Luciole, a new font for people with low vision

Abstract

Low vision affects the ability to read and can be a major barrier to educational success and insertion into the workplace. We designed a new font (Luciole) to improve the readability and comfort for people with low vision. In this study, we analyze the effect of the font type on readability. Luciole was compared to five other fonts (Arial, OpenDyslexic, Verdana, Eido and Frutiger) in 145 French readers (73 with low vision, and 72 normal sighted), aged 6 to 35 years old and divided into four reading expertise groups. Participants completed two tasks, first reading texts on paper and then reading false-words on screen using eye tracking. About half of the participants with low vision had a subjective preference for Luciole when reading on paper and on the screen; lower preference is noted for participants with normal vision. Other readability criteria show a slight advantage of the Luciole font over some fonts (e.g., Eido and OpenDyslexic) in both groups. The results obtained when taking into consideration the level of reading expertise confirm this trend.

Full text

Acta Psychologica 236 (2023) 103926
0001-6918/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Luciole, a new font for people with low vision
Anna Rita Galiano
a
,
*
, Vanessa Augereau-Depoix
a
, Nicolas Baltenneck
a
, Laura Latour
a
,
b
,
Hind Drissi
a
,
c
a
Laboratoire D´
eveloppement, Individu, Processus, Handicap et ´
Education (UR DIPHE), University Lumi`
ere Lyon 2, Lyon, France
b
IRSAM, Lyon, France
c
CTRDV (PEP69), Villeurbanne, France
ARTICLE INFO
Keywords:
Low vision
Readability
Typography
Visual impairment
ABSTRACT
Low vision affects the ability to read and can be a major barrier to educational success and insertion into the
workplace. We designed a new font (Luciole) to improve the readability and comfort for people with low vision.
In this study, we analyze the effect of the font type on readability. Luciole was compared to ve other fonts
(Arial, OpenDyslexic, Verdana, Eido and Frutiger) in 145 French readers (73 with low vision, and 72 normal
sighted), aged 6 to 35 years old and divided into four reading expertise groups. Participants completed two tasks,
rst reading texts on paper and then reading false-words on screen using eye tracking. About half of the par-
ticipants with low vision had a subjective preference for Luciole when reading on paper and on the screen; lower
preference is noted for participants with normal vision. Other readability criteria show a slight advantage of the
Luciole font over some fonts (e.g., Eido and OpenDyslexic) in both groups. The results obtained when taking into
consideration the level of reading expertise conrm this trend.
1. Introduction
Trends in prevalence of blindness and visual impairment over 30
years (1990–2020) estimate that in 2020, 43.3 million people are blind,
295 million people have moderate and severe vision impairment, and
258 million have mild vision impairment, and 510 million have visual
impairment from uncorrected presbyopia (Bourne et al., 2021). In
France, according to an HID (Handicaps-Incapacit´
es-D´
ependance) sur-
vey (Sander et al., 2005), an estimated 1.7 million people are visually
impaired (2.9 % of the population). There are 207,000 profoundly blind
people, with 61,000 completely blind, and 932,000 moderately visually
impaired people who have a visual impairment (severe difculties or
inability to read, write, or draw, near or distance vision). Finally, slightly
>560,000 visually impaired people are mildly visually impaired.
In most research on visual impairment, it has been reported that the
prole of visually impaired participants is complex and heterogeneous
(Cass et al., 1994; Sonksen & Dale, 2002).
In particular, low vision (LV) covers a broad spectrum of visual
impairment: varying degrees of sight loss (mild and moderate), can
impact central or peripheral vision (or both) or produce fragmented
vision (scotomas), and often include poor night vision and problems
with glare. Difculty with reading is the principal concern of people
with low vision. The inability to read the newspaper at a normal reading
distance (40 cm) with the best refractive correction can be considered as
low vision (Legge et al., 1985). A Swedish study that looked more spe-
cically at the literacy of 10-year-olds found that only 41 % of visually
impaired students reached the same reading level as 75 % of sighted
students. The author concluded that reading demands may be too high
for more than half of visually impaired students if their visual impair-
ments are not considered. In addition, 30 %, including many slow
readers, had difculty concentrating, tired quickly, rubbed their eyes or
felt like giving up (Fellenius, 1999). Finally, age is a stronger predictor of
reading rate than any of the clinical vision measures for children with
low vision (Lovie-Kitchin et al., 2001).
Font characteristics can affect the readability and legibility of text.
Legibility and readability are separate, though connected. Legibility is
an attribute of text that relates to its capability of being properly iden-
tied. Legibility refers to perception, and is directly related to the ease
with which the visual system can detect characters and words. This is
important for the initial data acquisition phase of reading (Sheedy et al.,
2005). Readability refers to the difculty or ease with which a reader
deciphers texts and gives meaning to the text read. Many studies on the
legibility of fonts have suggested that certain characteristics can affect
legibility for both sighted readers and those with low vision (see Russell-
* Corresponding author.
E-mail address: anna.galiano@univ-lyon2.fr (A.R. Galiano).
Contents lists available at ScienceDirect
Acta Psychologica
journal homepage: www.elsevier.com/locate/actpsy
https://doi.org/10.1016/j.actpsy.2023.103926
Received 6 December 2022; Received in revised form 6 April 2023; Accepted 21 April 2023

Acta Psychologica 236 (2023) 103926
2
Minda et al., 2007 for review).
In sighted people, several studies have found effect of type size
(Bernard et al., 2001; Bernard et al., 2002; Cornelissen et al., 1991;
Wilkins et al., 2009), of serif and sans serif type (Arditi & Cho, 2005;
Bernard et al., 2002; Dogusoy et al., 2016; Moret-Tatay & Perea, 2011;
Yager et al., 1998), height of the central body of letters (x-height)
(Bernard et al., 2001; Lockhead & Crist, 1980; Watts & Nisbet, 1974;
Wilkins et al., 2009; Wilkins et al., 2020), of letter spacing (Chung, 2002;
Hughes & Wilkins, 2000; Manseld et al., 1996; Moriarty & Scheiner,
1984; Woods et al., 2005) and of letter stroke width (Arditi et al., 1995;
Bigelow, 2019; Larson & Carter, 2016; Sheedy et al., 2005; Wendt,
1994) on readability and legibility.
In particular, sans serif fonts inuence on-screen reading and have a
positive effect on reading speed, although performance is related to
reader age (Bernard et al., 2002). Similarly, other studies have shown
that adult participants read in sans serif fonts faster and more accurately
than in serif fonts. Specically, serif fonts slow down reading at very
small sizes, but don't matter at larger sizes. In other words, rendering
serifs at small sizes may be counterproductive (Morris et al., 2002).
In addition, Lockhead and Crist (1980) showed that changing the x-
height can make letters distinct, which is an advantage for novice
readers and children with disabilities (Wendt, 1994). For children
reading on computer screens, the optimal font is a medium to large x-
height (Bernard et al., 2002).
Also, Moriarty and Scheiner (1984) found a reading speed advantage
when the spaces between letters were reduced by 18 % (Moriarty &
Scheiner, 1984).
Finally, the stroke width of letters affects legibility, e.g. stroke
thickness greater than the standard value does not help reading, and
excessively bold strokes result in slower reading (Bernard et al., 2013;
Sheedy et al., 2005).
Another criterion retained in the studies is subjective preference.
Readers' subjective preference is for sans serif fonts (Bernard et al.,
2002). Sans serif fonts are considered by school-aged children to be the
easiest and fastest to read, and most attractive (Dogusoy et al., 2016;
Yager et al., 1998).
In terms of size, 14-point fonts have been shown to be the easiest to
read at a xed distance of 57 cm from the computer screen (Bernard
et al., 2001; Bernard et al., 2002). Finally, the fonts that have either very
wide or narrow letter spacing (e.g. Courier and Times, respectively) are
popular among both children (Bernard et al., 2002).
Typography research usually does not include people with visual
impairments, and their results cannot always apply to readers with low
vision (Russell-Minda et al., 2007). Yet, readability is related to typo-
graphic aspects and depends on the reader's vision level (Bigelow,
2019). The main criteria for good reading experience for adults with low
vision are magnication, acuity reserve (which is print size relative to
threshold print size), contrast reserve (which is print contrast relative to
contrast threshold) (Whittaker & Lovie-Kitchin, 1993), and visual eld
(number of letters visible) (Alabdulkader & Leat, 2010). But these
criteria are not sufcient to ensure good reading for all people with low
vision, especially for children (Alabdulkader & Leat, 2010). A few
studies have investigated the characteristics of fonts that improve
reading for people with low vision. They have so far shown that font size
is an important factor for reading in low vision (Legge et al., 1985;
Arditi, 1999; Rubin et al., 2006). Indeed, reading speed on paper in-
creases with print size (Rubin et al., 2006). However, there is no
consensus on the optimal font size for low vision. According to Russell-
Minda et al. (2007), a larger print of at least 16- to 18-point type would
be preferable, but too much magnication can affect readability since it
requires more frequent eye movements (Legge, 2016). However,
considering the font size alone is not sufcient to impact readability for
the visually impaired (Arditi, 2004; Legge, 2016; Nersveen & Johansen,
2016).
