Sie sind auf Seite 1von 19

Page 1

Effects of Font Characteristics on Reading Speed



Jennifer Liu
Foothill High School
Pleasanton, CA

INTRODUCTION
Print, or written materials, has existed for more than 5000 years and is a crucial media for
communication. To communicate efficiently, the symbols used in the written or printed
materials should be highly legible and that the individual symbols should be distinct from one
another to minimize confusions among the symbols. In most western countries, the symbols for
print are alphabets. An alphabet is a set of letters basic written symbols each of which
roughly represents a phoneme of a spoken language. One of the oldest alphabet sets is the Greek
alphabet, from which many of the modern alphabet sets were descended, including the Roman
alphabet that subsequently evolved into the English alphabet that we use nowadays. Although
there are international guidelines as to how the 26 upper- and lowercase letters of the English
alphabet should be encoded and represented (ISO/IEC 646), with the development of computers
and computer software, the English alphabet can be printed in a variety of typographic forms
(fonts). Some fonts were designed to make printing most economical, e.g. by making print
more compact and thus fitting more letters on a printed page; while others were designed to be
the most pleasing and artistic. There are even software such as FontForge that allows people
who are not expert typographers to change the various physical dimensions of fonts, for instance,
the ratio between an ascender or a descender with respect to the letter body, the thickness of
Page 2
letter strokes, or the distance between lines of text. Given the abundance choice of font type, an
interesting question is whether or not the various font types affect reading performance.

It is well known that the characteristics of print limit reading performance (for a review, see
Legge, 2007). An obvious example is the difficulty in reading very fine print, or when the print
is very condensed such that there is very little space between words and between lines of text.
Indeed, factors such as print size (Latham & Whitaker, 1996; Chung, Mansfield & Legge, 1998),
contrast (Legge, Rubin & Luebker, 1987; OBrien, Mansfield & Legge, 2000), color (Legge &
Rubin, 1986), lettercases (Arditi & Cho, 2007), serifs versus non-serifs (Arditi & Cho, 2005) or
spaces between letters (Chung, 2002) or words or lines of text on reading speed (Chung, 2004;
Chung et al, 2008), have been studied extensively. There were also a few studies comparing
reading speed for different font types, however, in most cases, the comparison was only between
a pair of font types, because the investigators were interested in a certain text characteristic. For
instance, Mansfield, Legge & Bane (1996) compared reading performance for Courier and
Times-Roman font, because they were interested in whether reading speed was faster for a fixed-
width font (Courier) or a proportional-width font (Times Roman). Paterson and Tinker (1932)
measured reading speeds for 10 different fonts, but nine of them were very similar in appearance
(Scotch Roman, Garamond, Antique, Bodoni, Old Style, Caslon, Kabel Light, Cheltenham and
American Typewriter). Only one font (Old English, a Gothic style font) had an appearance that
was different from the others. Results of the study showed that reading speed for Old English
was the lowest, amounting to approximately 14% lower than the highest reading speed obtained
in the study (obtained using Scotch Roman font). To our knowledge, to date, there exists no
study that examines how reading speed change with the physical characteristics of fonts. The
Page 3
research question of this study was to examine the correlates of various physical properties of
fonts with reading speed, which we took to represent the efficiency of the font as a medium of
communication. Our results would be useful for selecting the most readable font for printed
materials for people who are visually impaired, as reading difficulty is the most frequent
complaint of these people (Kleen & Levoy, 1981; Elliott et al, 1997).

