dominant fac-
tor when learning to read; the fact that the one-storey ‘a’
references to the letter shape that most children learn to
write, has a positive influence on the inexperienced reader
(Sassoon 2001). e present study, however, focuses on the
experienced reader, where references to one’s own hand-
writing are less essential.
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure Distance Distance Distance
c errors      *
c errors    
c errors      *
Chi-square χ²()=.,
p>.
χ²()=.,
p>.
χ²()=.,
p>.
Χ²()=.
p>.
χ²()=.,
p>.
χ²()=.,
p=.
Statistically
Reliable
no no no No no yes
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
123 123 123 123 123 123
Short exposure Distance
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
* Post-hoc tests showed reliably more errors than versions .
Figure 6. Letter ‘c’
e hypothesis that the open aperture of a2 would
improve legibility was not confirmed, showing no reli-
able difference in performance between a1 and a2, except
for the SpencerTest distance study, where a2 performed
reliably poorer than versions a1 and a4. is is an
unexpected difference. A possible reason for this might
originate in the shape of the bowl. e upper part of the
bowl in version a2 is more diagonal in the SpencerTest
than in the two other fonts; furthermore it bends slightly
inwards, disrupting the dynamic movement of the curve,
and making it look more like a spine than a bowl.
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short Exposure Short Exposure Short Exposure Distance Distance Distance
e errors    
e errors   †   †
e errors       †
e errors   **  ***   ***** †† 
e errors  * †  **  **   †  ****
Chi-square χ²()=.,
p=.
χ²()=.,
p=.
χ²()=.,
p=.
χ²()=.,
p>.
χ²()=.,
p=.
χ²()=.,
p=.
Statistically
Reliable
Yes Yes yes No yes yes
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
Short exposure Distance
12345 12345 12345 12345 12345 12345
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
* Post-hoc tests showed reliably more errors than each of the other versions.
** Post-hoc tests showed reliably more errors than versions , , .
*** Post-hoc tests showed reliably more errors than version .
**** Post-hoc tests showed reliably more errors than versions ,  and .
***** Post-hoc tests showed reliably more errors than versions  and .
† A high frequency of misreadings for the letter ‘c’. SpencerTest Short Exposure e (), SpencerTest Distance e (), OvinkTest
Distance e (), PykeTest Distance e (), Pyketest Distance e ().
†† A high frequency of misreadings for the letter ‘o’ ().
Figure 7. Letter ‘e’
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short Exposure Short
Exposure
Short
Exposure Distance Distance Distance
n errors   
n errors    
n errors     *  *
Chi-square χ²()=.,
p>.
χ²()=.,
p>.
χ²()=.,
p>.
χ²()=.,
p>.
χ²()=.,
p=.
χ²()=.,
p=.
Statistically
Reliable
No no no No yes yes
Figure 8. Letter ‘n
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
Short exposure Distance
123 123 123
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
* Post-hoc tests showed reliably more errors than versions .
e initial hypothesis that closed apertures in ‘c’ and
e’ would lower legibility was, in most tests, not confirmed
in the case of the letter ‘c’ – only showing a statistically
reliable difference between the open c2 and the more
closed versions c1 and c3 in the distance study of the
PykeTest. e PykeTest c1 and c3 are the only versions
tested with a teardrop on top; this finding suggests that
teardrops do not improve legibility at distance. It further
appears that in the parafoveal vision, when the letter ‘c’ is
viewed in isolation, the viewer registers the cut-off area
in the circle regardless of the size of the area and there-
fore, in contrast to all existing recommendations by most
typographers, showed no difference in the characters
having closed or open apertures.
e hypothesis that closed apertures of e4 lower
legibility was only confirmed in the OvinkTest and
the PykeTest at short exposure and in the OvinkTest
at distance. e three remaining familiar e’ variations
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
showed no internal differences. Yet all versions except e1
demonstrate a high number of misreadings for the letters
c’ and ‘o. e e5 variation performed poorly. It appears
that the upper and lower crossbars are overdominating
the middle crossbar, which in some cases resulted in a
high number of misreadings for the letter c’.
e idea that detaching the bowl from the stem
would enhance legibility of versions n2 has not been
Figure 9. Letter ‘s’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
s1 errors 9 13 * 9 ** 38 ** 22 46 **
s2 errors - 4 - - 11 -
s3 errors 4 15 * 2 18 27 * 8
Chi-square χ²(1)=1.37,
p>.05
χ²(2)=7.61,
p=.02
χ²(1)=5.21,
p=.02
χ²(1)=12.09,
p=.0005
χ²(1)=10.05,
p=.007
χ²(1)=51.39,
p=.0001
Statistically
Reliable
no Yes yes yes yes yes
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
1 3 1 2 3 1 3 1 3 1 2 3 1 3
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Short exposure Distance
* Post-hoc tests showed reliably more errors than version .