Indeed, the font weight is a contributing factor to good contrast, as
long as it is not excessive (Arditi, 2004; Brozovi´
c et al., 2018; Nersveen &
Johansen, 2016; Rubin & Legge, 1989). It has been shown that excessive
type weight is not benecial for reading speed, especially for those with
central vision pathology (Bernard et al., 2013; Chung & Bernard, 2018).
In addition, narrow- and wide-spacing letter strings produce different
letter confusion matrices (Arditi et al., 1995; Liu & Arditi, 2001).
However, a study for two new fonts, Eido and Maxular Rx, designed
specically for individuals with people with macular degeneration
(MD), showed that spacing was a signicant predictor of reading per-
formance for subjects with MD (Xiong et al., 2018). More precisely,
narrower Spacing induces faster reading speed. In the same way, Arditi
and Cho (2007) suggest that for visually impaired readers, uppercase
letters are more legible than other case styles because lowercase letters
are, on average, smaller in height and width than uppercase characters,
offering an advantage to uppercase letters.
Research does not offer solid conclusions about the readability of
serif and sans serif fonts. Unlike sighted readers, no signicant effect of
serif presence or absence on reading performance has been demon-
strated for visually impaired individuals (Arditi, 2004; Arditi & Cho,
2005; Nersveen & Johansen, 2016; Brozovi´
c et al., 2018). Nevertheless,
it appears that there is a subjective preference among visually impaired
readers for sans serif fonts, similar to sighted readers (Russell-Minda
et al., 2007).
Few studies have been conducted on the specic typographic needs
of visually impaired children. Yet, several authors have pointed out that
the readability requirements for visually impaired children are not the
same as those for adults (Alabdulkader & Leat, 2010; Bessemans, 2016).
In general, visually impaired children need more time to read and un-
derstand text (Gompel et al., 2004). For most children, the primary
barrier to reading is font size. Children with low vision usually need
some magnication for text to reach their readability threshold. Farmer
and Morse (2007) suggest that children would benet more from using a
magnifying glass than an enlarged type. However, this idea is not shared
by other authors, who recommend enlarged types (Bessemans, 2016;
McLeish, 2007). Finally, an increase in inter-letter space inuences
reading speed, especially for slower readers. This increase reduces the
critical point size (McLeish, 2007).
No fonts specically designed to meet the reading needs of visually
impaired people have been developed. Only Eido was designed specif-
ically for older people with central eld loss (e.g. Age-Related Macular
Degeneration people).
The main aim of this study is to examine the readability of a new
font, Luciole,
1
designed by Centre Technique R´
egional pour la
D´
ecience Visuelle (CTRDV PEP69/ML) and the studio typographies.fr,
in collaboration with the developmental psychology laboratory (UR
DIPHE) of Universit´
e Lumi`
ere Lyon 2 (France). Luciole has an evidence-
based design, but it is also supported by the observations of professionals
(ophthalmologists, orthoptists, Braille Transcribers, typographers, etc.)
involved in the eld of visual impairment. The focus group method was
used to identify the most important typographic criteria for visually
impaired readers. Luciole font has been designed around a set of specic
design criteria to provide more comfortable reading for visually
impaired people. Finally, 10 criteria were selected to design Luciole
font: Sans serif, Open counterspace, Ample interletter spacing, distinc-
tiveness Glyph, Narrow width, Heavy diacritics, Large stem weight, Low
stroke modulation, High weight contrast and 4 variants of the font
(Regular, Italic, Bold, Bold Italic).
The typographic characteristics listed below as well as the ease of
use, the simplicity of the license, and the creation of an extended
character set (linguistic and mathematical support) suggest that Luciole
can potentially be a good font for readers with low vision. Luciole
contains >700 characters in each variant, as well as Greek and mathe-
matical symbols. Finally, Luciole is accessible free of charge to visually
impaired people and professionals.
1
https://luciole-vision.com/.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
3
Due to the existence of a specic font for visual impairment related to
aging and numerous studies into the visual requirements for reading in
adults with low vision, this rst study focused on readers of school age
and young adults. Three main hypotheses can be formulated. The rst
hypothesis is that the design of Luciole improves readability and
comprehension for people with low vision.
Secondly, previous research has shown that age is the strongest
predictor of reading performance in visually impaired children (Lovie-
Kitchin et al., 2001; Nersveen et al., 2018) and the level of reading
expertise also has an impact on the readability needs (Alabdulkader &
Leat, 2010; Bessemans, 2016). For this reason, we expect performance to
differ based on the level of reading expertise (French national
education).
Thirdly, the scientic literature suggests a number of similarities
between the results of normal vision and low vision readers (Legge et al.,
1985). For example, Luciole is a sans serif font and several studies
indicate that this characteristic has a positive inuence on reading
speed, especially for children of school age. This consistency leads us to
make a last hypothesis: a font designed primarily for low vision reading
might also improve the readability of normal vision readers.
2. Materials and methods
2.1. Participants
73 French readers with low vision (LV: 45 male and 28 female) aged
6 to 35 years old (M =13.7; SD =5.9), without associated pathologies
that could affect their reading ability, participated in this study. LV
group includes category 1 or 2, which corresponds to moderate or severe
according to the International Classication of Diseases 11th Revision
(WHO, ICD 11, 2022). Visual pathologies were reported by the children's
parents, or by themselves if the participant was an adult. 14.6 %
declared a pathology concerning damage to the optic nerve, 46 % the
retina, 9.4 % the anterior segment, 21.6 % had nystagmus and in 8.1 %
the pathology was unknown or undetermined.
72 readers with normal vision participated in this study (NV: 25 male
and 47 female), aged 6 to 34 years old (M =15.7; SD =7.7).
Children with LV were recruited in special centers for visual
impairment in France. Adult participants with LV were recruited from
the Valentin Haüy Association (AVH), the ofce of disability services of
Lyon 3 University, and the physiotherapist school for visually impaired
people (IFMKDV) in Lyon, France. The sighted children were selected
from a regular primary school and had normal or corrected to normal
vision. Adult sighted participants are students from the University of
Lyon. For the children with visual impairment, we asked the parent to
declare the pathology, the degree of visual impairment and the existence
of reading impairment. Adult participants answered the same questions.
In order to take into account, the level of expertise of the readers, we
formed four different subgroups, according to the French Education
system: Beginner Group, cycle 2 from CP, CE1 and CE2 (from 1st to 3rd
US grade), Intermediate Group, cycle 3 from CM1 to 6
`
eme
(from 4th to
6th US grade), Advanced Group, cycle 4 from 5
`
eme
to 3
`
eme
(from 7th to
10th US grade), and Expert Group composed of high school students and
young adults (Table 1). The youngest participants in CP class of Beginner
Group (1st US grade), were tested at the end of the school year.
This research respects ethical principles for research involving
human subjects (World Medical Association, 1964). Parents of minor
children were informed of the purpose of the study and how their
children's data would be used in accordance with ethical principles.
Participants aged 18 and over gave their informed consent directly. The
experiments were carried out after accepting the conditions of use of the
data according to the European standard (General Data Protection
Regulation-GDPR).
2.2. Materiel
Two stimuli were developed for this study: texts and false-words.
And 5 fonts were chosen for comparison with the Luciole font.
2.2.1. Texts conception
The texts were created using the database Manulex, a word fre-
quency list based on a corpus of readers used in French primary schools,
from 1st to 5th grades. This database allows words to be selected ac-
cording to their standard frequency index (SFI) by grade level, as well as
the number of letters in a word (L´
et´
e et al., 2004). The texts were
Table 1
Distribution of participants by reading expertise group.
Groups Low vision (LV) Normal vision (NV)
N Year month
(mean SD)
N Year month
(mean SD)
Beginner Group (6–9
years)
13 8.0 (1.1) 20 8.1 (1.1)
Intermediate Group (9–12
years)
20 10.4 (1.0) 15 10.2 (1.1)
Advanced Group (12–16
years)
27 14.2 (1.0) 10 13.3 (1.2)
Expert Group (16–35
years)
13 23.6 (6.1) 27 23.2 (3.5)
Total 73 13.7 (5.8) 72 15.4 (7.4)
Table 2
Characteristics of the six fonts type.
Arial Verdana Eido OpenDyslexic Frutiger Luciole
Free license ✓ ✓ ✓ – – ✓
Bold ✓ ✓ – ✓ ✓ ✓
Italic ✓ ✓ – ✓ ✓ ✓
Greek
alphabet
fonts
✓ ✓ ✓ Not specied – ✓
Font Math
Symbol
– – ✓ Not specied – ✓
Font legibility
for low
vision
– – ✓ – – ✓
Fig. 1. Description of the Arial, Verdana, Eido and OpenDyslexic, Frutiger and
Luciole fonts used in our experiments (x-height normalized).
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
4
developed from the 1500 most frequent words for each school level. The
texts consisted of 6 sentences, each sentence corresponding to a font.
The texts differed in content and length according to the reading
expertise group (number of words per text: 4, 5, 11, 65, 75 depending on
the expertise group). The texts contain couples of complex letters in
French: “cl”, “il”, “rm”, “ff” (“ff” except for Beginner Group) and
numbers.
