Non-visual exploration of geographic maps: Does soniﬁcation help?
, MASSIMILIANO PALMIERO
, STEFANO FEDERICI
, HAIXIA ZHAO
& OLIVETTI BELARDINELLI
Department of Psychology, ‘Sapienza’ University of Rome, Rome, Italy,
RIKEN, Brain Science Institute, Wako, Japan,
ECONA, Interuniversity Centre for research On Cognitive processing in Natural and Artiﬁcial systems, Rome, Italy,
di Scienze della Formazione, Universita` degli Studi di Perugia, Perugia, Italy, and
Institute for Advanced Computer Studies,
University of Maryland, Maryland, USA
Accepted June 2009
Purpose. This study aims at evaluating the effectiveness of soniﬁcation as a mean to provide access to geo-referenced
information to users with visual impairments.
Method. Thiry-ﬁve participants (10 congenitally blind, 10 with acquired blindness and 15 blindfolded sighted) completed
four tasks of progressive difﬁculty. During each task, participants ﬁrst explored a soniﬁed map by using either a tablet or a
keyboard to move across regions and listened to sounds giving information about the current location. Then the participants
were asked to identify, among four tactile maps, the one that crossmodally corresponds to the sonifed map they just explored.
Finally, participants answered a self-report questionnaire of understanding and satisfaction.
Results. Participants achieved high accuracy in all of the four tactile map discrimination tasks. No signiﬁcant performance
difference was found neither between subjects that used keyboard or tablet, nor between the three groups of blind and
sighted participants. Differences between groups and interfaces were found in the usage strategies. High levels of satisfaction
and understanding of the tools and tasks emerged from users’ reports.
Keywords: Soniﬁcation, blindness, mental mapping, non-visual exploration, haptics
There is a considerable amount of information that is
not available to blind individuals, in particular the
information which is encoded in spatial and graphi-
cal forms (e.g., diagrams, graphs, geo-political
maps). A subject that puzzles experimental research
is whether the issue is limited to a problem of access
to spatial information by users who are blind or
whether there are also cognitive constraints in the
construction of internal spatial representation for
people who have never had visual experience or who
have lost their vision in early infancy.
Experimental studies provide extensive evidence
that blind people experience difﬁculties in forming
mental representations of spatial structures [1–4] and
that they face difﬁculties in spatial imagery as well
as in navigating efﬁciently through spaces .
These results support the inference that visual
experience is necessary to the acquisition of spatial
concepts. However, there is no general agreement
about the nature of spatial processing difﬁculties
experienced by blind people. In particular, it is
unclear whether these spatial difﬁculties are due to
performance, competence, or ability deﬁcits .
Several studies question this sight-centered model
of spatial processing, arguing that spatial representa-
tions generated by different sensory modalities
(haptic and auditory, for instance) or by verbal input
are functionally equivalent to visual representations
of space [6,7]. This view supports the existence of an
amodal spatial representation system that receives
inputs from different sensory channels . Accord-
ing to this orientation, blind people should be able
to organize functionally equivalent spatial maps
using tactile, auditory, and kinaesthetic information.
Correspondence: Franco Delogu, Laboratory for Perceptual Dynamics, RIKEN BSI, 2–1 Hirosawa, Wako-shi, Saitama 351-0198, Japan.
Disability and Rehabilitation: Assistive Technology, May 2010; 5(3): 164–174
ISSN 1748-3107 print/ISSN 1748-3115 online ª2010 Informa UK Ltd.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 04/24/10
For personal use only.
The theoretical debate that sets a crucial role for
vision against a multisensory, or amodal, spatial
representation is an open question  and has
important empirical implications for designing effec-
tive assistive tools.
In recent decades, the attempts to provide easier
and full access to spatial information for visually-
impaired people have considerably increased. Owing
to more sophisticated integrations of non-visual
perceptual stimuli and conceptual information, new
possibilities of information access have been devel-
oped . One of the ﬁrst studies addressing the
issue that whether other sensory information can
compensate the lack of visual experience in forming
mental spatial imagery, was carried out by Kerr. She
used a mental scanning task in haptic learning. The
results showed a strong relation between scanning
time and Euclidean distances both for sighted and
for congenitally blind people, although the response
times were signiﬁcantly longer for blind than for
sighted subjects . The same chronometric
performance for blind, blindfolded, and sighted
participants was found by Ro¨der and Ro¨ sler using a
similar paradigm . Aleman et al. by asking
participants to memorize the position of a target
cube in two-dimensional (2D) and three-dimen-
sional (3D) matrices after haptic exploration con-
ﬁrmed that blind participants were capable of good
performance, although they made signiﬁcantly more
errors than sighted participants . Although there
is encouraging experimental evidence for effective
spatial mappings in absence of vision, other evi-
dences show that blind subjects have more dif-
ﬁculties than sighted ones in memorizing spatial
positions of target objects , and in main-
taining different items of spatial information simul-
Currently, the most common solution for display-
ing spatial information in the absence of vision is by
tactile mapping. For a long time, tactile maps were
the only means for communicating spatial, particu-
larly geographical information, to visual impaired
people. The design and the production of tactile
maps have involved researchers from several ﬁelds:
applied geographers, computer scientists and psy-
chologists, amongst others. Since the early work [16–
19], cartographers have adopted different solutions
both for designing and producing tactile maps: raised
inks, thermoforming, vacuum forming, etching, and
accretion for tactile maps . In parallel, psycho-
logical research investigated the impact of tactile
mapping on the cognitive system, quantifying acces-
sibility, identifying problems, possibilities and limita-
tions of tactile maps also in comparison to visually
presented maps. Implementations of haptic tools
within virtual navigation environments provided
encouraging results, as blind participants were able
to generate a verbal description and a physical model
of the explored virtual environment, with a general
improvement of navigation performance in real
Despite their effectiveness, tactile maps showed
several limitations and problems both logistically
and cognitively. Firstly, tactile maps are difﬁcult and
expensive to produce, are of low resolution, and
rarely reach a high level of quality . In com-
parison with the great abundance of visual maps, the
variety of raised-line maps is rather scarce. Braille
labeling is also a problem in small maps where there
is no enough room for labeling all the countries,
regions, or features within the map. Furthermore,
tactile spatial exploration is necessarily a slow serial
process compared with rapid visual exploration.
