Content uploaded by Samantha Jamson
Author content
All content in this area was uploaded by Samantha Jamson on Apr 02, 2014
Content may be subject to copyright.
White Rose Research Online
http://eprints.whiterose.ac.uk/
Institute of Transport Studies
University of Leeds
This is the author produced version of an article which appeared in Ergonomics
journal. It has been peer reviewed, but does not contain the publisher’s
formatting or pagination.
White Rose Repository URL for this paper:
http://eprints.whiterose.ac.uk/2511/
Published paper
Jamson, S.L.; Tate F.N.; Jamson, A.H. (2005) Evaluating the effects of bilingual
traffic signs on driver performance and safety - Ergonomics 48(15) 1734 – 1748
White Rose Consortium ePrints Repository
eprints@whiterose.ac.uk
Evaluating the effects of bilingual traffic signs on driver
performance and safety
S. L. JAMSON†*, F.N. TATE‡ and A.H. JAMSON†
†Institute for Transport Studies, University of Leeds, LS2 9JT, UK
‡ MWH New Zealand, 123 Taranaki St, PO Box 9624, Wellington, New Zealand
*Corresponding author. Email: S.L.Jamson@its.leeds.ac.uk
Variable Message Signs (VMS) can provide immediate and relevant information to road users and bilingual
VMS can provide great flexibility in countries where a significant proportion of the population speak an
alternative language to the majority. The study reported here evaluates the effect of various bilingual VMS
configurations on driver behaviour and safety. The aim of the study was to determine whether or not the
visual distraction associated with bilingual VMS signs of different configurations (length, complexity)
impacted on driving performance. A driving simulator was used to allow full control over the scenarios, road
environment and sign configuration and both longitudinal and lateral driver performance was assessed.
Drivers were able to read one and two-line monolingual signs and two-line bilingual signs without disruption
to their driving behaviour. However, drivers significantly reduced their speed in order to read four-line
monolingual and four-line bilingual signs, accompanied by an increase in headway to the vehicle in front.
This implies that drivers are possibly reading the irrelevant text on the bilingual sign and various methods for
reducing this effect are discussed.
Keywords: Variable Message Signs; Bilingual; Driver performance; Workload;
AMS Subject Classification:
1. Introduction
There are a number of bilingual countries (e.g. Wales, Belgium, Canada, New Zealand) where two languages
are given equal status, and provision needs to be made for speakers of both languages. This provision applies
to general signage including road signs and this has led to the need for research that evaluates drivers' ability
to perceive and comprehend them. In effect, when drivers encounter bilingual signs, they perceive double the
amount of text, half of which is superfluous to them. This paper presents a study that examines whether
bilingual signing affects driver performance and safety, using the local UK example of Welsh bilingual signs.
The Welsh language is the oldest living language of Great Britain and one of the oldest in Europe. The
population of Wales amounts to 5% of the UK total and it has a National Assembly responsible for
developing and implementing policies which reflect the needs of Welsh people. The study focussed on using
Variable Message Signs, due to their imminent introduction on Welsh roads.
1.1. The design of road signs
Road signs provide drivers with appropriate warnings and information as well as identifying legal
requirements or instructions. To fulfil this function efficiently and therefore safely, the signs must:
1) Be easily recognisable and locatable within a complex visual scene.
2) Clearly indicate the status of the message (legal, warning or information).
3) Convey the message efficiently thereby minimising visual distraction.
4) Be comprehensible so that drivers can recognise the action (or choice) to be taken.
5) Be located such that the driver has sufficient time to act on the message.
The perception and comprehension of a road sign (and indeed text in general) can be broken down into
three stages (Anderson 1990). The first stage comprises the perceptual processes by which the text on the road
sign is encoded. The second stage is termed parsing whereby the words in the message are transformed into a
mental representation of the combined meaning of the words. The third stage is the utilisation stage, in which
drivers actually use the mental representation of the sentence’s meaning. If the message is an assertion, the
drivers may simply store the meaning in memory; if it is an instruction, they may obey.
Many studies of visual information processing have involved determining what can be extracted from a
brief visual presentation and the resulting memory for this information (e.g. Sperling 1960). Displays of
letters are presented briefly to participants who are then asked to recall as many as possible. Usually they are
able to recall between three and six items, although they report they saw more, but could not identify them,
i.e. they faded away. These basic perceptual experiments demonstrate that visual information presented to
drivers should be limited in length and be in a relatively uncluttered environment. If information is presented
too quickly in succession, drivers will be unable to process the messages effectively.