To check that the text was understood, we used literal questions, 6
comprehension questions for each reading expertise group. For example:
“Grand-m`
ere a termin´
e sa leçon. Elle fait un spectacle demain” (Grand-
mother has nished her lesson. She is doing a show tomorrow).
Comprehension question: Que va faire la grand-m`
ere demain ? (What will
the grandmother do tomorrow?).
2.2.2. False-word conception
The 48 false-words were generated using a generator
2
of false-words
with the same criteria used for the texts.
2.2.3. Comparison between Luciole and the 5 other fonts
Luciole was compared to ve other sans serif fonts
3
(Arial, Frutiger,
Verdana, OpenDyslexic et Eido) (Table 2). The rst two, Arial and
Verdana, are often chosen by visually impaired people when adapting
written documents (Drummond et al., 2004). The second, Eido, is a font
created for people with macular degeneration, a disease that impairs the
central visual eld. The letters in this font were designed to reduce the
physical similarities between letters, reduce their visual complexity, and
design letters with which observers are familiar. The limit of Eido is that
it lacks bold and italics variants (Bernard et al., 2016). The third,
OpenDyslexic, is a font created to improve readability for readers with
dyslexia that we considered relevant to compare because the letters are
more differentiated compared to regular fonts. The fourth, Frutiger, is a
sans serif font developed to improve the specic legibility of airport
signage. It is currently very popular in magazines and brochures because
its shapes are designed to ensure that each individual font is quickly and
easily recognized.
Fig. 2. Mean rank of the Preference Rating for low vision (LV) and normal vision (NV) participants and by Groups (reading texts on paper).
Table 3
Means and standard error of mean (SEM) in reading time (s) for each group (reading expertise) for LV and NV.
Font LV1 - Beginner
Group (N =11)
NV1 - Beginner
Group (N =20)
LV2 - Intermediate
Group (N =18)
NV2 - Intermediate
Group (N =15)
LV3 - Advanced
Group (N =26)
NV3 - Advanced
Group (N =10)
LV4 – Expert
Group (N =
13)
NV4 – Expert
Group (N =
26)
Luciole 17.3 (18.3) 3.6 (1.6) 12.2 (9.3) 5.0 (1.7) 29.4 (10.5) 19.2 (2.5) 28.2 (4.1) 26.9 (4.2)
Verdana 15.7 (16.3) 3.7 (1.8) 10.5 (6.8) 5.2 (2.5) 30.4 (11.3) 19.4 (2.4) 27.8 (4.7) 27.9 (4.3)
Frutiger 12.4 (10.3) 3.9 (2.0) 10.9 (7.0) 4.5 (1.1) 29.4 (10.7) 18.6 (1.8) 28.1 (4.4) 27.5 (4.8)
Arial 10.4 (9.3) 4.0 (2.2) 10.2 (6.4) 5.3 (1.8) 28.8 (10.7) 19.3 (2.7) 28.1 (3.7) 27.8 (4.4)
OpenDyslexic 13.3 (11.8) 3.3 (1.9) 11.7 (7.5) 4.6 (1.3) 32.0 (10.9) 19.4 (2.5) 29.3 (5.5) 27.6 (4.1)
Eido 25.8 (25.8) 3.9 (2.0) 13.7 (8.5) 4.4 (0.8) 35.8 (15.9) 21.0 (2.7) 32.2 (6.5) 28.4 (5.1)
In bold font values that produced less reader time.
2
http://www.fredericboutot.com/.
3
Font is the term used here as a synonym for typeface because it is the
commonly used term outside of the typography world.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
5
To control for the real size of letters in a given font and to account for
font size discrepancies a normalization process has been employed. For
all fonts, the same point size effect has been neutralized, i.e. the geo-
metric heights (x-height) within a given point size have been aligned (for
x-height normalization see Wallace et al., 2022). For example, Arial size
31 point will correspond to Verdana size 29.5 point.
Luciole is the only font that meets the six criteria summarized in the
Table 2.
2.3. Procedure
The methodology used is inspired by Rello and Baeza-Yates (2016)
and Sparrow et al. (2016). In experiment 1, six texts were presented on
paper books in six fonts (Fig. 1). In experiment 2, false-words were
displayed on a screen of 24′′ with a resolution of 1024–768 pixels, and
eye movement characteristics during reading were recorded by Tobii
pro-spectrum (driven with software Tobii pro Lab). False-words are used
to control the visual decryption of each letter, the reader being unable to
Fig. 3. Reading Time box plots by Font Type for LV and NV groups.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
6
refer to their lexical eld for help or correction.
For each experiment, participants chose their comfortable reading
size: 12, 14, 16, 20, 24, 28, 34, 40. Afterwards, all texts and false-words
were presented with the same size for the tests on the 6 fonts. A within-
subject repeated-measures design was used to evaluate the font effect on
the reading for each group (LV and NV group).
To judge readability, we considered the objective measures reading
time (paper), reading accuracy (paper and screen) and xation time
(screen).
– Reading time: For quantifying readability, we used reading time (in
s) for reading text on paper but not on screen because the length of
the false-words was too short (six letters per false word).
– Reading accuracy: For this variable, we measured the number of
reading errors. A Reading error (paper and screen) is retained when
the phoneme does not correspond to the grapheme or the word is not
read according to the rules of reading in the French language.
Reading errors were coded 1 if error or 0.5 if the participant self
corrects. In addition, we asked the comprehension questions for the
texts and we scored 2 for a correct answer, 1 for incomplete or
partially correct answer, and 0 for wrong answers or no answer. We
use the comprehension questions as a control variable only to ensure
that the participants have understood the texts read (see Rello &
Baeza-Yates, 2016).
– Fixation duration: this variable denotes how long the eye rests still on
a single place of the text and we use the mean of the xation duration
obtained by the eye tracker. Previous research shows that shorter
xations are associated with better legibility (Rello & Baeza-Yates,
2016).
Finally, to collect the participants' preferences, we used subjective
Preference Ratings at the end of reading texts and false-words.
2.3.1. Experiment 1. Reading texts on paper
All participants (73 LV and 72 NV) took part in this rst experiment.
The subject, seated in front of a table, was free to manipulate the texts in
order to put them at the most comfortable reading distance. Subse-
quently, the participant was asked to keep the same distance while
reading the six texts. The subject was requested to read aloud the rst of
the six texts (in a chosen font size). The different texts were then pre-
sented in one of six sequences dened by a Latin square, in the same font
size. This is an experimental design in which fonts are administered in
sequences that are systematically varied so that each font appears with
the same frequency in each position of the sequence (rst, second, third,
etc.).
Next, the experimenter removed the text and asked the subject to
answer a comprehension question. The procedure (reading and
comprehension questions) was repeated for the other ve texts (in the
same font size, each text being printed in a different font). A short pause
was taken after the third text. Finally, the experimenter gave all six texts
back to the subject, in the same fonts and size in which they had read
them, and asked them to rank the fonts in order of subjective preference.
The reading of each text was recorded to facilitate data processing.
Reading time was measured when the results were tallied. Reading
duration was timed using a stopwatch. The number of reading errors was
recorded.
2.3.2. Experiment 2. Reading false-word in screen
The use of eye tracking is applicable with low vision but the cali-
bration was not possible for all the participants (Pel & Dalhoi, 2017;
Tatsumi et al., 2010). Thus, 68 LV participated in this second experi-
ment, where 42 of them were calibrated, and 71 NV, where 66 were
calibrated. Even if the calibration was not possible, the participant
continued the experiment to give their preference font on screen. The
number of reading errors was observed, but not xation duration.
The distance between the participant and the eye tracker was con-
stant (approximately 60 cm). A second camera was used to record verbal
responses. The participant was asked to choose the font size in which he/
she wanted to read on the screen. Each false word was displayed on the
screen as soon as the participant had nished reading aloud the word.
The false-words appeared in all six fonts in a Latin square order. When
all the false-words had been read, the French word “famille” (family)
written in the six fonts was displayed and the participant had to rank
them in order of subjective preference.
2.4. Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics Soft-
ware with the standard condition of p <.05 for signicant results. A
Shapiro-Wilk test showed that the data were not normally distributed.
Consequently, to investigate the signicant effects of Font Type on
subjective preferences, reading time, errors, and xation time, we used
Friedman's nonparametric test for repeated measures, as well as a
complete pairwise Dunn rank sum post-hoc comparison test with a
Bonferroni correction (Bonferroni, 1936) that includes the adjustment of
the signicance level. We chose this test in order to reduce the chances
of obtaining false-positive results (type I errors).
3. Results
3.1. Experiment 1. Reading texts on paper
3.1.1. Choice of font size
80.8 % of LV readers choose to read in one of the four largest sizes
(24 to 40), while 79.2 % of NV readers choose one of the rst four (12 to
20). The proportion of NV readers choosing one of the rst three sizes
(12 to 16) increases with age: 40 % in NV1 Beginner Group, 46.7 % in
NV2 Intermediate Group, 60 % in NV3 Advanced Group and 88.9 % for
NV4 Expert Group. This is not observed for LV participants. Consistent
with the literature, we observe a preference for larger characters in
participants with low vision compared to participants with normal
vision (Legge et al., 1985; Rubin et al., 2006).