Another limitation concerns the possibility to repre-
sent several features together (like in geopolitical
maps). Multi-feature representation in tactile maps
must be addressed with caution. When maps contain
too much information, the discriminability of single
feature can be impaired and even localization and
shape acquisition can be undermined. Clark and
Clark demonstrated that simpliﬁed maps, rather than
exact representations, lead to a better shape recogni-
In order to overcome these limitations, multi-
disciplinary researches have been developing new
ways of accessing map information, which should be
easier to learn and use by blind people.
The main alternatives to haptic technologies are
audio-based systems. Recent studies have demon-
strated that the auditory channel, with or without
tactile stimulation, can be a useful alternative for the
transmission of spatial information [23,24]. Speciﬁ-
cally, during the last few decades soniﬁcation has
been used to develop new tools for transmitting
spatial information. Soniﬁcation, namely the use of
non-speech audio to convey information, allows the
transformation of data relations into acoustical
ones in order to facilitate communication or inter-
pretation . Amongst others, soniﬁcation has
successfully been used to present geographical,
environmental, or census data . One of the ﬁrst
soniﬁcation systems was the Soundgraphs developed
by Mansur et al. . Soundgraphs allows the
presentation of line graphs in sound. Time is
mapped to the x-axis and pitch to the y-axis. The
shape of the graph can then be heard as a rising or
falling note playing over time. Mansur et al.’s initial
results were very promising as users were able to
identify the types of curves very easily along
with maximum/minimum points. Moreover, the
system allows listeners to get an overview by
listening to all data very quickly, which is not easily
provided using speech. Another soniﬁcation tool,
System, was developed by
Soniﬁcation of geographic maps 165
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 04/24/10
For personal use only.
Krueger and Gilden to present geographic informa-
tion to users with visual impairments . Users’
hands rested upon an illuminated surface covered
with a tactile grid and monitored by a ceiling-
mounted video camera. The video image was
analyzed by specialized processors and the location
of the user’s ﬁngertip on the light table was
determined. An invisible virtual map is deﬁned on
the desk surface and the feature that the user is
currently pointing to is signaled by a sound which
serves to identify the feature or kind of feature that
has been ‘touched’. Ramloll et al. found that using
non-speech sound in 2D numerical tables signiﬁ-
cantly improved the ability of vision impaired users
to locate and acquire data . Afonso et al.
implemented a virtual environment that incorporates
a high quality virtual 3D audio interface . The
aim was to determine how a verbal description and
the active exploration of an environment affected the
building up of a mental spatial representation.
The authors found that active exploration was
better than verbal learning in generating spatial
imagery. Blind participants were able to generate
correct spatial representations of an environment,
although they needed more time than sighted
Evidence from both theoretical and empirical
research seem to converge to the idea that a
combination of sound and touch will work better
than a single modality in non-visual displaying of
spatial information [29,31] Recently, the Human
Computer Interaction Laboratory of the University
of Maryland developed a new soniﬁcation tool,
iSonic, to facilitate the exploration of geo-referenced
information by users with visual impairments 
(Figure 1). In a pilot study, the authors showed that
blind users were able to recognize geographic maps
by using an interactive soniﬁcation system . A
later case study demonstrated that blind participants
could perform complex tasks using combinations of
soniﬁed tables and maps .
Considering multisensory integration as a promis-
ing way to improve the access to spatial information
to blind people, we aimed at verifying whether or not
blind users are able to generate effective mental
representations of geographical information when
using the iSonic soniﬁcation tool.
In the present study, we intended to ﬁnd out, both
in quantitative and qualitative terms, (1) whether
soniﬁcation is a useful mean to provide accurate geo-
political representations (speciﬁcally choropleth
maps), and (2) whether soniﬁed map exploration is
affected by different input modalities of exploration.
At a theoretical level, starting from the claim that
spatial representation is not mandatorily linked to
visual modality, we aimed at verifying that (1) blind
subjects (congenitally or acquired) do not differ from
blindfolded sighted subjects in the identiﬁcation of
soniﬁed maps, and (2) blind people use different
perceptual and cognitive strategies, e.g., focusing
more speciﬁcally on sound information than sighted
Thirty-ﬁve subjects, 16 females and 19 males,
participated in the study (mean age ¼32.46;
SD ¼6.73). The sample is comprised of three groups
(10 ‘early blind’ participants, 10 ‘late blind’ partici-
pants and 15 ‘sighted blindfolded’). Subjects were
considered early blind when visual acuity in their
better eye was below 1/300 within the third year after
birth. We considered those individuals with sight at
birth, and who lost sight after age 3 as late blind.
Motor performance and hearing abilities were
normal in all subjects.
All participants were right-handed. Subjects were
all Italians and therefore likely to be poorly
acquainted with the USA geographical representa-
tions used as stimulus material.