** Post-hoc tests showed reliably more errors than version .
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

confirmed – showing no statistically reliable difference
over n1 in any situation. Version n3 showed no noticeable
difference in most of the studies, except for the PykeTest
at distance presenting a statistically reliably bad perfor-
mance compared to both versions n1 and n2, and in the
OvinkTest at distance favoring version n2.
e hypothesis that the closed apertures of the
OvinkTest s2 would lower legibility was not confirmed.
e angle of the spine seems to have influenced the
performance, showing an advantage in favor of the closed
apertures of s2 compared to s1 in both the short- exposure
and the distance studies. e fact that the OvinkTest s1
has a diagonal spine and s2 a rounded spine might be the
reason for the advantage towards version s2. It appears that
the shape of the spine actually had a larger influence on
the legibility of the ‘s’ than the apertures being opened or
Figure 10. Letter ‘u’
0
10
20
30
40
50
60
errors
errors
1 2 1 2 1 3 1 2 1 2 1 3
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Short exposure
0
10
20
30
40
50
60
Distance
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
u1 errors 2 14 8 15 6 17
u2 errors 4 12 - 18 16 -
u3 errors - - 12 - - 20
Chi-square χ²(1)=0.18,
p>.05
χ²(1)=0.05,
p>.05
Χ²(1)=0.53,
p>.05
χ²(1)=0.17,
p>.05
χ²(1)=4.51,
p=.03
χ²(1)=0.16,
p>.05
Statistically
Reliable
No no No no yes no
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
closed, a finding that contradicts the recommendations of
the scholar G.W. Ovink (1938), who suggested a diagonal
spine of the ‘s’. e surprising performances of the Spen-
cerTest a2 and theOvinkTest s2 might therefore be related.
It seems that a diagonal stroke in the bowl and spine of
these letters lowers their legibility and that these areas
would benefit from being more rounded in shape.
e hypothesis that the legibility of ‘u’ would improve
by differentiating the letter from ‘n’ was not confirmed.
e tailless version u2 gave a poor performance in the
OvinkTest at distance; however, in other situations it
presented no statistically reliable difference from u1.
Furthermore, the PykeTest version u3 showed no differ-
ence from u1.
Figure 11. Letter ‘i’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
i1 errors 36 39 - 15 16 -
i2 errors 35 49 * - 29 ** 27 -
i3 errors 43 28 - 30 ** 27 -
Chi-square χ²(2)=2.52,
p>.05
χ²(2)=12.98,
p=.002
-χ²(2)=9.68,
p=.008
χ²(2)=5.66,
p>.05
-
Statistically
Reliable
No yes - yes no -
* Post-hoc tests showed reliably more errors than version .
** Post-hoc tests showed reliably more errors than version .
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
123 123 123 123 123 123
Short exposure Distance
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

e hypothesis that serifs on the letter i’ improve legibil-
ity was confirmed for distance viewing. In both the the
SpencerTest and the OvinkTest distance study, i1 with
the slab serif on top was recognized more oen than
i2 and i3; however, only with the SpencerTest showing
a statistically reliable difference, it seems as if the slab
serif on top of the stem helps to clarify the letterforms,
although when placed at the bottom, the character
becomes difficult to identify. It appears, however, that this
only happens at distance viewing, and not in the parafo-
veal view of short exposure, where i3 in the OvinkTest
performed reliably better than i2.
Figure 12. Letter ‘j’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
j1 errors 10 12 8 29 31 4
j2 errors 2 3 3 4 4 0
Chi-square χ²(1)=4.51,
p=.03
χ²(1)=4.79,
p=.03
χ²(1)=1.59,
p>.05
χ²(1)=24.08,
p=.0001
χ²(1)=27.27,
p=.0001
χ²(1)=2.33,
p>.05
Statistically
Reliable
yes yes no yes yes no
0
10
20
30
40
50
60
errors
errors
1 2 1 2 1 2
0
10
20
30
40
50
60
1 2 1 2 1 2
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Short exposure Distance
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
e hypothesis that broad characters improve legibility
was confirmed overall for the letter ‘j’, where the broad
j2 delivered a good performance on all accounts in the
SpencerTest and the OvinkTest; however, no statistically
reliable difference was demonstrated between j1 and j2
in the PykeTest. e broad ‘j’ is particularly successful
because it does not introduce any new confusions. is
differs from the broad form of the letter ‘t’ which intro-
duces a new confusion with the letter ‘c.