Table 4
Means percent and standard error of mean (SEM) of reading errors/words for All participants (LV and NV) and Groups.
Font LV - All
participants
(N =71)
NV - All
participants
(N =72)
LV1 -
Beginner
Group (N
=12)
NV1 -
Beginner
Group (N
=20)
LV2 -
Intermediate
Group (N =20)
NV2-
Intermediate
Group (N =15)
LV3 -
Advanced
Group (N =
26)
NV3 -
Advanced
Group (N =
10)
LV4 -
Expert
Group
(N =13)
NV4 -
Expert
Group
(N =27)
Luciole 3.8 (5.0) 3.1 (5.0) 10.4 (12.2) 9.5 (13.3) 2.00 (3.3) 0.9 (1.6) 3.2 (3.5) 0.7 (0.8) 1.4 (1.7) 0.5 (0.6)
Verdana 4.2 (5.2) 1.7 (2.6) 11.3 (15.2) 4.8 (7.1) 3.00 (4.1) 0.6 (1.1) 3.3 (3.0) 0.5 (0.6) 1.3 (1.4) 0.4 (0.6)
Frutiger 4.6 (6.3) 2.6 (4.1) 13.8 (15.0) 7.5 (11.3) 3.6 (5.1) 0.6 (1.1) 3.1 (3.7) 0.6 (0.9) 0.8 (0.8) 0.7 (0.8)
Arial 3.7 (4.9) 4.0 (5.9) 10.0 (13.3) 12.0 (13.6) 2.3 (3.0) 1.8 (2.4) 3.4 (3.7) 0.4 (0.6) 0.8 (0.9) 0.5 (0.7)
OpenDyslexic 3.9 (5.2) 1.7 (2.9) 10.8 (14.4) 5.3 (8.4) 3.2 (4.1) 0.6 (1.1) 2.9 (3.1) 0.2 (0.2) 0.9 (1.1) 0.3 (0.5)
Eido 7.5 (10.9) 2.8 (4.6) 26.3 (25.6) 8.8 (12.3) 5.0 (7.5) 0.9 (1.6) 4.1 (5.2) 0.6 (0.7) 0.7 (0.7) 0.4 (0.5)
In bold are the values of the fonts which produced less error.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
7
Fig. 4. Reading errors/words (percentage) Box plots by Font type for LV and NV groups.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
8
3.1.2. Preference ratings
144 participants indicated their subjective preferences for the fonts.
Luciole font obtained the rst place of subjective preference (average
rank) for LV All participants (i.e. 52 % preference) (Fig. 2). If we look at
the reading expertise groups, for the LV participants, Luciole is the fa-
vorite font for all Groups, except LV1 Beginner Group (respectively by
30,8 % LV1 Beginner Group, by 47.4 % of LV2 Intermediate Group, by
59.3 % of the LV3 Advanced Group and by 69.2 % of the LV4 Expert
Group). For the LV2 Intermediate Group and LV3 Advanced Group,
Luciole is tied with Frutiger. For the NV All participants Luciole is
ranked second (i.e. 29.2 % preference) tied with Verdana and achieved
the best average rank only for LV1 Beginner Group (i.e. 45 %
preference).
All participants. There was a signicant effect of font type on
subjective preference ratings for the LV participants (
χ
2
F(5) =182.127, p
<.05) and for NV participants (
χ
2
F(5) =129.811, p <.05). The results of
the post-hoc tests show that for both groups (all LV and NV) Luciole is
signicantly preferred over OpenDyslexic (p <.001) and Eido (p <
.001). The subjective preference is not signicant with Arial, Verdana
and Frutiger (p >.05).
Groups. Regarding the results of different reading, for young readers
(Beginner Group) the six fonts are not equivalent in terms of preference
(LV:
χ
2
F(5) =16.984, p <.05; NV:
χ
2
F(5) =36,371, p <.05). Luciole was
signicantly preferred over Eido (p <.05) by children in group LV but
not over the other fonts. In the NV group, Luciole was signicantly
preferred over OpenDyslexic (p <.05) and Eido (p <.001). For the
Intermediate Group, the fonts are signicantly different for LV partici-
pants (
χ
2
F(5) =58.466, p <.05), and for NV (
χ
2
F(5) =22.067, p <.05).
Luciole is signicantly preferred over Eido (p <.001) and OpenDyslexic
(p <.001) by LV, and only over Eido for NV (p <.001). For LV partic-
ipants in the Advanced Group, there is an effect of the font for the
preference ratings (LV:
χ
2
F(5) =69.187 p <.05; NV:
χ
2
F(5) =32.000, p <
.05). Luciole is signicantly preferred by LV participants over Eido (p <
.001) and OpenDyslexic (p <.001), and only signicantly preferred over
Eido (p <.001) for NV participants. Finally, for Expert readers there
were signicant differences between the 6 fonts for both groups (LV:
χ
2
F(5) =53.929 p <.05; NV:
χ
2
F(5) =51.415, p <.05). For LV partici-
pants, Luciole is signicantly preferred to Eido (p <.001) and Open-
Dyslexic (p <.001) and for NV participants, Luciole is signicantly
preferred over Eido (p <.001).
3.1.3. Reading time on paper by group
Because text size differs by reading expertise group, the analysis on
All Participants was not retained. Overall, reading times with Luciole
font are relatively high. It was Arial that allowed texts to be read more
quickly in Beginner, Intermediate and Advanced Groups, although not in
the Expert Group (Table 3).
On the contrary, Luciole produces low reader time in NV Expert
group (M =26,9). Globally, the reading time for the six fonts was higher
for all LV Groups compared to all NV Groups. Similarly, the intra-group
Fig. 5. Mean rank of the Preference Rating for low vision (LV) and normal vision (NV) participants and by Groups (reading texts on screen reading).
Table 5
Means and standard error of mean (SEM) of reading errors for all participants (LV and NV) and Groups.
Font LV - All
participants
(N =67)
NV - All
participants
(N =70)
LV1 -
Beginner
Group (N
=11)
NV1 -
Beginner
Group (N
=19)
LV2 -
Intermediate
Group (N =19)
NV2 -
Intermediate
Group (N =15)
LV3 -
Advanced
Group (N =
25)
NV3 -
Advanced
Group (N =
10)
LV4 -
Expert
Group
(N =12)
NV4 -
Expert
Group
(N =26)
Luciole 1.4 (1.4) 0.9 (0.8) 2.2 (1.6) 1.3 (1.1) 1.5 (1.3) 0.9 (0.6) 1.5 (1.2) 0.7 (0.8) 0.4 (0.5) 0.6 (0.7)
Verdana 1.7 (1.5) 1.0 (0.8) 2.5 (1.4) 1.7 (0.9) 1.8 (1.5) 1.0 (0.8) 1.7 (1.6) 0.5 (0.5) 0.8 (0.8) 0.6 (0.6)
Frutiger 1.5 (1.4) 0.8 (0.9) 2.0 (1.3) 1.7 (1.3) 1.7 (1.4) 0.5 (0.6) 1.2 (1.2) 0.4 (0.5) 1.2 (1.2) 0.5 (0.7)
Arial 1.8 (1.3) 0.8 (0.7) 2.4 (1.6) 1.3 (1.0) 1.9 (1.2) 0.7 (0.7) 1.7 (1.1) 0.4 (0.6) 1.2 (1.4) 0.7 (0.6)
OpenDyslexic 2.6 (1.7) 0.8 (0.7) 2.5 (1.5) 1.2 (1.1) 2.7 (1.8) 0.8 (0.6) 2.9 (1.7) 0.5 (0.4) 1.9 (1.4) 0.6 (0.6)
Eido 2.0 (1.5) 0.9 (0.8) 2.3 (1.7) 1.6 (1.0) 2.1 (1.4) 0.9 (0.9) 2.0 (1.6) 0.5 (0.5) 1.4 (1.7) 0.5 (0.6)
In bold the font that generated less error.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
9
Fig. 6. Reading errors for All participants (LV and NV) and Groups (screen).
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
10
variability was greater in the LV Groups (Fig. 3).
Groups. For LV group, there was a signicant effect of Font Type on
Reading Time for LV2 Intermediate Group (
χ
2
F(5) =14.222, p <.05) and
LV3 Advanced Group (
χ
2
F(5) =18.879, p <.05) and LV4 Expert Group
(
χ
2
F(5) =13.132 p <.05) but not for LV1 Beginner Group (
χ
2
F(5) =1.857,
p =.869). The results of the post-hoc tests show a signicant difference
between Luciole and Eido (p <.001) for LV2 Intermediate Group, (p <
.05) but there are no signicant differences for the other Groups.
In contrast, no signicant effect of font type on Reading Time was
found for NV groups: NV1 Beginner Group (
χ
2
F(5) =7.629, p =.178),
NV2 Intermediate Group (
χ
2
F(5) =5.895, p =.317), NV3 Advanced
Group (
χ
2
F(5) =7.771, p =.169), and NV4 Expert Group (
χ
2
F(5) =7.033,
p =.218).