Materials and tasks
Four soniﬁed auditory maps and 16 tactile plastic
maps were used as experimental materials. Auditory
maps represented patterns of unemployment rates in
Figure 1. The iSonic software being used by a user with visual
impairments. As the user moved her ﬁnger on the touch-tablets
stereo a sound is produced representing the value of a selected
variable for the state at the ﬁnger location. Commands can be
issued using the keyboard, such as an automatic sweep of all the
states in a region.
166 F. Delogu et al.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 04/24/10
For personal use only.
USA. The rate for each state was soniﬁed by a
musical pitch. Three levels of pitches were used,
with a high pitch (G4 at 783.9 Hz) for high
unemployment, a medium pitch (E4 at 659.2 Hz)
for medium levels, and a low pitch (C4 at 523.2 Hz)
for low unemployment rates. Patterns of three maps
were of progressive complexity, from the ﬁrst being
clearly divided into three areas, to the third being
more varied and realistic (See leftmost maps in
The fourth auditory map represented the Idaho
state and its counties (Figure 3). For this map, the
three levels of unemployments were not provided as
the map was used to assess how subjects explore and
represent shape, size, and external boundaries of a
map, not the pattern of values.
A total of 16 tactile plastic maps (42 cm630 cm)
were used in the four tasks of the experiment. For
each task, four tactile maps were used, with one
corresponding to the soniﬁed map (target), and the
remaining three being the distractors. The distractor
maps, identical in shape to the target, vary only in
their pattern of unemployment (see the smaller maps
in Figure 2). In order to control whether subjects
were able to detect subtle variations we also included
distractors presenting patterns very similar to the
State and county borders consisted of embossed
lines. For the ﬁrst three tasks, three different textures
represented the three categories of the value (rate of
unemployment) on the tactile maps: dotted texture
was equal to high value, herringbone texture to
medium value, and striped texture to low value (see
the picture in Figure 4 for an example).
For the fourth task, four tactile maps of four
different states were used: with one target map
corresponding to the soniﬁed map (Idaho see
Figure 3), and the three distractors (New Hampshire,
Utah, and Washington). The tactile maps repre-
sented the general shapes of the states. There were
no textures inside the counties of the four tactile
A paper questionnaire was given at the end about
user satisfaction and general understanding of the
tools and tasks.
Figure 2. Maps of unemployment patterns across the USA. The
unemployment value is represented by different shades of grey (the
darker, the higher). In the soniﬁed maps users would hear a high
pitch tone for high values and lower pitch tones for the lower
Figure 3. Auditory map of Idaho shape. The map is blank.
Figure 4. An example of tactile maps with different textures
representing the three values of unemployment rate.
Soniﬁcation of geographic maps 167
For personal use only.
Map exploration was conducted using an opera-
tional interactive software, iSonic, an environment
to assist users with visual impairment in exploring
geo-referenced data using coordinated maps and
tables, augmented with nontextual sounds and
speech output . Only a fraction of the features
of iSonic were used to conduct this study. In
iSonic, different pitches of a violin timbre indicate
different levels of the chosen variable (such as
scholarization, unemployment, or crime rate sta-
tistics). Stereo sounds provide information about
the left–right location of the geographical area.
Table I lists the soniﬁcation features of iSonic that
iSonic can be operated using two input devices: a
computer keyboard and a touch-tablet. The key-
board interface allows a combination of techniques
to navigate the map. Users can perform the
following actions: start an automatic sweep to
obtain a quick overview of the data patterns;
perform a relative navigation using arrow keys to
move from one region to the neighboring ones; use
the numerical keypad to explore the 363 zones into
which the map is divided (see Figure 5), with
pressing 1 to play the left-bottom zone, pressing 9
to play the right-top zone, and so on; request details
of a data item (e.g., the name of the current state).
The modality of keyboard exploration was struc-
tured as shown in Table II.
The touch-tablet interface allows users to point to
and explore the display in a more direct way: when
users touch the smooth surface they hear the sound
whose pitch indicates the value for the region at the
ﬁnger position. They can also drag their ﬁnger on the
surface and hear as a continuous sound plays when
their ﬁnger moves within the region. The sound
stops when the user lifts her/his ﬁnger. When the
ﬁnger crosses a region border, a border sound is
played. Another sound is played when the user
moves outside the map. In both interfaces, by
pressing ‘0’ on the number pad (as shown in the
Table II) participants could hear a sweep of the
The two interfaces show important differences:
keyboard exploration could be deﬁned as qualita-
tive, discrete, and symbolic whereas touch-tablet
exploration as quantitative, continuous, and analo-
gical. More precisely, keyboard exploration is:
qualitative, because it informs about the state
values; discrete, because it progresses by means of
all-or-nothing steps; symbolic, because arrow-keys
typically move in only four directions. In contrast,
the touch-tablet interaction is: quantitative, because
each movement gives information about shape and
size of each state; continuous, because it keeps
informing users also when they stay still on a spot;
analogical-proprioceptive, because the movement of
a part of the body mirrors an analogous movement
on the map.
Figure 5. Participants were asked to indicate which level of
unemployment (low, medium, high) was the most represented
within each one of the nine above depicted map zones.
Table I. Sound and their corresponding parameters.
Sound parameter Map feature
High state with a high level
Medium state with a medium level
Low state with a low level
Single click region border
Fret guitar background (area out of
the map borders)
(relative loudness of left
and right channels)
(position on the left-right axis)
Table II. Keyboard exploration.
168 F. Delogu et al.
For personal use only.
Each experimental session was conducted in a
soundproof and dimly lighted room and lasted for
*1 h and 30 min. The whole session consisted of
three phases: a pre-experimental phase, the experi-
ment, and a post-experimental questionnaire. Sub-
jects were randomly assigned to either the keyboard
or the touch-tablet interface condition.