Figure 13. Letter ‘l’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
l1 errors 37 † 42 45 43 †† 38 †† 41
l2 errors 30 37 30 18 33 40 †
l3 errors - 40 - - 46 * -
Chi-square χ²(1)=1.15,
p>.05
χ²(2)=0.75,
p>.05
χ²(1)=6.46,
p=.01
χ²(1)=19.21,
p=.0001
χ²(2)=6.30,
p=.04
χ²(1)=0.00,
p>.05
Statistically
Reliable
No No yes yes yes no
* Post-hoc tests showed reliably more errors than versions .
† A high frequency of misreadings for the letter ‘t, SpencerTest Short Exposure l (), PykeTest Distance ().
†† A high frequency of misreadings for the letter ‘i’. SpencerTest l (), OvinkTest l (), OvinkTest l ().
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
123 123 123 123 123 123
Short exposure Distance
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

e hypothesis that broad characters improve legibility
was confirmed overall for the letter ‘l’. e broad version
l2 showed a reliably better performance in the Tinker
distance study compared to the narrower l1, in the
OvinkTest distance study compared to the straight stem
l3, and in the PykeTest short exposure study compared to
the serifed l1.
Figure 14. Letter ‘t’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
t1 errors 13 24 28 36 * 29 40
t2 errors 10 19 24 23 19 31
t3 errors 13 15 25 35 * † 39 * † 49 †
Chi-square χ²(2)=62,
p>.05
χ²(2)=2.92,
p>.05
χ²(2)=0.57,
p>.05
χ²(2)=6.99,
p=.03
χ²(2)=13.35,
p=.001
χ²(2)=12.50,
p=.002
Statistically
Reliable
No no no yes Yes yes
* Post-hoc tests showed reliably more errors than versions .
† A high frequency of misreadings for the letter ‘r’. SpencerTest t (), OvinkTest t (). PykeTest t ().
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
123 123 123 123 123 123
SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
e hypothesis that broad characters improve legibility
was confirmed in the letter ‘t’ at distance viewing. All
distance studies presented a small difference between t1
and t2 in favor of the latter, although only with a statisti-
cally reliable difference in the SpencerTest.
In all three fonts, the version t3 was frequently
misread for the letter ‘r’ in the distance threshold study,
and delivered a statistically reliably poorer performance
compared to other versions of the ‘t’. On the other hand,
in the short-exposure study this kind of misreading
was non-existent, and the three versions of the letter
performed in general quite similarly.
Figure 15. Letter ‘f’
Font SpencerTest OvinkTest PykeTest SpencerTest OvinkTest PykeTest
Study Short
Exposure
Short
Exposure
Short
Exposure
Distance Distance Distance
f1 errors 14 18 16 32 30 28 *
f2 errors 6 17 12 21 29 28 *
f3 errors 15 23 7 26 26 7
Chi-square χ²(2)=5.12,
p>.05
χ²(2)=1.49,
p>.05
χ²(2)=4.28,
p>.05
χ²(2)=4.11,
p>.05
χ²(2)=0.58,
p>.05
χ²(2)=21.54,
p=.0001
Statistically
Reliable
No no no no no yes
* Post-hoc tests showed reliably more errors than version .
0
10
20
30
40
50
60
0
10
20
30
40
50
60
errors
errors
123 123 123 123 123 123
Short exposure Distance
SpencerTest OvinkTest PykeTest SpencerTest O vinkTest PykeTest
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

e hypothesis that broad characters improve legibility was
not confirmed in the letter ‘f. e descending f3 showed
no difference in the SpencerTest and the OvinkTest;
however, a reliably better performance was demonstrated
in the PykeTest at distance. is result may be due to the
f3 version of the PykeTest being broader in shape than the
f3 version of the two other fonts. Contrary to the broad
versions of the ‘j’ and ‘l’ groups, the wide f2 did not perform
reliably better than any of the other tested variations.
Conclusion
Our technique of comparing letter variations within a
typeface has provided insights about letter legibility that
was not previously available. Earlier studies such as those
presented by Waller (2007) and Fox et.al. (2007) that
examine legibility by comparing typefaces, struggle to
make comparisons because every letter differs on several
dimensions. Our studies are complimentary to this work
by investigating letters that come from the same font
with many fewer differences. is allows us to be more
confident in understanding why one letter performs
better than another.