3.1.4. Reading accuracy
3.1.4.1. Reading errors. Overall, very few errors were produced in
Experiment 1. Beginner Groups LV1 and NV1 produce the most errors
(Table 4). Luciole did not differ from other fonts except for Intermediate
LV2 Group. For this group, it is the font that produced the least reading
errors. The intragroup variability is also greater for the LV group (Fig. 4).
All participants. We did not nd a signicant effect of fonts on the
percentage of reading errors in two groups (LV:
χ
2
F(5) =3.811, p =.577;
NV:
χ
2
F(5) =8.270, p =.142).
Groups. We observed the same trend for the expertise groups, i.e.,
there is no effect of the font on the number of errors produced according
to the level of expertise of the readers: for the Beginner Group (LV:
χ
2
F(5)
=8.103, p =.151; NV:
χ
2
F(5) =5.428, p =.366), Intermediate Group
(LV:
χ
2
F(5) =1.156, p =.949; NV:
χ
2
F(5) =5.201, p =.392), Advanced
Group (LV:
χ
2
F(5) =6.449, p =.265; NV:
χ
2
F(5) =2.300, p =.806) and
Expert Group (LV:
χ
2
F(5) =0.678, p =.984; NV:
χ
2
F(5) =4.709, p =.452).
3.1.4.2. Comprehension questions. All participants. A large majority of
subjects answered the comprehension questions correctly for all the
fonts studied. There were no differences between the fonts for the LV
(
χ
2
F(5) =6.089, p =.298) and for the NV participants (
χ
2
F(5) =6.972, p
=.223).
Groups. No effect of fonts on comprehension was found for all
reading expertise groups Beginner Group (LV:
χ
2
F(5) =6.026, p =.304;
NV:
χ
2
F(5) =3.462, p =.629), Intermediate Group (LV:
χ
2
F(5) =7.564, p
=.182; NV:
χ
2
F(5) =3.760, p =.584), Advanced Group (LV:
χ
2
F(5) =
7.831, p =.166; NV:
χ
2
F(5) =2.300, p =.806) and Expert Group (LV:
χ
2
F(5) =5.000, p =.416; NV:
χ
2
F(5) =4.218, p =.518).
3.2. Experiment 2. Reading false-words on screen
3.2.1. Choice of font size
The eye tracker calibration was achieved for 42 LV participants and
66 NV participants. However, for reading preference and screen reading
accuracy, we analyzed responses of all participants independently of eye
tracker calibration success. When reading from the screen, 68.6 % of the
LV participants chose to read at a size between 24 and 40, while 78.9 %
of the NV subjects chose a size between 12 and 20. We observe here the
same trend as for reading text on paper.
3.2.2. Preference ratings
The Preference rating was assessed independently of the success of
the eye-tracking calibration. A total of 68 LV and 71 NV participants
gave their preference. For on-screen reading, the mean rank conrms
that Luciole was the most popular font by LV all participants (i.e. 47,9 %
preference) and in all Groups (respectively by 54.5 % for LV1 Beginner
Group, 52.6 % for LV2 Intermediate Group, 37.9 % for LV3 Advanced
Group and 58.3 % for LV4 Expert Group) (Fig. 5). For NV All partici-
pants, Luciole was ranked fourth (i.e. 17,3 % preference) and got the
best rank only for NV3 Advanced Group (i.e. 30 % preference).
All participants. The effect of the font on preference is signicant
for LV participants (
χ
2
F(5) =142.608, p <.050) and NV (
χ
2
F(5) =
114.925, p <.05). Pairwise post-hoc comparisons showed signicant
differences between Luciole and Eido (p <.001), OpenDyslexic (p <
.001) and Arial (p <.05) for LV participants. For NV participants Luciole
does not produce signicant differences with other fonts.
Groups. The differences are also signicant for LV readers for LV2
Intermediate Group (
χ
2
F(5) =45.448, p <.050), LV3 Advanced Group
(
χ
2
F(5) =59.973, p <.05), and LV4 Expert Group (
χ
2
F(5) =39.245, p <
.05) but not on LV1 Beginner Group (
χ
2
F(5) =7.992, p =.157). For the
NV participants, there was a signicant effect on all expertise Groups:
NV1 Beginner Group (
χ
2
F(5) =15.943, <.05), NV2 Intermediate Group
(
χ
2
F(5) =33.933, <.050), NV3 Advanced Group (
χ
2
F(5) =35.886, p <
.050), and NV4 Expert Group (
χ
2
F(5) =49.652, p <.05).
Regarding LV participants, Luciole was signicantly preferred over
Eido for LV2 Intermediate Group (p <.001), LV3 Advanced Group (p <
.001) and LV4 Expert Group (p <.001). Luciole was also preferred over
OpenDyslexic for LV2 Intermediate Group (p <.001), LV3 Advanced
Group (p <.001), and LV4 Expert Group (p <.05). As regards the NV
Groups, Luciole was signicantly preferred over Eido for NV2 Interme-
diate Group (p <.05), NV3 Advanced Group (p <.001), and NV4 Expert
Group (p <.01) and over OpenDyslexic for NV3 Advanced Group (p <
.05).
Table 6
Means and standard error of mean (SEM) of xation duration (s) in screen reading for all participants (LV and NV) and Groups.
Font LV - All
participants
(N =42)
NV - All
participants
(N =66)
LV1 -
Beginner
Group (N
=5)
NV1 -
Beginner
Group (N
=17)
LV2 -
Intermediate
Group (N =16)
NV2 -
Intermediate
Group (N =14)
LV3 -
Intermediate
Group (N =11)
NV3 -
Advanced
Group (N =
10)
LV4 -
Expert
Group
(N =10)
NV4 –
Expert
Group
(N =25)
Luciole 10.4 (10.0) 13.6 (13.0) 15.3 (9.7) 17.3 (6.1) 10.3 (6.9) 14.7 (2.9) 11.0 (6.7) 12.6 (2.2) 7.6 (4.3) 10.9
(1.5)
Verdana 11.5 (9.5) 13.8 (13.0) 20.2 (14.1) 16.6 (5.3) 10.9 (7.4) 16.4 (2.9) 11.9 (7.0) 12.5 (2.1) 7.7 (4.6) 11.0
(1.7)
Frutiger 10.4 (9.7) 14.0 (13.1) 18.1 (14.0) 17.3 (4.1) 10.5 (6.8) 15.3 (1.9) 10.2 (6.4) 14.1 (4.5) 6.6 (4.3) 11.1
(1.6)
Arial 11.6 (9.5) 13.9 (13.0) 18.4 (11.0) 16.8 (6.0) 12.1 (8.1) 15.5 (2.6) 11.6 (7.2) 12.2 (1.9) 7.6 (5.0) 11.6
(2.3)
OpenDyslexic 11.2 (8.7) 13.5 (13.1) 18.8 (14.5) 16.2 (6.5) 12.1 (8.0) 15.6 (2.5) 11.0 (7.3) 12.5 (2.1) 6.0 (3.7) 10.8
(1.4)
Eido 12.6 (9.5) 14.2 (13.0) 22.0 (16.1) 18.6 (6.6) 12.1 (8.1) 15.6 (2.9) 12.3 (7.0) 12.1 (2.3) 9.0 (4.7) 11.3
(1.5)
In bold the font that generated less xation duration.
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
11
Fig. 7. Fixation duration (s) box plots for all participants (LV and NV) and Groups (screen).
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
12
3.2.3. Reading accuracy
Reading False-words on screen with Luciole generates fewer errors in
participants with a LV, except for in the LV3 Advanced Group (Luciole is
behind Frutiger) (Table 5). However, the intragroup variability is
greater in the LV group than in the NV group (Fig. 6). For NV partici-
pants, it is Frutiger which produced on average less reading errors in
NV2 Intermediate Group; this is also the case for the NV3 Advanced
Group results but with error levels equal to those of the Arial font. For
NV All participants the reading error rates are also equal to those of two
other fonts: Arial and OpenDyslexic. Finally, OpenDyslexic obtained the
lowest error average in NV1 Beginner Group.
Given that the number of false-words was the same for all partici-
pants of experiment 2, the calculation of the average was done for all
groups, including for All participants Group (LV and NV).
All participants. There was a signicant effect of Font Type on
reading errors (
χ
2
F(5) =34.399, p <.05) for LV participants. The results
of the post-hoc tests show that Luciole produces signicantly fewer er-
rors than OpenDyslexic (p <.005), whereas for other fonts the differ-
ences are smaller. On the other hand, there is no effect of the font on
reading errors in NV participants (X
2
(5) =2.883, p =.718).
Groups. Regarding LV participants, we did nd a signicant effect
for LV2 Intermediate Group (X
2
(5) =11.399, p <.05), LV3 Advanced
Group (
χ
2
F(5) =17.236; p <.05) and LV4 Expert Group (X
2
(5) =11.962;
p <.05). Pairwise post-hoc comparisons showed no signicant differ-
ences between Luciole and the other fonts. Finally, we did not nd a
signicant effect for LV1 Beginner Group (
χ
2
F(5) =3147, p =.677).