In the pre-experimental phase, participants were
tested for their basic auditory discrimination abilities
with an auditory localization task and a pitch
discrimination task. Then, in a training session of
about 10 min, they could explore soniﬁed training
maps (different from the ones used in the experi-
mental tasks) to practice with the interface and the
logic of the experimental tasks.
In the experimental phase, subjects performed four
tasks. Each task consisted of auditory exploration,
tactile recognition, and questionnaire ﬁlling. The
interface condition (keyboard or touch-screen) for
the exploration depended on the experimental group
each subject was assigned to. Subjects were in-
structed to explore the soniﬁed map in order to learn
the overall data pattern. The ﬁrst three tasks were to
explore the patterns of unemployment rate across
states. The fourth task was to explore the general
shape of the map. The purpose of the ﬁrst three tasks
was to investigate the exploration and representation
of geo-referenced data patterns on maps, whereas the
fourth task was designed for assessing how subjects
explore and represent shape, size, and external
boundaries of the map. For each task, immediately
after exploring the auditory map, subjects were
presented with four tactile maps, one target and three
distractors. Subjects were invited to rate each of the
four tactile maps for its correspondence with the
soniﬁed map, along a scale ranging from 0 to 10 (0
standing for ‘no correspondence at all’ and 10
standing for ‘perfect match’). Finally, participants
answered questions about the distribution of values
across the states (for task1, 2, and 3), or about the
approximate shape of the map (rectangular,
L-shaped, triangular, squared, circular, or rectangu-
lar) and the number of regions (for task 4). There was
a 5-min break between the tasks.
In the post-experimental phase, participants answered
a questionnaire about their satisfaction and under-
standing of the tool and tasks, based on a 6-point
Likert scale (from 0 completely in disagreement to 6
completely in agreement). The questionnaire covered
the following aspects: pitch sound detection, useful-
ness of stereo-panning, map shape discrimination,
discrimination of the shape of the regions inside the
Idaho map and their numerosity, understanding of
unemployment level distribution, software usability,
software easiness, pleasantness of the software.
During each task, we asked participants to rate (from
0 to 10) the level of matching of each one of the four
tactile maps with the soniﬁed map they explored. Only
one of the tactile maps (target) matched the soniﬁed
map, whereas the other three were distractors.
An analysis of variance was performed for each one
of the four tasks, using the factor map(target and
three distractors) as the independent within-subjects
factor. High rates of matching between the target
tactile map and the soniﬁed map (mean ¼7.30,
SD ¼0.14) and low rates of matching between
distractors and the soniﬁed map (mean ¼3.75,
SD ¼0.49) indicated high accuracy.
The effect was signiﬁcant in all four tasks with the
following values: F(3, 102) ¼19.33, p50.01 in task
1, F(3, 102) ¼14.08, p50.01 in task 2, F(3, 102) ¼
12.40, p50.01 in task 3 and F(3, 102) ¼25.03,
p50.01 in task 4. The post hoc analyses (Fisher’s
LSD) showed that in all tasks the target tactile map
was rated higher than all distractors. Considering
that the differences between the patterns of target
and distractors are in some cases very subtle (see for
example the second distractor map of task 1 and all
the distractors of task 3 in Figure 2), subjects
demonstrated to have a clear representation of the
unemployment pattern of target maps.
A separate analysis was conducted considering
Interface (keyboard or touch-screen) and the Group
(sighted vs. late blind vs. early blind subjects) as
between factors and Task as within factor. The
algebraic difference between the target tactile map
matching rates and the distractors map matching rates
was used as a measure for subject’s accuracy. This
measure was the dependent variable of the analysis.
The results showed that congenitally blind,
acquired blind, and sighted blindfolded subjects
did not differ in discriminating targets in all four
tasks: F(2, 29) ¼2.25, p¼0.12.
Regarding the inﬂuence of the interface, the use of
keyboard or touch-tablet during the exploration did
not lead to signiﬁcant differences in discriminating
targets from distractors: F(1, 29) ¼0.075, p¼0.78.
The inﬂuence of TASK is not signiﬁcant,
F(3, 87) ¼1.39, p¼0.25, indicating that the discri-
mination of target maps from distractors was of
comparable difﬁculty in the four different tasks.
After each one of the ﬁrst three tasks, subjects were
asked to report the level of value (low, medium, high
unemployment, or water/background) that was the
Soniﬁcation of geographic maps 169
For personal use only.
most represented in each one of the following nine
zones of the map:
For task, four subjects were asked to estimate the
number of regions included in the map. Subjects were
particularly accurate in the ﬁrst task (about 80% of
correct identiﬁcation), in which the levels of unem-
ployment were clearly organized in three big patterns,
while in tasks 2 and 3, where the spatial organization
was more complex, they performed progressively
more poorly (62% and 53%, respectively). A one-
way within ANOVA conﬁrmed that this difference is
statistically signiﬁcant F(2, 16) ¼8.34, p50.01. Task
1 differed from task 2 and task 3 which in turn did not
differ from each other (post hoc Fisher’s Least
Signiﬁcant Difference (LSD): p50.05).
We did not ﬁnd any difference, F(2, 24) ¼1.7,
p40.05, in the percentage of correct identiﬁcation
of values into quadrants among sighted, congenitally
blind, and acquired blind subjects.
Interestingly, in task 4, when subjects had to estimate
the number of regions inside the map, there was, in all
conditions, a general tendency to underestimate the
number of regions (mean ¼25.02; SD ¼14.02), re-
porting 544, the actual number of Idaho counties.