METHODS
This research project comprised two parts. In the first part, we surveyed the list of fonts
available in Microsoft Word and selected a set of 19 fonts that appeared to be unique and not
similar to one another. We then characterized the physical properties of this set of 19 selected
fonts with respect to seven measurements: the letter height, letter width, perimeter, ink area,
perimetric complexity, skeleton length and boldness. To do so, we first created an image for
each of the 26 lowercase letters of each font using Microsoft PowerPoint, for a fixed point size of
140 points. Then we used several functions written in MATLAB (Mathworks, MA; version 7.1)
to extract the different measurements for each letter template of each font. The letter height (in
pixels) was defined as the vertical distance between the topmost and the bottom-most pixel of a
letter. The letter width (in pixels) was defined as the horizontal distance between the leftmost
and the rightmost pixel of a letter. The perimeter (in pixels) was calculated using the MATLAB
built-in function bwperim. The ink area (in pixels square) was the sum of all the pixels of a
letter. The perimetric complexity was defined as the ratio between the square of the perimeter
and ink area. Previously, Pelli, Burns, Farell & Moore-Page (2006) suggested that the efficiency
of recognizing single letters is inversely proportional to the perimetric complexity of a letter. A
letter with a lower perimetric complexity can be recognized with a higher efficiency than its
Page 4
higher-complexity counterpart. The skeleton length (in pixels) is another measurement of the
spatial complexity of a letter. It was determined by using the MATLAB built-in function
bwmorph to calculate the length of the skeleton connecting the centers of the maximal disks
that could be drawn within the stroke of a letter (Bernard & Chung, 2011). The boldness of a
letter was approximated by (2!Area)/Perimeter. We chose these seven characteristics to
investigate their effects on reading performance because letter height and letter width are directly
related to letter size and it is well known that there is a resolution limit on reading. It is also
commonly believed that print that is bolder is easier to read, therefore we also examined the
effects of ink area and boldness. The perimetric complexity (taking into account the perimeter
and ink area measurements) and skeleton length are two different representations of the spatial
complexity of letters. As mentioned above, human observers recognize letters that are less
complex with a higher efficiency (Pelli et al, 2006), therefore, we hypothesized that reading
speed should also correlate with the complexity of letters of a font. Figure 1 shows how six of
the characteristics (the perimetric complexity is a calculated value and cannot be represented)
were defined for the letter w rendered in Neuropol font (one of the fonts chosen for our reading
speed measurement). Table 1 summarizes the seven measurements for each of the 19 fonts.
These measurements represent the averaged value across the 26 lowercase letters of each font.






Figure 1. Definitions of six of the physical characteristics used in our study to characterize each letter of each font.
Page 5
Table 1. Physical measurements for the 19 fonts surveyed in this study. Measurements were taken for a fixed letter
height of 60 pixels. Values are the averages across the 26 lowercase letters of each font. The highlighted ones were
the fonts chosen for reading speed measurements.

Font
Letter
Height
Letter
Width Perimeter
Ink
Area
Perimetric
Complexity
Skeleton
Length Boldness
Arial 60 39.23 250.11 1022.01 63.87 116.68 12.75
Aristocrat 60 59.91 287.74 1398.98 74.62 153.21 9.61
Bauhaus 60 44.22 228.43 1865.67 35.13 85.05 10.33
Boopee 60 33.40 214.40 614.19 89.06 104.89 12.84
Bradley Hand ITC 60 42.84 239.45 339.47 186.56 125.76 11.18
Chiller 60 33.69 195.84 592.09 91.96 100.81 11.62
Courier 60 54.00 296.93 1193.69 89.21 145.49 11.00
Curlz 60 38.32 269.31 452.42 175.04 151.06 14.06
Edwardian Script 60 57.38 219.95 570.29 83.02 130.26 7.67
Frenchscript 60 48.86 256.65 676.14 161.04 139.30 10.51
Gigi 60 43.53 259.55 498.44 154.40 137.60 11.92
Harlow Solid Italic 60 49.49 228.68 1444.76 27.15 104.35 9.24
Jokerman 60 43.71 295.76 1030.99 92.15 156.82 13.53
Magneto 60 100.16 567.88 3318.44 90.48 331.86 11.34
Neuropol 60 65.88 308.63 1696.15 77.63 141.82 9.37
Old English Text 60 35.87 225.82 864.10 69.27 121.82 12.59
Planet Benson 60 48.12 427.56 1290.75 57.63 222.87 17.77
Ravie 60 63.17 332.34 1969.07 51.41 174.73 10.52
Times New Roman 60 44.40 261.91 821.88 97.27 140.27 11.80



Figure 2. Examples of the same sentence rendered in the six font types used in the study. The font types are, from
top to bottom: Bauhaus, Curlz, Edwardian, Old English, Neuropol and Times. The x-height (height of the lowercase
letter x) was equated for the six font types in this illustration.