Based on the findings we recommend wide versions
of narrow letters. e wide j2, l2, and t2 all showed
versions that performed better than their more narrow
forms. e SpencerTest wide f2 also performed better
than the narrow ‘f. Only for the letter ‘i’ was there no
clear benefit for a wide form. Yet applying the broad
variations of j2 and l2 in a typeface will possibly result in
spacing problems: j2 will overlap with descending char-
acters to the le, an issue causing potential trouble in the
Scandinavian languages which have a high number of gj
letter combinations. Version l2 would create a disrupt-
ing area of extra white space when placed to the le of
another stem. When implementing these variations in a
final typeface, it could be necessary to apply a number of
extra ligatures and kerning pairs.
Based on the findings we can also recommend
extending letters into the ascending and descending
areas. Both of the ascending ‘a’ and descending ‘s’ versions
performed better than x-height versions. e PykeTest
version of the descending ‘f also performed better than
the non-descending forms. In the case of the SpencerTest
and the PykeTest distance studies, and the the PykeTest
exposure study, s3 showed a reliably better performance
than the x-height s1, and a4 showed in general no statisti-
cally reliable differences compared to other two-storey ‘a
versions. Implementing the high performing unfamiliar
versions in a font within a new typeface would theoreti-
cally place the font on an equal legibility level to a font
of familiar letterforms within the same typeface. e two
fonts will have the same level of legibility but very differ-
ent familiarity levels. Studying readers’ experience with
these different versions would be an interesting subject
for future research into typeface familiarity.
e hypothesis that a single storey ‘a’ is less legible
than the double storey ‘a’ was confirmed. With a high
level of misreadings for ‘o’ and ‘q, we recommend against
the single storey ‘a.
We cannot conclude that creating differences between
the letters ‘n’ and ‘u’ increases the legibility of a typeface.
Neither a tailless form of the letter ‘u’ nor lowering the
connection point of the letter ‘n’ had the intended effect.
e recognition rates were comparable to the more
common letterform.
e hypothesis that more open versions of ‘c’ and
e’ are more legible than more closed forms was not
confirmed. While there was some indication that this
was true for the letter ‘e, there was no indication this was
also true for the letter ‘c’. e extremely closed e4 versions
performed worse than the more open e1, e2, and e3.
Except for the PykeTest at distance, the more closed c3
did not perform reliably worse than c1 or c2.
e study confirmed the notion that the perfor-
mance of letter shapes varies according to the situa-
© 2010. John Benjamins Publishing Company
All rights reserved

Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -
tion in which it is presented, and that some features
are most important in distance viewing and others are
most important in the parafoveal view.ere are many
differences between a letter from one typeface and the
same letter in another typeface. e present method of
studying within-font matters provides data that has a
practical use for the design of new typefaces. We found
that within a single typeface design that wide letters
‘j’, ‘l’, and ‘t’ performed better than the narrow form in
the same design. If we had instead compared a type-
face like Courier with wide letter designs to a typeface
like Helvetica with narrow letter designs, it would be
much harder to reach the same conclusions because
of the many inherent differences between Courier and
Helvetica. Similarly, it is difficult to compare Futura’s
single storey ‘a’ to Times New Roman’s double storey
‘a. is test compared the single and double storey ‘a
within the same typeface and found the single storey
‘a’ to be less legible. is technique can help us reach a
much stronger conclusion about the legibility of differ-
ent typeface designs, and would be a useful technique
to apply to the design of any new typeface.
Note
1. This paper is partially based on the PhD thesis of the first
author while affiliated with the Royal College of Art, London, UK.
References
Bouma, H. (). Visual recognition of isolated lower-case letters.
Vision Research, , –.
Chaparro, B. S., Shaikh, A. D., & Chaparro (). Examining
the legibility of two new clearType fonts [online]. Usability
News, (), available from http://www.surl.org [Accessed 
October ].
Fox, D., Chaparro, B.S., & Merkle, E. (). Examining legibility of
the letter “e” and number “” using classification tree analysis
[online]. Usability News, (), available from http://www.surl.
org [Accessed  October ].
Garvey, P.M., Pietrucha, M.T., & Meeker, D. (). Effects of font
and capitalization on legibility of guide signs. Transportation
Research Record, , –.
Geyer, L. H. (). Recognition and confusion of the lowercase
alphabet. Perception & Psychophysics, (), –.
Harris, J. (). Confusions in letter recognition. Professional
Printer, (), –.