After all tasks were completed, participants answered
a set of questions (Likert-scale from 0 to 6) regarding
the general features of the task and the soniﬁcation
tool. A two-factor between-subject ANOVA was
performed, considering Interface and Group as
independent variables. Table III shows the complete
list of the questions and the level of signiﬁcance both
for group and interface factors.
Subjects reported high levels of understanding and
satisfaction of the tool and the tasks, even though less
in the more speciﬁc aspects and features of the maps
(particularly for the number and shape of regions and
general shape of the maps). We found no statistically
signiﬁcant difference between sighted and blind
participants and between the two interface groups
except for a few exceptions. Regarding the stereo
panning feature, (which set the proportions of the left
and right signal amplitude as a function on the
position of the source), blind subjects paid more
attention to stereo panning than sighted participants.
Both groups of blind subjects reported being helped
by the stereo panning more than sighted subjects
(post hoc LSD: p50.05).
The groups differed also in their perception of the
easiness of the interface: blind subjects (both early
and late blind) rated the interface higher for ease of
use than the sighted participants, probably due to
their experience with non-visual interfaces.
Finally, no signiﬁcant difference was found in
blind and sighted subjects’ answers about the
effectiveness of the interface.
Exploration modalities and strategies
From the log ﬁles of tasks 1 and 3, the easiest and the
hardest tasks respectively, we extracted several
quantitative parameters to analyze modalities and
strategies of exploration.
Regarding the number of steps, (where step is
deﬁned as a move from a state to another), the data
showed that touch-tablet users performed more steps
than keyboard users (295 vs. 118, respectively in
average). This difference is signiﬁcant, F(1, 43) ¼
17.66, p50.001. The higher number of steps
performed by touch-tablet users is reﬂected in higher
exhaustivity, measured with the percent number of
states explored. In fact, we found that touch-tablet
users explored more states (83.20%) than keyboard
users (62.98%), and this difference is signiﬁcant,
F(1, 39) ¼27.93, p50.001. There is no signiﬁcant
inﬂuence of group either on the number of steps, or
on the exhaustivity of exploration. However, group-
ing together late blind and sighted participants, they
showed greater exhaustivity than those early blind,
F(1, 41) ¼4.49, p50.05.
Table III. Questions and signiﬁcance level of the ﬁnal questionnaire.
Question Rate (from 1 to 6) Group Interface
It is easy to distinguish the different sounds 5.1 F(2,29) ¼0.26, p¼0.76 F(1,29) ¼0.00, p¼0.99
Stereo panning is useful for the orientation 4.6 F(2,29) ¼6.22, p¼0.00** F(1,29) ¼1.10, p¼0.30
The shape of the maps is clear to me 3.7 F(2,29) ¼0.18, p¼0.83 F(1,29) ¼0.63 p¼0.43
The shape of the single regions was clear to me 3.7 F(2,29) ¼0.18, p¼0.83 F(1,29) ¼0.63 p¼0.43
The number of the region into the maps was clear to me 2 F(2,29) ¼0.02, p¼0.97 F(1,29) ¼0.47 p¼0.49
I could understand the distribution
of the level of unemployment during the exploration
4.2 F(2,29) ¼1,96, p¼0.15 F(1,29) ¼0.04 p¼0.83
The tool is easy to use 5.4 F(2,29) ¼4.05, p¼0.02* F(1,29) ¼1.54 p¼0.22
The tool is fun to use 4.4 F(2,29) ¼0.98, p¼0.38 F(1,29) ¼8.8 p¼0.00**
The tool is easy to learn 5.1 F(2,29) ¼1.53, p¼0.23 F(1,29) ¼1.61 p¼0.21
*The levels of signiﬁcance at 0.05.
**The levels of signiﬁcance at 0.01.
170 F. Delogu et al.
For personal use only.
We analyzed whether participants tended to
exhibit linear movements by keeping a clear direction
for more than two steps and whether they maintained
speciﬁc directions while moving. The percentage of
‘keep-a-direction steps’ on the total number of
movements is high (54.69%). Through a uni-
factorial ANOVA, considering keep (percentage of
steps that does not change direction for at least two
movements) as the dependent variable and interface
as the between factor, we observed that keyboard
users tended to keep a direction more than touch-
tablet explorers, F(1, 39) ¼17.64, p50.001. There
was not a signiﬁcant inﬂuence of group on keep,
F(2, 39) ¼2.01, p40.05.
There was signiﬁcant preference for speciﬁc
directions, F(3, 117) ¼8.43, p50.001. The post
hoc analysis showed that rightward movements
(39.31%) were the most performed, followed by
leftward movements (30.60%), whereas downward
(18.08%) and upward (11.99%) movements were
performed less frequently. There was no inﬂuence of
the group on the direction preference, F(6, 117) ¼
1.96, p40.05, but there was inﬂuence of the
interface, F(3, 117) ¼8.43, p50.001. As shown in
Figure 6, while touch-tablet users showed clear
preference for rightward movements (which in the
post hoc is signiﬁcantly higher than all the other
directions), keyboard users showed that they prefer
horizontal movements (both rightward and leftward)
to vertical ones (both upward and downward).
In all four tasks, participants showed high accuracy
in recognizing the tactile map corresponding to the
soniﬁed map explored. This result indicates that, at
least to some extent, auditory stimulation can be
effectively used to present geo-referenced informa-
tion. Considering the abovementioned limitations of
traditional tactile maps, our ﬁndings have relevant
empirical implications for the accessibility of blind
individuals to geopolitical data and, more in general,
to spatial information.