The number of errors when reading on screen is not signicantly
different for people with NV: all participants (
χ
2
F(5) =2.883, p =.718),
NV1 Beginner Group (
χ
2
F(5) =4.751, p =447), NV2 Intermediate Group
(
χ
2
F(5) 6.529, p =.258), NV3 Advanced Group (
χ
2
F(5) =0.684, p =.984),
and NV4 Expert Group (
χ
2
F(5) =5.914, p =.315).
3.2.4. Fixation duration
Table 6 shows that Luciole generates a lower mean duration of x-
ations than other fonts for the LV participants (except for LV3 Advanced
Group and LV4 Expert Group). For all participants LV, Luciole produced
the same average xation duration as Frutiger (M =10,4). For the NV
participants, Luciole produced the lowest average xation duration only
in NV2 Intermediate Group (M =14,7). OpenDyslexic produced less
xation duration in three subgroups (all participants, NV1 Beginner
Group and NV4 Expert Group), and Eido for NV3 Advanced Group.
Finally, xation durations are systematically shorter for LV participants
compared to NV participants for all fonts (Fig. 7).
All participants. We found signicant effect of Font Type on Fixa-
tion duration for LV participants (
χ
2
F(5) =21.732, p <.05) but not for
NV participants (
χ
2
F(5) =9.853, p =.08). The post-hoc test for LV par-
ticipants showed signicant differences uniquely between Luciole and
Eido (p <.05).
Groups. For LV participants, there was a signicant effect of Font
Type on Fixation duration only for LV4 Expert Group (
χ
2
F(5) =11,429, p
<.05). The differences for the other Groups are not signicant: LV1
Beginner Group (
χ
2
F(5) =6.257, p =.282), LV2 Intermediate Group
(
χ
2
F(5) =9.848, p =.08) and LV3 Advanced Group (
χ
2
F(5) =7.312, p =
.198). For LV4 Expert Group, the Fixation duration reading with Luciole
is not signicantly different from other fonts (p >.05).
Finally, we did not nd a signicant effect of Font Type on Fixation
duration for NV Groups: NV1 Beginner Group (
χ
2
F(5) =7.857, p =.164),
NV2 Intermediate Group (
χ
2
F(5) =7.796, p =.168), NV3 Advanced
Group (
χ
2
F(5) =0.971, p =.965), and NV4 Expert Group (
χ
2
F(5) =7.834,
p =.166).
4. Discussion
Text magnication, acuity reserve and contrast reserve are not
always sufcient for improving reading for people with LV. It has been
shown that certain characteristics of fonts can positively inuence
reading in people with LV. In this study, we wanted to nd out if using a
font specically designed for people with LV affects readability. In other
words, the goal was to investigate whether Luciole is a font that would
be perceived as more attractive or comfortable for visually impaired
readers but also for readers with normal vision. To do this, we rst
looked at the subjective preference of French readers. We considered the
French educational system as a criterion for reading prociency.
Around half of the participants with LV have a subjective preference
for Luciole when reading on paper and on a screen. When reading on
paper, Luciole is the preferred font among the 6 fonts, with the exception
of the Beginner Group. In this group, Luciole is second after Arial. When
reading on screen, Luciole obtained the best subjective preference for all
groups of reading expertise. However, the preference for Luciole is only
signicant when the font is compared to Eido and OpenDyslexic and
Arial.
Participants with NV prefer Frutiger (All participants) for reading on
paper. If we consider reading expertise, Luciole obtains the best average
rank in the Beginner Group but for the other groups, Frutiger obtained
the best subjective preference except for the Expert Group.
For screen reading, the preferences are quite variable from one group
to another. Finally, a trend emerges from the results: Luciole is signi-
cantly preferred over Eido and Open Dyslexic. We hypothesize that
Luciole's preference over these two fonts may be explained by the fact
that these fonts are not familiar to visually impaired young people nor to
participants with normal vision. Indeed, Eido is used with people
suffering of age-related macular degeneration (AMD) (Bernard et al.,
2016) and OpenDyslexic for dyslexic disorder (Rello & Baeza-Yates,
2016). Similarly, Luciole is a font that was little known in France
when we carried out the experiments. This could explain that it did not
produce differences with Arial and Verdana. Indeed, these two fonts are
widely used in texts adapted to the visually impaired.
Yet, it has been shown that subjective preferences do not necessarily
correspond to better performance (Legge et al., 1985; Legge & Rubin,
1986). One of the readability criteria is reading speed (Chung et al.,
2008; Manseld et al., 1996; Legge, 2016). Previous studies have shown
that faster reading is related to more readable texts (Williams et al.,
2003; Rello & Baeza-Yates, 2016). In this regard, Luciole does not allow
a signicantly faster reading (texts on paper) for the LV participants. The
only difference observed concerns the comparison with Eido for all
participants and the LV2 Intermediate Group. However, Luciole had no
effect on reading speed for NV participants. Our hypothesis that
Luciole's design has an effect on reading speed is not conrmed, but the
lack of a signicant effect of a font on reading time is recurrent in
readability studies (Arditi & Cho, 2007; Rubin et al., 2006; Tarita-Nistor
et al., 2013). This trend is also found in the studies on Eido font which do
not show signicant gain in reading speed when compared to other
fonts. In contrast, there is a signicant effect on the recognition of letters
and words (Bernard et al., 2016).
Likewise, to measure the reading accuracy, it is advisable to consider
the number of errors produced during reading (Legge, 2016). In our
study, we evaluated reading accuracy using two parameters, number of
errors while reading the text out loud and a comprehension question.
Reading accuracy for texts read on paper is not inuenced by the tested
fonts. In other words, Luciole did not increase or decrease errors in text
reading and did not impact text comprehension. Therefore, Luciole does
not seem to inuence reading accuracy. In response to these data, the
design of the texts may also partially explain this absence of effect.
Indeed, we used a rigorous methodology that took into account the
words most used by the children according to their school level (from 1st
to 5th grades). We have also controlled the complexity of the letters, by
removing complex letter couples and the length of the texts. Therefore,
the absence of differences could be due to the comfort of reading the
texts generated by these two criteria.
In contrast, the results show that reading false-words on screen is
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
13
inuenced by the fonts. Readers with LV made fewer errors on average
with Luciole for all participants and Groups (except LV3 Advanced
Group) but the only signicant difference was observed with Open-
Dyslexic (for All participants). However, the number of errors made by
NV participants is not related to fonts. This result shows that Luciole can
improve on-screen reading accuracy to a lesser extent for people with
LV, even if the effects are small.
Finally, it is observed that Fixation Duration is an indicator of
readability and that there is a relationship between short xations and
readability (Hy¨
on¨
a & Olson, 1995; Rayner & Duffy, 1986; Rello &
Baeza-Yates, 2016). The results of our study indicate an effect of font on
xation duration. Nevertheless, Luciole produced shorter xations only
in contrast to Eido for all participants with LV but no difference between
Luciole and other fonts for group NV. Moreover, the level of reading
prociency did not show any benet to using Luciole. Surprisingly, LV
participants produced shorter xation duration for all fonts. The only
explanation for this result is the size of the font used. Indeed, the LV
participants chose a larger size to read the texts, compared to the NV
participants. On the other hand, the intragroup variability is greater for
the LV participating, in particular for the less expert readers.
In summary, from a clinical point of view, Luciole is the only font
that offers the benet of open access and that includes all the typo-
graphic criteria that a visually impaired person needs to read text. The
results of the experiments indicate that Luciole is preferred over other
fonts. Measurements performed show a signicant effect compared to
Eido and OpenDyslexic but not compared to fonts usually used by people
with LV (Arial and Verdana).
5. Limitations and future work
Although our study is based on a design already used in studies that
compare several fonts, our study has some limitations. These limits
relate to: 1) the characteristics of the population of the LV group, 2) the
choice of font size, 3) the measurement of angular size, 4) the knowledge
of the fonts tested, and 5) reading aloud.
First, low vision is a highly heterogeneous category of visual
impairment because it can affect central and/or peripheral vision. This
heterogeneity of proles has been reported in many studies (Cass et al.,
1994). Also, the level of visual activity differs from one person to
another and can sometimes change over time. These criteria are difcult
to control, and a thorough visual examination by an ophthalmologist
could not be conducted for this study. Future studies may benet from
careful ophthalmological examination during the experiments to limit
effects of heterogeneity in low vision.
Second, we believe that leaving the choice of font size to the par-
ticipants may have inuenced the results. Especially since the partici-
pants with low vision chose a larger size than they would normally use.
It would be useful to control for this variable in future studies.
Third, following the second limitation, some studies with visually
impaired participants incorporate the measurement of angular size
which takes into account both the physical size of the characters and the
viewing distance (see Legge, 2016). While this measure is particularly
relevant when it comes to comparing different subjects, the use of
angular size seems less necessary in this research. As a matter of fact, the
different fonts were standardized for x-height and the participants chose
their most comfortable font size. When reading text on paper, it was not
possible to impose an equal distance for all participants. However,
within a participant, the same x-height and viewing distance were used
across all six fonts, resulting in a consistent angular size across fonts. If
comparison within a participant is valid here, the retinal size is repre-
sentative of how texts are processed by the visual system and may
provide a unied system for comparisons across studies. This may be a
limitation for future comparisons.