Larson, K. (). The science of word recognition. Typo, vol. ,
(pp. –).
Legros, L.A., & Grant, J.C. (). Typographical printing surfaces:
the technology and mechanism of their production. London:
Longmans, Green, and Co.
Lonsdale, M.d.S. (). Does typographic design of examination
materials affect performance? Information Design Journal,
(), –.
McClelland, J.L., & Johnston, J.C. (). The role of familiar units
in perception of words and nonwords’. Perception & Psycho-
physics, (), –.
McClelland, J.L., & Rumelhart, D.E. (). An interactive activation
model of context effects in letter perception: part . an ac-
count of basic findings. Psychological Review, (), –.
Ovink, G. W. (). Legibility, atmosphere-value, and forms of
printing types. Leiden.
Paap, K.R., & Noel, R.W. (). Dual-route models of print and
sound: still a good horse race. Psychological Research, , –.
Pelli, D.G., & Tillman, K.A. (). ‘Parts, wholes, and context in
reading: a triple dissociation’ [online]. PLoS ONE (): e,
New York University. Available from: http://psych.nyu.edu/
pelli [Accessed . November ].
Pyke, R. L. (). Report on the Legibility of Print. Medical Re-
search Council, London: His Majesty’s Stationery Office.
Rayner, K. (). Foveal and parafoveal cues in reading. In J. Re-
quin (ed), Attention and performance VII. Hillsdale, NJ: Erlbaum.
Rayner, K. & Pollatsek, A. (). The Psychology of Reading. Engle-
wood Cliffs, NJ:Prentice-Hall International.
Rayner, K., McConkie, G.W., & Ehrlich, S. (). Eye movements
and integrating information across fixations. Journal of
Experimental Psychology: Human Perception and Performance,
, –.
Rumelhart, D.E., & McClelland, J. L. (). An interactive activa-
tion model of context effects in letter perception: part . The
contextual enhancement effect and some tests and exten-
sions of the model. Psychology Review, (), –.
© 2010. John Benjamins Publishing Company
All rights reserved
Sofie Beier and Kevin Larson Design Improvements for Frequently Misrecognized Letters idj (), , -

Sasson, R. (). ‘Through the eyes of a child: perception and
type design’. in: D. Jury, (ed), Typographic Writing, ISTD.
Sheedy, J.E., Subbaram, M.V., Zimmerman, A.B., Hayes, J.R. ().
Text legibility and the letter superiority effect. Human Fac-
tors, , –.
Tinker, M.A. (). Legibilit y of print. U.S.A.: Iowa State University
Press.
Tracy, W. (). Letters of credit: a view of type design. Boston:
David R. Godine.
Waller R. (). Comparing typefaces for airport signs. Informa-
tion Design Journal, (), –.
About the Authors
Sofie Beier is a designer, researcher and lec-
turer employed at The Danish Design School.
She has a PhD from the Royal College of Art in
London, on the subject of typeface familiar-
ity and its relation to legibility. In the study
of legibility related matters, her academic research focuses on
integrating design approaches with methods applied by the
scientific communities.
Her typefaces are published by T, Die Gestalten Verlag, and
FontShop.
Email: sbe@dkds.dk
Kevin Larson received his PhD in cognitive
psychology in  from the University of Texas
at Austin. His academic research was on word
recognition and reading acquisition. His passion
is understanding the impact of typography on
the reading experience, and applying that un-
derstanding towards improving the on-screen
reading experience. He is a member of Microsoft’s Advanced
Reading Technologies team.
Email: kevlar@microsoft.com
Contact
The Danish Design School
Strandboulevarden 
 Copenhagen Ø
Denmark
Microsoft Advanced Reading Technologies
 Microsoft Way,
Redmond, WA 
USA
... More recent studies take this into account. Beier and Larson (2010) created and tested several letter variations within three different font styles, aiming at investigating the effect of different letter skeletons. Using a single-letter presentation paradigm, the experiment included different versions of open and closed apertures of the letters 'c' and 'e' and showed significantly lower letter recognition of closed apertures at both short exposure and great reading distances. ...
... This indicates that findings reached through experiments with digit recognition can be translated into letter recognition as well. We further replicated findings concerning individual recognition of the lowercase letters 'c' and 'e' from an investigation using different experimental designs and employed an early version of the test font Pyke (Beier and Larson, 2010). While we here looked at the mean recognition of the seven aperture letters collectively, the previous experiment investigated the letters individually. ...