Considering the differences between blind and
sighted subjects, according to the hypothesis that
claims that vision plays a crucial role in setting up
spatial mapping during a critical or sensitive period
of development , we would expect early blind
subjects to perform poorer than late blind and
blindfolded ones. However, our results go against
this theoretical expectation, suggesting the possibility
of equivalent performance of blind and sighted
subjects in spatial representation and orientation
obtained through modalities other than vision. In
fact, the three groups do not differ in performing the
map recognition task. Our results are consistent with
the hypothesis of an equivalence between sensory
modalities in transmitting spatial information [6,7]
and with the amodal theory of spatial representation
Nevertheless, we should consider that subjects,
both sighted and blind, had difﬁculties in recognizing
speciﬁc details of the map. In particular, all
participants had difﬁculty remembering the distribu-
tion of the values in complex patterns (task 2 and
task 3) and the number of regions inside a map (task
4). In accordance with experimental evidence, it
seems that soniﬁcation, like other non-visual display
tools , at least in its present modalities, is more
indicated for navigating and learning the macro-
structure of a spatial conﬁguration than for transmit-
ting spatial details. The latter would require a higher
engagement of working memory. Moreover in the
speciﬁc case of our experiment, the mandatorily
sequential information acquisition may have pro-
gressively shadowed previously acquired details.
Future research has to take into more direct account
the relationships between display characteristics and
cognitive processes involved in task exploration.
Regarding the relative effectiveness of the two
interfaces, we found that touch-tablet and keyboard
users showed no signiﬁcant difference in the
performance. Both keyboard (exploring the map by
moving step-by-step with the arrow keys) and
touch-tablet (exploring the map by touching the
screen with a ﬁnger) exploration seem to lead to an
effective overall spatial representation. Consistent
with the amodal theory of spatial representations
[8,36], this result supports that different perceptual
experiences may result in comparable representa-
tions of the same spatial conﬁguration. However, the
presence of representational differences, not detect-
able at a performance level, cannot be excluded.
Some differences in strategies and modalities
emerged from the log ﬁle analysis. Touch-tablet
users performed more steps than keyboard ones,
Figure 6. Percentage of movements towards the four directions in
function of the interface.
Soniﬁcation of geographic maps 171
For personal use only.
moving faster along the map. This data could be due
to the positional spatial feedback in proprioceptive
exploration (by touch-tablet) which is lacking in
discrete exploration (by keyboard). It seems that a
more direct, analogical, on-line spatial processing is
guaranteed by touch-screen and that keyboard
exploration, being strictly symbolic and discrete,
requires a more conspicuous working memory
involvement for reconstructing the position after
each step. Exploring by arrow keys requires spatial
reasoning, thus it is slower. It is possible that the
greater cognitive effort with the keyboard interface
may have an effect on the duration of learning. This
aspect needs further investigation. Keyboard users
also tended to be less exhaustive and may have
missed some parts of the map when moving by
discrete steps through states with irregular shapes.
The fact that arrow exploration is discrete, symbolic,
and less sensitive to irregularities leads keyboard
users to keep a speciﬁc direction more than touch-
tablet users. On the contrary, touch-tablet explora-
tion, which provides information about the shape of
states, can promote a more dynamic exploration
behavior, and consequently, more frequent changes
Data show that there was a general preference for
rightward movements. In particular, touch-tablet
users preferred to move to the right more than the
other directions and this may be tied to hand
lateralization and reading direction that is a well
acquired way of processing space and acting in it by
both blind and sighted individuals. Keyboard users,
instead, showed a general preference to horizontal
movements (both rightward and leftward). It must be
noted that the two keys (rightward and leftward) are
on the same row and may allow rapid backward/
forward control along the writing direction while the
hand maintains the same posture.
To summarize the analysis of exploration mod-
alities and strategies, the touch-tablet interface seems
to foster better exploration compared with the
keyboard interface. With proprioceptive cues, the
touch-screen interface allows subjects to have con-
stant awareness about the current exploration posi-
tion. On the contrary, with the keyboard interface,
subjects were forced to construct a mental mapping
by maintaining in memory and interpolating a large
amount of data that are only partially present and
mostly extracted by previous positions. Moreover,
the lack of information could cause shape and
location distortions in subjects’ mental maps. In
fact, as subjects had no information about state size,
they could tend to normalize the dimensions of each
state, and as a consequence, to misplace the centre of
the map. This problem is particularly relevant in
cases like the USA map, considering the concentra-
tion of small states in the North-East and the
presence of the biggest states in the South-West,
distribution unfamiliar to Italian subjects. Another
limitation of keyboard exploration is that the absolute
position is only obtainable by the rehearsal of the
previous steps while with the touch-tablet it is
immediately inferable following the proprioceptive
feedback. However, a possible advantage of keyboard
exploration is that it allows a simpler categorization
of the map division into states. For example, it is
easier to categorize the number of states along the
left–right and bottom–up axes. Obviously this is an
advantage only for regular maps. On the contrary it
seems that touch-tablet is more adapted for exploring
more complex maps and for detecting ﬁner scale
Regarding the inﬂuence of vision in exploration
modalities and strategies, we did not ﬁnd differences
between early blind, late blind, and blindfolded
subjects: the equivalent performances by our three
groups are further evidences for the amodal theory of
spatial representations [8,36].