Also, Luciole was compared to some well-known fonts for people
with low vision, which could have impacted the results because of their
experience with these fonts. Since Luciole is a new type design, none of
the participants were familiar or experienced with the font, unlike
others like Arial or Verdana. The novelty of the characteristics of the font
may have had an impact in particular on the reading time. Therefore, we
believe that future studies should be carried out on readers who
frequently use Luciole.
Finally, the test paradigm could also carry another limitation. We
asked participants to read texts aloud in order to identify errors, but this
may have penalized some readers because speed is related to oral
articulation. However, practically it is more difcult to evaluate reading
speed when reading silently. Even with the adoption of comprehension
tests, it is difcult to ensure that the text read silently is actually read,
not just skimmed (Rubin, 2013). While silent reading is generally faster
than reading aloud, both forms of reading are equally affected by
changes in the type size (Chung et al., 1998) and are predicted by the
same clinical tests (Lovie-Kitchin et al., 2000).
For these reasons, we believe that the results of this study should be
viewed with caution.
6. Conclusion
For this study, we tested the effect of a new font designed specically
for individuals with LV on objectively measured readability and pref-
erences on paper and on screen. We tested this font also with sighted
participants. The main conclusion is that Luciole was preferred over the
other 5 fonts tested by people with LV for both tasks (paper and screen).
This trend was also observed when considering the level of reading
expertise of the participants, but Luciole was signicantly preferred to
fonts not often used by this population.
Finally, the objective measures (reading time, reading accuracy and
xation duration) show less effect of the Luciole font on readability. No
effect was found for the NV group.
Further studies should consider the limitations indicated above, in
particular the effect of visual pathologies, and also include an older
population.
Funding
This work was supported by the Ceres Foundation.
Declaration of competing interest
The authors declare no conict of interest.
Data availability
Data will be made available on request.
Acknowledgments
We would like to thank the participants who took part in the study
and our associative and socio-educational partners who helped us in the
constitution of the groups. In particular, Centre Technique R´
egional
pour la D´
ecience Visuelle CTRDV; also IFMKDV and S3AS (Association
PEP69), the Cit´
e Scolaire Ren´
e Pellet (EREA-DV), and CPHV (Fondation
Asile des aveugles). We would also like to thank Erell Collet, Master's
student, for her participation in the collection of data during her
research internship in the DIPHE laboratory (University of Lumi`
ere Lyon
2).
References
Alabdulkader, B., & Leat, S. J. (2010). Reading in children with low vision. Journal of
Optometry, 3(2), 68–73. https://doi.org/10.1016/S1888-4296(10)70010-8
Arditi, A., & Cho, J. (2007). Letter case and text legibility in normal and low vision.
Vision Research, 47(19), 2499–2505.
Arditi, A., & Cho, J. (2005). Serifs and font legibility. Vision Research, 45(23),
2926–2933. https://doi.org/10.1016/j.visres.2007.06.010
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
14
Arditi, A. (2004). Adjustable typography: An approach to enhancing low vision text
accessibility. Ergonomics, 47(5), 469–482. https://doi.org/10.1080/
0014013031000085680
Arditi, A. (1999). Making text legible. Available online under. Lighthouse International
http://www.lighthouse.org/accessibility/legible/.
Arditi, A., Cagenello, R., & Jacobs, B. (1995). Letter stroke width, spacing, and legibility.
In Technical digest series, Vol 1 -Vision Science and its applications (pp. 324–327). New
Mexico: Santa Fe.
Bernard, J. B., Aguilar, C., & Castet, E. (2016). A new font, specically designed for
peripheral vision, improves peripheral letter and word recognition, but not eye-
mediated reading performance. PloS one, 11(4). https://doi.org/10.1371/journal.
pone.0152506
Bernard, J. B., Kumar, G., Junge, J., & Chung, S. T. (2013). The effect of letter-stroke
boldness on reading speed in central and peripheral vision. Vision Research, 84,
33–42. https://doi.org/10.1016/j.visres.2013.03.005
Bernard, M. L., Chaparro, B. S., Mills, M. M., & Halcomb, C. G. (2002). Examining
children's reading performance and preference for different computer-displayed text.
Behaviour & Information Technology, 21(2), 87–96. https://doi.org/10.1080/
01449290210146737
Bernard, M., Liao, C. H., & Mills, M. (2001). The effects of font type and size on the
legibility and reading time of online text by older adults. In CHI'01 extended abstracts
on Human factors in computing systems (pp. 175–176). ACM.
Bessemans, A. (2016). Matilda: A typeface for children with low vision. In M. C. Dyson, &
C. Y. Suen (Eds.), Digital fonts and reading (pp. 19–36). Singapore: World Scientic.
https://doi.org/10.1142/9789814759540_0002.
Bigelow, C. (2019). Typeface features and legibility research. Vision Research, 165,
162–172. https://doi.org/10.1016/j.visres.2019.05.003
Bonferroni, C. E. (1936). Teoria statistica delle classi e calcolo delle probabilita. In , 8.
Pubblicazioni Dell’Istituto Superiore Di Scienze Economiche E Commerciali Di Firenze (pp.
3–62).
Bourne, R., Steinmetz, J. D., Flaxman, S., Briant, P. S., Taylor, H. R., Resnikoff, S.
Tareque, M. I., … (2021). Trends in prevalence of blindness and distance and near
vision impairment over 30 years: An analysis for the global burden of disease study.
The Lancet Global Health, 9(2), e130–e143. https://doi.org/10.1016/S2214-109X
(20)30425-3
Brozovi´
c, M., Dunđer, V., Alihodˇ
zi´
c, J., Milin, M., Bota, J., & Kovaˇ
cevi´
c, D. (2018).
Guidelines for designing text in printed media for people with low vision. Acta
graphica: znanstveni ˇ
casopis za tiskarstvo i graˇ
cke komunikacije, 29(1), 25–30.
Cass, H. D., Sonksen, P. M., & McConachie, H. R. (1994). Developmental setback in
severe visual impairment. Archives of Disease in Childhood, 70(3), 192–196.
Cornelissen, P., Bradley, L., Fowler, S., & Stein, J. (1991). What children see affects how
they read. Developmental Medicine & Child Neurology, 33(9), 755–762.
Chung, S. T., Manseld, J. S., & Legge, G. E. (1998). Psychophysics of reading. XVIII. The
effect of print size on reading speed in normal peripheral vision. Vision Research, 38
(19), 2949–2962.
Chung, S. T., & Bernard, J. B. (2018). Bolder print does not increase reading speed in
people with central vision loss. Vision Research, 153, 98–104. https://doi.org/
10.1016/j.visres.2018.10.012
Chung, S. T., Jarvis, S. H., Woo, S. Y., Hanson, K., & Jose, R. T. (2008). Reading speed
does not benet from increased line spacing in AMD patients. Optometry and Vision
Science: Ofcial Publication of the American Academy of Optometry, 85(9), 827.
https://doi.org/10.1097/OPX.0b013e31818527ea
Chung, S. T. (2002). The effect of letter spacing on reading speed in central and
peripheral vision. Investigative Ophthalmology & Visual Science, 43(4), 1270–1276.
https://doi.org/10.1016/j.visres.2013.03.005
Dogusoy, B., Cicek, F., & Cagiltay, K. (2016, July). How serif and sans serif typefaces
inuence reading on screen: An eye tracking study. In International conference of
design, user experience, and usability (pp. 578–586). Cham: Springer. https://doi.org/
10.1007/978-3-319-40355-7_55.
Drummond, S. R., Drummond, R. S., & Dutton, G. N. (2004). Visual acuity and the ability
of the visually impaired to read medication instructions. British Journal of
Ophthalmology, 88(12), 1541–1542. https://doi.org/10.1136/bjo.2003.029918
Farmer, J., & Morse, S. E. (2007). Project magnify: Increasing reading skills in students
with low vision. Journal of Visual Impairment & Blindness, 101(12), 763–768. https://
doi.org/10.1177/0145482X0710101206
Fellenius, K. (1999). Swedish 9-year-old readers with visual impairments: A
heterogeneous group. Journal of Visual Impairment & Blindness, 93(6), 370–380.
Gompel, M., Van Bon, W. H., & Schreuder, R. (2004). Reading by children with low
vision. Journal of Visual Impairment & Blindness, 98(2), 77–89. https://doi.org/
10.1177/0145482X0409800208
Hy¨
on¨
a, J., & Olson, R. K. (1995). Journal of Experimental Psychology: Learning, Memory,
and Cognition, 21(6), 1430. https://doi.org/10.1037/0278-7393.21.6.1430
Hughes, L., & Wilkins, A. (2000). Typography in children’s reading schemes may be
suboptimal: Evidence from measures of reading rate. Journal of Research in Reading,
23(3), 314–324.
Larson, K., & Carter, M. (2016). Sitka: A collaboration between type design and science.
In M. C. Dyson, & C. Y. Suen (Eds.), Digital fonts and reading (Vol. 1, pp. 37–53).