Article
An often-repeated piece of advice when choosing fonts for great legibility is to use fonts with large counters and apertures. To identify effects of open and closed apertures on the letters ‘a’, ‘c’, ‘e’, ‘r’, ‘s’, ‘t’ and ‘f’, we ran an experiment using the serif font Pyke as stimulus. The letters in focus were designed for this experiment with three variations of open apertures (Open, Medium and Closed). The experimental paradigm was to present a letter either with or without flankers in the parafovea at 2◦ eccentricity. The findings showed that participants had more trouble identifying the letter if it was set in a font variation with closed apertures.
... Few studies have looked into the effects that font style might have on letter recognition and on lexical processing in glance reading. In letter recognition research, the focus is on the effects of letter structure (Beier, Bernard, and Castet 2018;Beier and Dyson 2014;Beier and Larson 2010;Bernard, Aguilar, and Castet 2016) and letter weight (Beier and Oderkerk 2019). In research on the lexical processing of words, a similar focus is seen on effects of letter structure (Dobres et al. , 2015 and letter weight (Dobres, Reimer, and Chahine 2016), and also letter width (Dyson and Beier 2016;Sawyer et al. 2017). ...
Article
Full-text available
Most text on modern electronic displays is set in fonts of regular letter width. Little is known about whether this is the optimal font width for letter recognition. We tested three variants of the font family Helvetica Neue (Condensed, Standard, and Extended). We ran two separate experiments at different distances and different retinal locations. In Experiment 1, the stimuli were presented in the parafovea at 2° eccentricity; in Experiment 2, the stimuli were presented in the periphery at 9° eccentricity. In both experiments, we employed a short-exposure single-report trigram paradigm in which a string of three letters was presented left or right off-centre. Participants were instructed to report the middle letter while maintaining fixation on the fixation cross. Wider fonts resulted in better recognition and fewer misreadings for neighbouring letters than narrower fonts, which demonstrated that wider letter shapes improve recognition at glance reading in the peripheral visual view. Practitioner summary: Most of the text is set in fonts of regular letter width. In two single-target trigram letter recognition experiments, we showed that wider letter shapes facilitate better recognition than narrower letter shapes. This indicates that when letter identification is a priority, it is beneficial to choose fonts of wider letter shapes.
... Low stroke contrast improves word recognition (Minakata et al. 2020). Simple letter skeletons result in greater letter recognition (Beier & Larson 2010;Beier et al. 2018). Condensed fonts impair letter recognition (Oderker & Beier 2020), and so do heavy and light letter weight fonts (Beier & Oderkerk 2019a), which also slow down reading speed (Chung & Bernard 2018). ...
Preprint
Readability is on the cusp of a revolution. Fixed text is becoming fluid as a proliferation of digital reading devices rewrite what a document can do. As past constraints make way for more flexible opportunities, there is great need to understand how reading formats can be tuned to the situation and the individual. We aim to provide a firm foundation for readability research, a comprehensive framework for modern, multi-disciplinary readability research. Readability refers to aspects of visual information design which impact information flow from the page to the reader. Readability can be enhanced by changes to the set of typographical characteristics of a text. These aspects can be modified on-demand, instantly improving the ease with which a reader can process and derive meaning from text. We call on a multi-disciplinary research community to take up these challenges to elevate reading outcomes and provide the tools to do so effectively.
... Multiple experiments within design and vision research have demonstrated that font style can affect both letter and word identification. Examples include serifs at vertical extremes improving distance letter recognition (Beier and Dyson, 2014), small-size sans serif resulting in faster reading speed (Morris et al., 2002), simple letter shapes causing faster recognition of trigrams (Beier et al., 2018), and greater letter differentiation improving letter recognition (Beier and Larson, 2010;Bernard et al., 2016). One typographical feature that is yet to be investigated is the impact on the perception of bold fonts of high stroke contrast on letter recognition. ...
Article
Full-text available
To make graphical user interfaces look more fashionable, designers often make use of high-stroke-contrast fonts. We are yet to understand how these fonts affect reading. We examined the effect of letter-stroke contrast on three bold fonts, one with extreme contrast between thick and thin strokes, one with no contrast, and one in between. The fonts were designed for this experiment to enable control of font variables. Participants identified the middle letter in a lowercase letter trigram in each trial, briefly presented in the parafovea (at 2° left and right of fixation) and at the foveal fixation point. There was evidence for letter recognition impairment for the font with high stroke contrast compared to the fonts with low and medium stroke contrast, while there was no significant difference in performance between the medium- and low-stroke-contrast fonts. The results suggest that bold fonts with high stroke contrast should not be considered for designs where letter recognition is a priority.