Page 6
Based on Table 1, we selected a set of six fonts that differed in terms of the seven physical
properties to be used in the second part of the study. The selected fonts are highlighted in Table
1, and examples of these fonts are shown in Figure 2.

In the second part of the study, we examined how oral reading speed changed with font type, for
the six selected fonts. Six observers with normal vision, aged between 17 and 32 participated in
this study. All were native English speakers. Written informed consent was obtained from each
observer after the procedures of the experiment were explained and prior to the commencement
of data collection. We measured oral reading speed using the rapid serial visual presentation
(RSVP) paradigm in which text was presented one word at a time for a given duration at the
same location (left justified in our case) on a display (Rubin & Turano, 1992; 1994; Chung et al,
1998; Chung, 2002; 2004; 2007). We used the same reading materials, and followed the same
experimental paradigm for reading speed measurement as used in Chung et al (1998) and Chung
(2002; 2004; 2007). In brief, on each trial, a single sentence was chosen randomly from a pool
of 2630 sentences. Each sentence contained between 8 and 14 words (mean = 111.7). All the
words used were from the 5000 most frequently used words in written English, according to
word-frequency tables derived from the British National Corpus (Kilgarriff, 1997). Words were
presented as high contrast (~98%), black letters on a white background of 80 cd/m
2
. None of the
observers read any sentence more than once.

For each font, we determined reading speed for five print sizes, defined as degrees (deg) in visual
angle of the lowercase letter x (x-height), so that we could construct plots of reading speed as a
function of print size. The order of testing the five print sizes was randomized and was different
Page 7
for each observer. To determine reading speed, for each print size, we presented words at five
different durations. Observers were instructed to read as quickly and as accurately as possible.
The number of words read correctly was recorded by the experimenter and saved in a datafile.
Each experimental condition (font ! print size) was tested twice, in separate blocks of trials.
Fifteen sentences were tested in each block, three for each duration, in a randomized sequence.
When pooled across the two blocks of trials for each condition, a total of six sentences were
tested for each duration, with the total number of words read ranging from 52 to 78 (mean =
65.94.2). We then fit each set of data relating the proportion of words read correctly as a
function of word exposure duration using a cumulative Gaussian curve to construct a
psychometric function. From the best-fitted psychometric function (see Figure 3 for a sample
psychometric function), we determined the word exposure duration that corresponded to 80%
1
of
the words read correctly. This duration was then converted to reading speed according to the
following equation:

!
Reading speed (wpm) =
60
RSVP word exposure duration (sec)

where wpm stands for words per minute.

The six font types were tested in a random order that was different for each observer. All
observers completed testing in 23 sessions of approximately 1.5 hrs per session.


1
Although reading speed could be determined based on different criteria of proportion-correct of words read
correctly, we used 80% in this study because previous studies used the same criterion, and thus results can be
compared across studies.
Page 8

Figure 3. Proportion of words read correctly is plotted as a function of word exposure duration (sec) for a given
font type and print size. We determined the criterion reading speed based on the word exposure duration that
yielded 80% of words read correctly. In this example, the word exposure duration yielding 80% of words read
correctly is 0.13 sec, corresponding to a reading speed of 461.5 wpm.