World Scientic.
Legge, G. E. (2016). Reading digital with low vision. Visible Language, 50(2), 102. PMID:
29242668.
Legge, G. E., Pelli, D. G., Rubin, G. S., & Schleske, M. M. (1985). Psychophysics of
reading-I. Normal vision. Vision Research, 25(2), 239–252. https://doi.org/10.1016/
0042-6989(85)90117-8
Legge, G. E., & Rubin, G. S. (1986). Psychophysics of reading. IV. Wavelength effects in
normal and low vision. JOSA A, 3(1), 40–51.
L´
et´
e, B., Sprenger-Charolles, L., & Col´
e, P. (2004). MANULEX: A grade-level lexical
database from french elementary school readers. Behavior Research Methods,
Instruments, & Computers, 36(1), 156–166.
Liu, L., & Arditi, A. (2001). How crowding affects letter confusion. Optometry and Vision
Science, 78(1), 50–55.
Lockhead, G. R., & Crist, W. B. (1980). Making letters distinctive. Journal of Educational
Psychology, 72(4), 483–493. https://doi.org/10.1037/0022-0663.72.4.483
Lovie-Kitchin, J. E., Bevanm, J. D., & Hein, B. (2001). Reading performance in children
with low vision. Clinical and Experimental Optometry, 84(3), 148–154. https://doi.
org/10.1111/j.1444-0938.2001.tb04958.x
Lovie-Kitchin, J. E., Bowers, A. R., & Woods, R. L. (2000). Oral and silent reading
performance with macular degeneration. Ophthalmic and Physiological Optics, 20(5),
360–370.
Manseld, J. S., Legge, G. E., & Bane, M. C. (1996). Psychophysics of reading. XV: Font
effects in normal and low vision. Investigative Ophthalmology & Visual Science, 37(8),
1492–1501.
McLeish, E. (2007). A study of the effect of letter spacing on the reading speed of young
readers with low vision. British Journal of Visual Impairment, 25(2), 133–143. https://
doi.org/10.1177/0264619607075995
Moret-Tatay, C., & Perea, M. (2011). Do serifs provide an advantage in the recognition of
written words? Journal of Cognitive Psychology, 23(5), 619–624.
Moriarty, S. E., & Scheiner, E. C. (1984). A study of close-set text type. Journal of Applied
Psychology, 69(4), 700–702. https://doi.org/10.1037/0021-9010.69.4.700
Morris, R. A., Aquilante, K., Yager, D., & Bigelow, C. (2002). P-13: Serifs slow RSVP
reading at very small sizes, but don't matter at larger sizes. In , 33(1). SID symposium
digest of technical papers (pp. 244–247). Oxford, UK: Blackwell Publishing Ltd.
https://doi.org/10.1889/1.1830242.
Nersveen, J., Kvitle, A., & Johansen, E. (2018). Legibility in print text for people with
impaired vision. In G. Craddock, et al. (Eds.), Transforming our world through design,
diversity and education (pp. 862–869). IOS Press. https://doi.org/10.3233/978-1-
61499-923-2-862.
Nersveen, J., & Johansen, E. A. (2016). Lesbarhet av trykt tekst for svaksynte. Norsk
forskningslaboratorium for universell utforming. NTNU. https://ntnuopen.ntnu.no/nt
nu-xmlui/handle/11250/2427663.
Pel, J., & Dalhoi, B. (2017). Eye-tracking technology in low vision: implications for
functional diagnosis and rehabilitation. In 12th international conference by the
international society for low vision research and rehabilitation (ISLRR). The
Netheirlands: The Hague.
Rayner, K., & Duffy, S. A. (1986). Lexical complexity and xation times in reading:
Effects of word frequency, verb complexity, and lexical ambiguity. Memory &
Cognition, 14(3), 191–201. https://doi.org/10.3758/BF03197692
Rello, L., & Baeza-Yates, R. (2016). The effect of font type on screen readability by people
with dyslexia. ACM Transactions on Accessible Computing (TACCESS), 8(4), 1–33.
https://doi.org/10.1145/2897736
Rubin, G. S., Feely, M., Perera, S., Ekstrom, K., & Williamson, E. (2006). The effect of font
and line width on reading speed in people with mild to moderate vision loss.
Ophthalmic and Physiological Optics, 26(6), 545–554. https://doi.org/10.1111/
j.1475-1313.2006.00409.x
Rubin, G. S., & Legge, G. E. (1989). Psychophysics of reading. VI—The role of contrast in
low vision. Vision Research, 29(1), 79–91. https://doi.org/10.1016/0042-6989(89)
90175-2
Rubin, G. S. (2013). Measuring reading performance. Vision Research, 90, 43–51.
Russell-Minda, E., Jutai, J. W., Strong, J. G., Campbell, K. A., Gold, D., Pretty, L., &
Wilmot, L. (2007). The legibility of typefaces for readers with low vision: A research
review. Journal of Visual Impairment & Blindness, 101(7), 402–415. https://doi.org/
10.1177/0145482X0710100703
Sander, M. S., Bournot, M. C., Leli`
evre, M., & Tallec, A. (2005). Les personnes ayant un
handicap visuel : Les apports de l’enquˆ
ete handicaps-Incapacit´
es-D´
ependance.
DRESS ´
Etudes et r´
esultats, 416, 1–12.
Sheedy, J. E., Subbaram, M., Zimmerman, A. B., & Hayes, J. R. (2005). Text legibility and
the letter superiority effect. Human Factors, 47(4), 797–815. https://doi.org/
10.1518/001872005775570998
Sonksen, P. M., & Dale, N. (2002). Visual impairment in infancy: Impact on
neurodevelopmental and neurobiological processes. Developmental Medicine and
Child Neurology, 44(11), 782–791.
Sparrow, L., Dei Cas, P., & Kheli, R. (2016). Une police de caract`
eres adapt´
ee permet-
elle Une am´
elioration de la lecture chez les enfants dyslexiques ? ANAE Approche
Neuropsychologique des Apprentissages chez l'Enfant, 28(5), 551–559.
Tarita-Nistor, L., Lam, D., Brent, M. H., Steinbach, M. J., & Gonz´
alez, E. G. (2013).
Courier: A better font for reading with age-related macular degeneration. Canadian
Journal of Ophthalmology, 48(1), 56–62. https://doi.org/10.1016/j.jcjo.2012.09.017
Tatsumi, H., Murai, Y., Kawahara, M., Sekita, I., & Miyakawa, M. (2010). Eye tracking
application for low vision. In 2010 IEEE international conference on systems, man and
cybernetics (pp. 1578–1583). IEEE. https://doi.org/10.1109/ICSMC.2010.5642352.
Wallace, S., Bylinskii, Z., Dobres, J., Kerr, B., Berlow, S., Treitman, R.Sawyer, B. D., …
(2022). Towards individuated Reading experiences: Different fonts increase Reading
speed for different individuals. ACM Transactions on Computer-Human Interaction
(TOCHI), 29(4), 1–56.
Watts, L., & Nisbet, J. (1974). Legibility in children's books: A review of research. London:
NFER Publishing Company.
Wendt, D. (1994). Legibility. In Font technology (pp. 271–306). Berlin, Heidelberg:
Springer.
Whittaker, S. G., & Lovie-Kitchin, J. A. N. (1993). Visual requirements for reading.
Optometry and Vision Science, 70(1), 54–65.
Williams, S., Reiter, E., & Osman, L. (2003, April). Experiments with discourse-level
choices and readability. In Proceedings of the 9th European workshop on natural
A.R. Galiano et al.

Acta Psychologica 236 (2023) 103926
15
language generation (ENLG-2003) (pp. 127–134), 13-14 Apr 2003, Budapest,
Hungary.
Wilkins, A., Cleave, R., Grayson, N., & Wilson, L. (2009). Typography for children may be
inappropriately designed. Journal of Research in Reading, 32(4), 402–412. https://
doi.org/10.1111/j.1467-9817.2009.01402.x
Wilkins, A., Smith, K., & Penacchio, O. (2020). The inuence of typography on
algorithms that predict the speed and comfort of reading. Vision, 4(1), 18. https://
doi.org/10.3390/vision4010018
Woods, R. J., Davis, K., & Scharff, L. F. (2005). Effects of typeface and font size on
legibility for children. American Journal of Psychological Research, 1(1), 86–102.
World Health Organization. (2022). International statistical classication of diseases and
related health problems Available from (11th ed.). https://icd.who.int/en.
World Medical Association. (1964). Declaration of Helsinki. Ethical Principles for Medical
Research Involving Human Subjects. Helsinki, Finland: World Medical Association.
Xiong, Y. Z., Lorsung, E. A., Manseld, J. S., Bigelow, C., & Legge, G. E. (2018). Fonts
designed for macular degeneration: Impact on reading. Investigative Ophthalmology &
Visual Science, 59(10), 4182–4189.
Yager, D., Aquilante, K., & Plass, R. (1998). High and low luminance letters, acuity
reserve, and font effects on reading speed. Vision Research, 38(17), 2527–2531.
https://doi.org/10.1016/S0042-6989(98)00116-3
A.R. Galiano et al.