... Isolation of a given variable requires manipulation of this variable only, while the others are kept constant. When researchers are able to alter the test fonts so that only one variable is changed, they may succeed in identifying the effect of specific typographical features such as serifs (Arditi & Cho, 2005;Beier & Dyson, 2014;Morris et al., 2002), letter skeleton (Beier et al., 2018;Beier & Larson, 2010;Larson & Carter, 2016) and letter boldness (Beier & Oderkerk, 2019b). The present experiment employs this methodological paradigm to demonstrate that isolated font variables alone can induce significant differences in reading acuity in AMD patients. ...
Article
Full-text available
Low vision readers depend on magnification, but magnification reduces the amount of text that can be overviewed and hampers text navigation. In this study, we evaluate the effects that font variations letter spacing, letter width, and letter boldness have on low vision reading. We tested 20 low-vision patients with age-related macular degenera-tion (AMD) and used the Radner Reading Chart, which measures reading acuity (logRAD), maximum reading speed, and critical print size. The results demonstrated a small, but measurable effect of letter spacing and letter width on reading acuity near critical font sizes.
... M (2020). Detalle del diseño de los carácteres en Roboto, Open sans y Lato.MÉTODOPara este estudio se revisaron algunas investigaciones recientes, como(Beier and Larson 2010;Chaparro et al. 2010) que abordan estudios de legibilidad con el uso de tipografía digital y que han servido de referente para platear en método utilizado en este estudio. ...
Chapter
Full-text available
The objective of this research is to determine the degree of typographic readability of three of the most popular fonts on the google fonts platform, through a comparative analysis, based on the individual recognition of 37 characters in common, which include box letters. Low and Arabic numerals, since these characters are the ones with the highest formal similarities to each other. As part of the sample, the typographic fonts Roboto, Open Sans, and Lato were select in their regular version. This study focuses on the collection and analysis of qualitative data obtained from the visual perception of the participants in this experiment. This study allows us to identify the range of typographic legibility of the typefaces evaluated and permit us to observe the characters most confused or missed, depending on the typeface design.
... We can conclude, nevertheless, having reviewed and compared the various findings, that in the field of typeface legibility, certain fundamental principles do exist, and that they are based on differentiation (Beier & Larson, 2010;Nedeljković, Puškarević, Banjanin, & Pinćjer, 2013) and familiarity (Sanocki & Dyson, 2012). Although not recognised as actual typographical theoretic knowledge, those principles can be found in the epistemological studies of Dirk Wendt (1994), Ole Lund (1999) and Sofie Beier (2009). ...
Article
Full-text available
At the threshold of the digital era, Zuzana Licko was of the opinion that familiar letterforms owe legibility to centuries-long exposure and that all new, prototypically unmatching forms would be equally legible if used as frequently. This paper examined the legibility in the context of familiarity – is it affected by the time of exposure to a particular typeface or a typeface’s universal structure. We ran repeated measures tests with exposure period in-between. The experiment was conducted using for this purpose designed typefaces as stimuli, and the eye-tracking on-screen reading technology. The results confirmed that one’s familiarity with a typeface influences one’s reading speed. The universal letter structure, recognised by Frutiger as the prototype skeleton, is the constant that a priori provides legibility. On the other hand, the period of exposure to uncommon letterforms also has a positive impact on legibility. Therefore, considering that the period of familiarity with the humanist letterforms has been continuous since their establishment, the maxim from the dawn of the digital era can be regarded as valid.
Book
Full-text available
Type Tricks: User design is a reference book and user manual. This book disseminates the author’s research into typeface legibility, offering tips on what to consider when designing for running text, skim-reading, expressive typefaces, reading from afar, micro-type, spatial graphics, and struggling readers (older age, low-vision, children, dyslexia). The book provides readers – including designers and anyone interested in text design – with a quick, visual overview along with a comprehensive reference list.
Chapter
Typefaces are artifacts that express the dynamic relationship between form and content, encode ideas and convey meanings. Type design requires broad skills in letter shapes, in-depth knowledge of print and digital media, the reading process and an artistic sensitivity to form. The aim of this study is to explore how functional and perceptive characteristics contribute to type design activities. By employing a qualitative approach, through observations, reviewing the scientific studies, both by designers, and psychology and reading researchers, and merging them with type designers craft knowledge, this study seeks to find out which factors nurture and must be taken into account for type design. The results suggest that the relations between ascenders, descenders and x-height; contrast; color and weight; serifs; distinctive character features; counters; joins; and familiar shapes, seem to be the most important structural, perceptive and stylistic factors that contribute to type design.