RESULTS
Reading speed in words per minute (wpm) is plotted as a function of print size (deg) in Figure 4.
Each panel shows the data for one observer (S1 S6), with the results for the six font types
plotted as different colored symbols (see legend). For any given font, reading speed increases
with print size when print size is small, and reaches a plateau when print size is large (Latham &
Whitaker, 1996; Chung et al, 1998). To compare how font type affects reading performance, we
followed previous research and extracted the two most important parameters from each set of
reading speed versus print size data the maximum reading speed and the critical print size
(Chung et al., 1998; Cheung, Kallie, Legge & Cheong, 2008). The maximum reading speed
(MRS), as suggested by the name, refers to the highest reading speed that could be attained. The
critical print size (CPS) refers to the smallest print size that yields a criterion percentage of the
MRS. Two commonly used mathematical functions to describe reading speed versus print size
data in log-log scale are the two-limb fit (e.g. Chung et al., 1998; Chung, 2002; 2007) and the
exponential-decay fit (Cheung et al, 2008; Chung, Jarvis & Cheung, 2007). Cheung et al (2008)
1.0
0.8
0.6
0.4
0.2
0.0
P
r
o
p
o
r
t
i
o
n

C
o
r
r
e
c
t
0.01
2 3 4 5 6 7
0.1
2 3 4 5
Word Exposure Duration (sec)
Page 9
showed that for most data-set, the two functions yielded highly comparable results, but the
exponential-decay fit gave a slightly better fit. Therefore, in this study, we used the exponential-
decay function of the following form to describe our data:

!
Reading speed (wpm) = MRS+ Amplitude " e
#(x/ $)

where MRS is the maximum reading speed in wpm, Amplitude is the drop in reading speed from
MRS (maximum amplitude occurs when x = 0) and " represents the rate of change of reading
speed with print size. Curve-fitting was performed using Igor Pro. The best-fitted curves are
shown in Figure 4 (continuous lines through data symbols). Based on these best-fitted curves,
we derived the MRS, the critical print size that corresponded to two criteria 80% (CPS
80
) and
90% (CPS
90
) of MRS, and ", the rate of change of reading speed with print size. These values
are summarized in Table 2.

The group-average values of MRS, CPS
80
, CPS
90
and " are compared across the different font
types in Figure 5. We performed a repeated-measures single-factor design ANOVA, using the
statistical package R, to determine if there was any main effect due to font type on the MRS,
CPS
80
, CPS
90
and ". After parceling out the variability due to observers, we found a main effect
of font type on MRS (F
(df=5,25)
= 3.19, p = 0.023), but not on the critical print size for both
criterion levels (CPS
80
: F
(df=5,25)
= 2.22, p = 0.084; CPS
890
: F
(df=5,25)
= 2.00, p = 0.114) or on " (F

(df=5,25)
= 1.61, p = 0.193). In other words, the MRS was different for different font types, while
the critical print size and the rate of change of reading speed with print size were similar across
the six font types (there were also more variability across observers for these measurements,
possibly the reason for the non-significant effects). The highest MRS was attained for Times
(geometric mean 95% confidence intervals = 839.7 359.5 wpm), followed by Neuropol
Page 10
(663.0 275.8 wpm), Curlz (647.7 162.5 wpm), Edwardian (580.5 176.8 wpm), Old English
(570.6 188.8 wpm) and the lowest MRS was for Bauhaus (544.9 175.2 wpm).

Figure 4. Reading speed (wpm) is plotted as a function of print size (deg) for the six font types for each of the six
observers (S1S6). For each set of reading speed versus print size data, we fit an exponential-decay function (see
text for details) from which we derived the maximum reading speed (MRS) and critical print sizes that corresponded
to 80% and 90% of the MRS.
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S6
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S5
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S4
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S3
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S2
Edwardian
Curlz
Old English
Bauhaus
Neuropol
Times
10
2
4
6
8
100
2
4
6
8
1000
2
4 5 6 7 8 9
0.1
2 3 4 5
S1
Print Size (deg)
R
e
a
d
i
n
g

S
p
e
e
d

(
w
p
m
)
Page 11
Table 2. Fitted parameters for individual observers, as well as the group-average (geometric means).