When rating the efﬁciency and the manageability
of the soniﬁcation tool and the tasks, our subjects
reported high levels of understanding and satisfac-
tion, even when they had difﬁculties in comprehend-
ing some speciﬁc features of the maps. We noticed
that blind subjects reported to be helped by the
stereo-panning more than sighted subjects. This
result is comprehensible, as, thanks to their experi-
ence, blind people pay more attention to non-visual
environmental information and perform better than
sighted people in sound localization [38,39]. Groups
differed also in their perception of the easiness of the
interface. Blind subjects (both early and late ones)
rated the tool easier to use than sighted ones. This
could again be due to their previous greater
experience and attention with non-visual events
To conclude, the soniﬁcation tool iSonic is useful to
transmit geo-referenced information in the absence
of visual information conﬁrming the hypothesis that
visual experience is not necessary for an efﬁcient
spatial cognition . Consistent with Millar ,
our results show that different senses are able to
transmit spatial information, as spatial representa-
tions are not necessarily linked to any speciﬁc
sensory modality. However, all participants, both
blind and sighted, found difﬁculties in acquiring ﬁne
details of soniﬁed maps. Such difﬁculties seem to be
linked to the greater amount of time necessary to
acquire data by touching and hearing, compared with
by seeing and to a heavy working memory load
172 F. Delogu et al.
For personal use only.
Touch-screen exploration seems to be preferable
to keyboard exploration for several reasons: among
others, it allows direct integration between haptic-
spatial cues and soniﬁcation, promotes more ex-
haustive explorations, causes less representational
distortions than keyboard, and ﬁnally is considered
by users as easier and funnier than keyboard.
The results, limited by the speciﬁcity of the empirical
applied research and by the use of a restricted
ecological cartographic pattern, should be general-
ized by further studies. The design of more
articulated experimental paradigms implying a con-
trolled manipulation of number, size, and shape of
states inside the map could improve the external
validity. Moreover, ﬁner grained dependent mea-
sures allowing a ﬁner description of how the spatial
patterns and mentally represented are needed.
Despite of the limited generalizability of our data,
this research conﬁrms the importance of multi-
modal integration in the transmission of spatial
information. In fact, jointly with evidence from both
theoretical and empirical research, our research
supports the idea that a combination of sound and
touch will work better than a single modality for non-
visual display of spatial information. In particular,
the dynamic integration of auditory and haptic
capabilities could represent a new frontier for the
study of new effective non-visual displays allowing
users to access spatial information in more ﬂexible
Declaration of interest: The authors report no
conﬂicts of interest. The authors alone are respon-
sible for the content and writing of the article.
1. Millar S. Understanding and representing space: theory and
evidence from studies with blind and sighted children.
Oxford: Oxford University Press; 1994.
2. Millar S. Understanding and representing spatial information.
Br J Vis Impair 1995;13:8–11.
3. Millar S. Theory, experiment and practical application in
research on visual impairment. Eur J Psychol Edu 1997;
4. Ungar S, Blades M, Spencer C. The construction of cognitive
maps by children with visual impairments. In: Portugali J,
editor. The construction of cognitive maps. Dordrecht:
Kluwer Academic Publishers; 1996. pp 247–273.
5. Lahav O, Mioduser D. A blind person’s cognitive mapping of
new spaces using a haptic virtual environment. J Res Spec Edu
6. Avraamides M, Loomis J, Klatzky RL, Golledge RG. Func-
tional equivalence of spatial representations derived from
vision and language: evidence from allocentric judgments.
J Exp Psychol Learn Mem Cogn 2004;30:801–814.
7. de Vega M, Cocude M, Denis M, Rodrigo MJ, Zimmer H.
The interface between language and visuo-spatial representa-
tions. In: Denis M, Logie RH, Cornoldi C, de Vega M,
Engelkamp J, editors. Imagery, language and visuo-spatial
thinking. London: Psychology Press; 2001. pp 109–136.
8. Bryant DJ. A Spatial Representation System in Humans.
Psycholoquy. [Serial Online] 1992;3(16):Space (1). Electronic
3.16 VIA The INTERNET. Last accessed on March 11, 2008.
9. Thinus-Blanc C, Gaunet F. Representation of space in blind
persons: vision as a spatial sense? Psychol Bull 1997;
10. Golledge R, Klatzky R, Loomis J. Cognitive mapping and
wayﬁnding by adults without vision. In: Portugali J, editors.
The construction of cognitive maps. The Netherlands:
Kluwer; 1996. pp 215–246.
11. Kerr NH. The role of vision in ‘visual imagery’ experiments:
evidence from the congenitally blind. J Exp Psychol Gen
12. Ro¨ der B, Ro¨ sler F. Visual input does not facilitate the scanning
of spatial images. J Ment Imagery 1998;22:165–181.
13. Aleman A, van Lee L, Mantione M, Verkoijnen I, de Haan E.
Visual imagery without visual experiences: evidence from
congenitally totally blind people. Neuroreport 2001;12:
14. Vecchi T. Visuo-spatial imagery in congenitally totally blind
people. Memory 1998;6:91–102.
15. Vecchi T, Tinti C, Cornoldi C. Spatial memory and
integration processes in congenital blindness. Neuroreport
16. Sherman JA. Current map resources and existing map needs
for the blind. Twenty-third Annual Meeting of the American
Congress on Surveying and Mapping. Washington DC.
Gaithersburg, MD: American Congress on Surveying and
Mapping; 1963. pp 26–29.
17. Sherman JA. Needs and resources in maps for the blind. New
Outlook Blind 1965;59:130–134.
18. Wiedel JW, Groves PA. Tactual Mapping: design, reproduc-
tion, reading and interpretation. Washington, DC: Depart-
ment of Health, Education and Welfare; 1969. p 116.