Article
Full-text available
This article introduces six new ClearType fonts developed by Microsoft. Legibility of two of the serif fonts, Cambria and Constantia, is compared to the traditional serif font Times New Roman. Results show that the legibility, as measured by the number of correct identifications of briefly presented characters, was highest for the new font Cambria, followed by Constantia, and then Times New Roman. Old style digits, such as 0,1, and 2, used in Constantia resulted in confusion with the letters o, l, and z. Times New Roman symbols were confused with both letters and other symbols.
Article
Full-text available
This study investigated the legibility of onscreen typefaces and the influence of individual character features on correct identification. Specific attributes of alphanumeric characters and symbols shown to be the least legible were measured and analyzed using a statistical method called classification tree analysis. Results from this analysis for the letter "e" and the number zero are discussed.
Article
Full-text available
The research objective was to improve highway guide sign legibility by replacing the 40-year-old guide sign font with a new font called Clearview. It was believed that the current guide sign font's thick stroke design, made with high-brightness materials and displayed to older vehicle operators, exhibited a phenomenon known as irradiation or halation. Irradiation becomes a problem if a stroke is so bright that it visually bleeds into the character's open spaces, creating a blobbing effect that reduces legibility. The Clearview font's wider open spaces allow irradiation without decreasing the distance at which the alphabet is legible. Results are presented of two daytime and two nighttime controlled field experiments that exposed 48 older drivers to high-brightness guide signs displaying either the current or the Clearview font. The Clearview font allowed nighttime recognition distances 16 percent greater than those allowed by the Standard Highway Series E(M) font, without increasing overall sign dimensions.
Article
Full-text available
The present paper reports three experiments investigating the effect typographic design of examination materials has on performance. Significant differences in performance were always found in all three experiments. These differences were always in favour of text and/or question and answer sheet layouts conforming to legibility guidelines. Participants also considered the questions with these layouts easy to answer. The main conclusion was that text and question and answer sheet layouts displaying a combination of typographic features intended to improve legibility facilitate efficient search reading and answering of questions at the perceptual level of reading. The effect of typographic layout on performance should therefore be considered in order to construct valid and reliable examination materials.
Book
Reading is a highly complex skill that is prerequisite to success in many societies in which a great deal of information is communicated in written form. Since the 1970s, much has been learned about the reading process from research by cognitive psychologists. This book summarizes that important work and puts it into a coherent framework. Note that the full-text of this book is not available.
Article
The interactive activation model of context effects in letter perception is reviewed, elaborated, and tested. According to the model, context aids the perception of target letters as they are processed in the perceptual system. The implication that the duration and timing of the context in which a letter occurs should greatly influence the perceptibility of the target is confirmed by 9 experiments with 128 undergraduates, demonstrating that early or enhanced presentations of word and pronounceable-pseudoword contexts greatly increased the perceptibility of target letters. According to the model, letters in strings that share several letters with words should be equally perceptible whether they are orthographically regular and pronounceable (SLET) or irregular (SLNT) and should be much more perceptible than letters in contexts that share few letters with any word (XLQJ). The prediction was tested and confirmed. Overall results are accounted for, with some modification of parameters, although there are some discrepancies in detail. Several recent findings that seem to challenge the model are considered, and a number of extensions are proposed. (34 ref)
Article
This study combined three research methodologies to inform the choice of a typeface for signs at London's Heathrow Airport. The methodologies were legibility testing, qualitative consumer research, and expert review. The study showed that, contrary to a number of expert predictions, the serifed typeface performed as well as the sans serif in legibility testing. Character width was a more significant factor in legibility, with condensed sans serif performing relatively poorly. The use of multiple methodologies led to a richer basis for decision-making: the qualitative research revealed clear genre expectations among airport users for sans serif signs; the expert reviewers raised a range of additional issues of genre, culture and context.
Article
The confusion matrix for the full lowercase English alphabet is estimated, based upon 25 trials per letter for each of seven subjects. Average correct recognition was controlled to 0.5 by limiting brightness and duration of displays to individually determined levels. Comparison of the obtained data to that reported by Bouma 11971t for eccentric vision supports the conclusion that limited energy foveal recognition is qualitatively different from eccentric vision recognition. Comparison of the obtained data to the uppercase confusion matrix reported by Townsend (1971) supports the inference that recognition performance has more between-letter variability for both recognizability and confusion pairings for the lowercase alphabet than for the upper.