A. Maximum Reading Speed (MRS, wpm)
Font S1 S2 S3 S4 S5 S6
Geometric Mean95%
confidence intervals
Bauhaus 753.96 884.45 493.73 564.52 249.81 563.97 544.93175.22
Old English 929.79 781.92 473.25 520.41 275.51 699.71 570.61188.82
Edwardian 566.27 775.61 725.82 865.19 240.74 576.54 580.54176.76
Curlz 734.03 976.23 543.13 800.96 413.64 572.44 647.67162.67
Neuropol 1232.6 1005.3 747.05 655.99 242.61 576.54 663.01275.84
Times 1144.4 1545 665.41 1271.9 431.11 543.55 839.74359.50

B. "
Font S1 S2 S3 S4 S5 S6
Geometric Mean95%
confidence intervals
Bauhaus 0.044 0.070 0.048 0.038 0.034 0.056 0.0470.010
Old English 0.076 0.066 0.079 0.057 0.056 0.139 0.0750.025
Edwardian 0.081 0.097 0.132 0.170 0.029 0.040 0.0770.043
Curlz 0.073 0.098 0.070 0.093 0.135 0.065 0.0860.021
Neuropol 0.077 0.014 0.133 0.013 0.013 0.032 0.0300.039
Times 0.100 0.106 0.062 0.105 0.054 0.032 0.0700.025

C. Critical Print Size
80
(deg)
Font S1 S2 S3 S4 S5 S6
Geometric Mean95%
confidence intervals
Bauhaus 0.132 0.176 0.149 0.129 0.105 0.154 0.1390.019
Old English 0.173 0.172 0.206 0.161 0.150 0.292 0.1880.042
Edwardian 0.180 0.224 0.269 0.301 0.113 0.100 0.1820.066
Curlz 0.181 0.227 0.175 0.218 0.280 0.175 0.2060.033
Neuropol 0.158 0.080 0.249 0.087 0.070 0.090 0.1100.056
Times 0.202 0.202 0.149 0.213 0.136 0.092 0.1590.038

D. Critical Print Size
90
(deg)
Font S1 S2 S3 S4 S5 S6
Geometric Mean95%
confidence intervals
Bauhaus 0.163 0.224 0.182 0.156 0.129 0.193 0.1720.026
Old English 0.226 0.218 0.260 0.201 0.189 0.388 0.2400.059
Edwardian 0.236 0.292 0.360 0.419 0.133 0.128 0.2370.095
Curlz 0.232 0.294 0.224 0.282 0.374 0.220 0.2660.048
Neuropol 0.211 0.090 0.341 0.097 0.079 0.112 0.1330.082
Times 0.271 0.275 0.192 0.286 0.173 0.114 0.2080.056

Page 12

Figure 5. Group-average values are compared across the six font types for (A) maximum reading speed (MRS,
wpm), (B) ", the rate of change of reading speed with print size, (C) the critical print size corresponding to 80%
MRS (deg) and (D) the critical print size corresponding to 90% MRS (deg). The averaged values are the geometric
means. Error bars represent the 95% confidence intervals. Gray symbols represent individual observers data.

Our research question was to examine whether there were any correlation of the various physical
properties of fonts with reading speed. Although our finding that Times yielded the highest
MRS, yet it never ranks top or least on any of the physical measurement scores according to
Table 1 might imply that there was no correlation between MRS and any of the physical
characteristics, this was only for one font. To better examine the correlation between MRS and
the physical characteristics of fonts, we plotted in Figure 6 the MRS as a function of each of the
5
6
7
8
9
0.1
2
3
4
5
C
r
i
t
i
c
a
l

P
r
i
n
t

S
i
z
e
9
0

(
d
e
g
)
B
a
u
h
a
u
s
O
l
d

E
n
g
l
i
s
h
E
d
w
a
r
d
i
a
n
C
u
r
l
z
N
e
u
r
o
p
o
l
T
i
m
e
s
D
5
6
7
8
9
0.1
2
3
4
5
C
r
i
t
i
c
a
l

P
r
i
n
t

S
i
z
e
8
0

(
d
e
g
)
B
a
u
h
a
u
s
O
l
d

E
n
g
l
i
s
h
E
d
w
a
r
d
i
a
n
C
u
r
l
z
N
e
u
r
o
p
o
l
T
i
m
e
s
C
0.01
2
3
4
5
6
7
0.1
2
!
B
a
u
h
a
u
s
O
l
d