(Reprinted as Occasional Paper in Geography No. 2,
Department of Geography, University of Maryland, 1972.).
19. Andrews SK. Spatial cognition through tactile maps. First
International Symposium on Maps and Graphics for the
Visually Handicapped. March 10–12, 1983. Washington, DC:
Association American Cartographer; 1983. pp 30–40.
20. Salisbury J, Sirinivasan M. Virtual environment Technology
for training (VETT). BBN Report No. 7661. Cambridge,
MA: VETREC, MIT; 1992.
21. Jacobson RD. Navigating maps with little or no sight: a novel
audio-tactile approach. Workshop on Content Visualization
and Intermedia Representations held at the University of
Montreal. 1998, August 15; Montreal, Quebec. New Bruns-
wick: Association for Computational Linguistics (ACL)
publishers; 1998. pp 95–102.
22. Clark J, Clark DD. Creating tactile maps for the blind using a
GIS. ASPRS/ACSM AnnualConvention & Exposition, Bethes-
da, MD: J. Lyon, April 25 1994 Reno, Nevada. pp 283–288.
23. Kawai Y, Kobayashi M, Minagawa H, Miyakawa M,
Tomita F. A support system for visually impaired persons
using three-dimensional virtual sound. Seventh International
Conference on Computer Helping People with Special Needs
(ICCHP). July 17–21, 2000, Karlsruhe. Wien: Osterrei-
chische Computer Gesellshaft Publishers; 2000. pp 327–334.
24. Parente P, Bishop G. BATS: the blind audio tactile mapping
system. Forty-ﬁrst ACM Annual Southeast Regional Con-
ference held at Amstrong Atlantic State University, March
7–8, 2003. Savannah, Georgia.
Soniﬁcation of geographic maps 173
For personal use only.
25. Kramer G, Walker B, Bonebright TP, Cook J, Flowers N,
Miner J. Soniﬁcation report: status of the ﬁeld and research
agenda. International Community for Auditory Display
(online). Internet. 1997. Electronic Citation. http://www.
icad.org/websiteV2.0/References/nsf.html. Last accessed on
January 23, 2008.
26. Kramer G, editor. Auditory display.Reading, MA: Addison-
27. Mansur DL, Blattner M, Joy K. Sound-Graphs: a numerical
data analysis method or the blind. J Med Syst 1985;
28. Krueger MW, Gilden D. KnowWhere
: an audio/spatial
interface for blind people. Fourth International Conference
for Auditory Display (ICAD) held at Xerox Palo Alto
Research Center (Parc). November 2–5, 1997, Palo Alto, CA.
29. Ramloll R, Yu W, Riedel B, Brewster SA. Using non-speech
sounds to improve access to 2D tabular numerical informa-
tion for visually impaired users. Fifteenth Annual conference
of British Computer Society (BCS) IHM-HCI, September
10–14, 2001, Lille. Berlin: Springer publishers; 2001.
30. Afonso A, Katz BFG, Blum A, Jacquemin C, Denis M. A
study of spatial cognition in an immersive virtual audio
environment: comparing blind and blindfolded individuals.
Eleventh Meeting of the International Conference on Audi-
tory Display (ICAD) held at the University of Limerick,
Ireland: Limerick, July 6–9 2005, Ireland. pp 228–235.
31. Wickens CD. Processing resources in attention. In:
Parasuraman R. Davies DR. editors. Varieties of attention.
New York: Academic Press Publishers; 1984. pp 63–101.
32. Zhao H, Shneiderman B, Plaisant C. Listening to choropleth
maps: interactive soniﬁcation of geo-referenced data for users
with vision impairment. In: Lazar J, editor. Universal usability.
New York: Hoboken John Wiley & Sons Ltd. Publishers;
2008. pp 141–174.
33. Zhao H, Smith BK, Norman K, Plaisant C, Shneiderman B.
Interactive soniﬁcation of choropleth maps: design and
evaluation. IEEE Multimed 2005;12:26–35.
34. Zhao H, Plaisant C, Shneiderman B, Lazar J. Data soniﬁca-
tion for users with visual impairment: a case study with geo-
referenced data. ACM T Hum Comput Interact 2008;15:
35. Fortin M, Voss P, Rainville C, Lassonde M, Lepore F. Impact
of vision on the developmental of topographical orientation
abilities. Neuroreport 2006;17:443–446.
36. Grush R. The emulation theory of representation: motor
control, imagery and perception. Behav Brain Sci 2004;
37. Gaunet F, Martinez JL, Thinus-Blanc C. Early-blind subjects’
spatial representation of manipulatory space: exploratory
strategies and reaction to change. Perception 1997;
38. Doucet ME, Gagne´ JP, Leclerc C, Lassonde M, Guillemot JP,
Lepore F. Blind subjects process auditory spectral cues more
efﬁciently than sighted individuals. Exp Brain Res 2005;
39. Yabe T, Kaga K. Sound lateralization test in adolescent blind
individuals. Neuroreport 2005;16:939–942.
40. Tinti C, Adenzato M, Tamietto M, Cornoldi C. Visual
experience is not necessary for efﬁcient survey spatial
cognition: evidence from blindness. Q J Exp Psychol 2006;
41. Foulke E. Perception, cognition and the mobility of blind
pedestrians. In: Potegal M, editor. Spatial abilities: develop-
ment and physiological foundations. San Diego, CA: Aca-
demic Press; 1982. pp 55–76.
42. Millar S. Crossmodal and intersensory perception and the
blind. In: Walk RD, Pick HC, editors. Intersensory perception
and sensory integration. New York: Plenum; 1981.
174 F. Delogu et al.
For personal use only.