E
n
g
l
i
s
h
E
d
w
a
r
d
i
a
n
C
u
r
l
z
N
e
u
r
o
p
o
l
T
i
m
e
s
B
100
2
3
4
5
6
7
1000
2
M
a
x
i
m
u
m

R
e
a
d
i
n
g

S
p
e
e
d

(
w
p
m
)
B
a
u
h
a
u
s
O
l
d

E
n
g
l
i
s
h
E
d
w
a
r
d
i
a
n
C
u
r
l
z
N
e
u
r
o
p
o
l
T
i
m
e
s
A
Font Type
Page 13
six physical characteristics (letter height was not plotted because the other measurements were
obtained for a fixed letter-height). A regression line (taking into account the data were plotted
on log-log axes) was used to fit the data and the correlation coefficient was given in each panel.
From the figure, perimeter, perimetric complexity and skeleton length yielded relatively high
correlation coefficients, suggesting a reasonable relationship between MRS and each of these
physical characteristics. We performed a t-statistics to determine if each of these correlation
coefficients was statistically different from a slope of 0 (no correlation). None of these
correlation coefficients turned out to be different from a slope of 0, but we suspected that it could
be due to the relatively small number of fonts used in the study. Future studies using more fonts
might find a statistically significant correlation coefficient.

Figure 6. Maximum reading speed is plotted as a function of six of the seven physical characteristics of fonts (letter
height was not plotted because it was held constant for the measurements of the other characteristics). A regression
line was used to fit each set of data and the correlation coefficient, r, is given in each panel.

100
2
3
4
5
6
7
8
9
1000
10
2 3 4 5 6 7 8 9
100
Letter Width (pixels)
r = 0.08
100
2
3
4
5
6
7
8
9
1000
2 3 4 5 6 7 8 9
100
2
Perimetric Complexity
r = 0.51
100
2
3
4
5
6
7
8
9
1000
3 4 5 6 7 8 9
10
2 3
Boldness
r = 0.26
100
2
3
4
5
6
7
8
9
1000
5 6 7 8 9
100
2 3 4 5
Skeleton Length (pixels)
r = 0.61
100
2
3
4
5
6
7
8
9
1000
3 4 5 6 7 8 9
1000
2 3
Ink Area (pixels
2
)
r = -0.17
100
2
3
4
5
6
7
8
9
1000
100
2 3 4 5 6 7 8 9
1000
Perimeter (pixels)
r = 0.58
M
a
x
i
m
u
m

R
e
a
d
i
n
g

S
p
e
e
d

(
w
p
m
)
Page 14
DISCUSSION
By measuring reading speed for six font types that are very different with respect to the seven
aspects of physical characteristics of font that we defined, we were hoping to find some
correlation between reading performance and some of the physical characteristics. Based on the
statistical results, the MRS was the only parameter that showed an effect of font type. The CPS
(defined using two criterion levels) and the rate of change of reading speed with print size did
not show any effect of font type. We then determined if MRS correlated with any of the physical
characteristics of font. We found that MRS had a reasonable correlation with the perimeter,
perimetric complexity and skeleton length of fonts, but not for letter width, ink area or boldness
of fonts.

Of the seven physical characteristics of fonts that we examined, letter height and letter width are
directly related to letter size. Previous research has shown that reading speed increases with
print size, up to a certain print size, beyond which reading speed shows a plateau and does not
depend on print size (Legge et al, 1985; Latham & Whitaker, 1996; Chung et al, 1998). Figure 4
clearly illustrates that we obtained the same relationship between reading speed and print size,
for all font types. Therefore, one might expect a correlation between reading speed and
measurements of letter size (letter height or letter width). However, note that what we failed to
find was a correlation between the maximum reading speed with letter width. This is not entirely
surprising considering that the maximum reading speed, the plateau of reading speed versus print
size function, is independent of print size.

Page 15
It is often assumed that bold print is more readable than regular print. However, we found a low
correlation coefficient between MRS and boldness (Figure 6). A closely related measurement to
boldness is the ink area of the letters and again, there was a low correlation between MRS and
ink area (Figure 6). Indeed, Bauhaus was the font that had the highest score in boldness and ink
area, yet, it yielded the lowest MRS. Nevertheless, it remains possible that for a given font (as
opposed to comparing across fonts), making the strokes bolder would increase reading speed. In
my mentors lab, we are in the process of investigating how boldness per se affects reading
speed in a separate study.

We also examined whether reading speed correlates with the spatial complexity of fonts. Using
two measurements to represent the spatial complexity the perimetric complexity and the
skeleton length of letters, we found a reasonable correlation between MRS and spatial
complexity. Previously, Pelli et al (2006) reported that humans recognize single letters more
efficiently when letters are less complex (lower spatial complexity scores). Here, we found that
the result could be generalized to reading, in that MRS is also higher for fonts that have a lower
spatial complexity score.

Although our results seem to imply that the MRS may not depend on physical characteristics of
fonts such as letter width, the ink area and boldness, our analyses only examined whether or not
there exists a correlation between MRS and the physical characteristics one at a time. It remains
possible that there are interactions among the different physical characteristics to limit MRS. An
extensive investigation of the interactions of the various factors that co-determine MRS would
Page 16
require more fonts, and our use of six fonts is unlikely to be able to reveal any of the interaction
effects.

As mentioned in the Introduction, we were hoping that our results would be useful for selecting
the most readable font for printed materials for people who are visually impaired. The
conventional advice for people with visual impairment is the 3Bs bigger, brighter and bolder.
This means that if we make an object larger (by magnification), brighter (using better lighting)
and bolder, it would be easier for the visually impaired to see the object. However, based on our
results, making print larger than the critical print size, or increasing the boldness or ink area of
the font may not improve the reading performance of the visually impaired. This interpretation
should be treated with caution, though, because when print size is smaller than the CPS,
increasing the print size always leads to a higher reading speed, regardless of the font type. Also,
for the boldness effect, we were comparing the boldness measurement across fonts, it remains
possible that increasing the boldness of a given font would improve reading speed. For example,
we found that Times offers the highest MRS, and it is likely that the Times-bold print would
yield an even higher MRS.

In this study, we found that for a given print size, the Times font was consistently read faster
than other fonts. It was also the font type that yielded the highest MRS. It is interesting to note
that in Figure 6, the data point for Times, the highest MRS in all the panels, does not fall along
the trend of the other five fonts. This might imply that there is something special about the
Times font. Times is the most common font used in the printing industries nowadays (Times or
some variants of Times are the top choices for newspaper print and books), and it is likely that
Page 17
the highest reading speed obtained with Times, and the fact that it does not follow the trend of
how MRS varies with most of the physical characteristics of fonts, are simply due to a practice or
learning effect. In other words, because observers have seen print in Times more often than print
in other font types, print in Times is processed faster, akin to the perceptual learning effect (any
relatively permanent and consistent change in the perception of a stimulus array, after practice or
experience with this array Gibson, 1963). One way to test this hypothesis is to conduct a
perceptual learning experiment in which observers are trained repeatedly with other less
commonly used fonts.

CONCLUSIONS
Reading performance, characterized by the maximum reading speed, critical print sizes that
yielded 80% and 90% of the maximum reading speed, and the rate of change of reading speed
with print size, was compared for six font types that differ in their physical characteristics. The
six font types included Bauhaus, Curlz, Edwardian, Neuropol, Old English, Times. Averaged
across six human observers, Times font was read consistently faster than the other five fonts.
The maximum reading speed was highest for Times, and lowest for Bauhaus. The critical print
sizes that yielded 80% and 90% of the maximum reading speed, and the rate of change of
reading speed with print size, did not vary significantly across font types. By correlating the
maximum reading speed with different measurements of the physical characteristics of font, we
found that the maximum reading speed correlated relatively well with the perimeter, perimetric
complexity and skeleton length of fonts.
Page 18
REFERENCES

Arditi A, Cho J (2005). Serifs and font legibility. Vision Research, 45, 29262933.

Arditi A, Cho J (2007). Letter case and text legibility in normal and low vision. Vision Research,
47, 24992505.

Bernard J-B, Chung STL (2011). The dependence of crowding on flanker complexity and target-
flanker similarity. Journal of Vision, 11(8):1, 116.

Cheung S-H, Kallie CS, Legge GE, Cheong AMY (2008). Nonlinear mixed-effects modeling of
MNREAD data. Investigative Ophthalmology and Visual Science, 49, 828835.

Chung STL (2002). The effect of letter spacing on reading speed in central and peripheral vision.
Investigative Ophthalmology and Visual Science, 43, 12701276.

Chung STL (2004). Reading speed benefits from increased vertical word spacing in normal
peripheral vision. Optometry & Vision Science, 81, 52535.

Chung STL (2007). Learning to identify crowded letters: Does it improve reading speed? Vision
Research, 47, 31503159.

Chung STL, Jarvis SH, Cheung S-H (2007). The effect of dioptric blur on reading performance.
Vision Research, 47, 15841594.

Chung STL, Jarvis SH, Woo SY, Hanson K, Jose RT (2008). Reading speed does not benefit
from increased line spacing in AMD patients. Optometry & Vision Science, 85, 827833.

Chung STL, Mansfield JS, Legge GE (1998). Psychophysics of reading. XVIII. The effect of
print size on reading speed in normal peripheral vision. Vision Research, 38, 29492962.

Elliott DB, Trukolo-Ilic M, Strong JG, Pace R, Plotkin A, Bevers P (1997). Demographic
characteristics of the vision-disabled elderly. Investigative Ophthalmology and Visual
Science, 38, 25662575.

Gibson EJ (1963). Perceptual learning. Annual Review of Psychology, 14, 2956.

Kilgarriff A (1997). Putting frequencies in the dictionary. International Journal of Lexicography,
10, 135155.

Kleen SR, Levoy RJ (1981). Low vision care: correlation of patient age, visual goals, and aids
prescribed. American Journal of Optometry & Physiological Optics, 58, 200205.

Latham K, Whitaker D (1996). A comparison of word recognition and reading performance in
foveal and peripheral vision. Vision Research, 36, 26652674.

Page 19
Legge GE (2007). Psychophysics of reading in normal and low vision. Mahwah, NJ & London:
Lawrence Erlbaum Associates.

Legge GE, Pelli DG, Rubin GS, Schleske MM (1985). Psychophysics of reading. I. Normal
vision. Vision Research, 25, 239252.

Legge GE, Rubin GS (1986). Psychophysics of reading. IV. Wavelength effects in normal and
low vision. Journal of the Optical Society of America, A, 3, 4051.

Legge GE, Rubin GS, Luebker A (1987). Psychophysics of reading. V. The role of contrast in
normal vision. Vision Research, 27, 11651177.

Mansfield JS, Legge GE, Bane MC (1996). Psychophysics of reading. XV. Font effects in
normal and low vision. Investigative Ophthalmology and Visual Science, 37, 14921501.

OBrien BA, Mansfield JS, Legge GE (2000). The effect of contrast on reading speed in dyslexia.
Vision Research, 40, 19211935.

Patterson DG, Tinker MA (1932). Studies of typographical factors influencing speed of reading:
X. Style of Type Face. Journal of Applied Psychology, 16, 605-613.

Pelli DG, Burns CW, Farell B, Moore-Page DC (2006). Feature detection and letter
identification. Vision Research, 46, 46464674.

Powell BB (1991). Homer and the Origin of the Greek Alphabet.

Rubin GS, Turano K (1992). Reading without saccadic eye movements. Vision Research, 32,
895902.

Rubin GS, Turano K (1994). Low vision reading with sequential word presentation. Vision
Research, 34, 17231733.

Das könnte Ihnen auch gefallen