Once the road sign has been perceived, it then requires recognition. It is generally agreed that recognising a
message involves feature analysis (e.g. Kinney et al. 1966). Here, the overall pattern is decomposed into a set
of 'mini' features, which are recognised and then used to identify the pattern. Thus the letter 'H' consists of
two vertical and one horizontal line and a specification of how they should be combined. In terms of
recognition of the individual letters on a VMS sign, this would imply that as long as the individual features
are readily recognisable, drivers should be able to interpret the message. Where problems might occur is
where there are limitations in the sign design that do not allow the display of the features naturally used in text
processing.
A further important consideration in sign design is that of context. When context or general world-
knowledge guides perception, processing is known as top-down processing, because high-level general
knowledge determines the interpretation of the low-level perceptual units (Tulving et al. 1964). With context,
less information needs to be extracted from the word itself in order to identify it. These results strongly
suggest that signing should be as context-relevant as possible and use familiar phrases or symbols.
As drivers, we are faced with enormous amounts of information to process in such a way that the
environment makes sense. The more frequently processes have been practised, the less attention they require.
Highly practised processes, which require little or no attention are referred to as automatic, while those
processes that require attention are known as
controlled (Schneider et al. 1977). With reference to the driving
task, and in particular the reading of road signs, a designer should aim to implement a message or pictogram
that is already known to the driving population or is already used in another context. This limits the time
needed to perceive and parse a text message.
Some attempts at producing design guidelines for signs have been made, based on the time taken for
drivers to read traffic signs (Forbes et al. 1965, Johansson et al. 1970). Research indicates that reading times
typically increase in a linear manner with the number of names on a sign (Hall et al. 1991). This has
implications for the design of bilingual signs, as they inherently display more information than their
monolingual counterparts.
1.2. The use of Variable Message Signs
VMS have gained more importance over recent years due to the improved flexibility of signs brought about
by the use of electronics and the widespread availability of communications. They have been used to supply
drivers with up-to-date information about road and weather conditions and can be interfaced to traffic
monitoring systems. The majority of studies on VMS focus on rates of compliance, either by using surveys or
via observational studies (see Bonsall et al. 1999 for an overview). Some behavioural work has also been
carried out, for example Rämä et al. (2000) found that VMS warning signs reduced the mean speed on
slippery roads by 1–2 km/h. Other research on VMS has concentrated on legibility criteria in terms of
character luminance (Kerr et al. 1987, Padmos et al. 1987, Colomb et al. 1991, Upchurch et al. 1992, Ullman
et al. 2001). Research has indicated that drivers require approximately 1s per word presented on a VMS in
order to perceive and process the information (Dudek et al. 1986). A limit of eight words, for drivers
travelling at motorway speeds, was deemed to be the maximum that can be handled within the limits of
human information processing.
The most common use for VMS is to display speed limit restrictions: some studies have reported good
compliance rates, especially when the reason for the restriction is also displayed (McCoy et al. 1995, Garber
et al. 1995). This is the main advantage of VMS: messages can be made to appear personalised and up-to-
date. Their inclusion in a highway authority's traffic management plan is therefore understandable.
1.3. Bilingual signing
There has been little work carried out which has attempted to evaluate if drivers are able to glean the
appropriate information from bilingual signs, without there being a degradation in their driving performance.
Attempts to discover the optimum method of displaying bilingual text have been limited in scope, usually due
to financial and technological constraints. For example, Anttila et al. (2000) tested an alternating bilingual
sign (swapping between Swedish and Finnish every 2 seconds) against one which displayed the messages
simultaneously. By recording eye movements, they inferred the visual demand of the signs. However the
study was limited to one level of sign complexity and did not assess any other (driving) performance
measures.
Research on static bilingual signs has confirmed increased reading times - by up to 15% - compared to
their monolingual counterparts (Rutley 1972, Lesage 1978, Anttila et al. 2000). This is despite the fact that
drivers need only read half of the information presented in order to glean the appropriate message. An
increase in reading time for a particular sign would logically be accompanied by a decrease in the amount of
visual attention paid to the road. Whether this leads to an increase in accident risk depends on the context of
the driving situation (e.g. cars ahead, relative speed, proximity of cars in the adjacent lane). When
considering the visual distraction of in-car systems, guidelines exist as to 'how much' distraction is acceptable
(e.g. ISO 1997, Zwahlen et al. 1988). However, there is no baseline against which we can evaluate the level
of visual distraction posed by road signs. As a result, researchers and policy makers have simply sought to
minimise the reading time.
As the English and Welsh languages use the same alphabet (although the combinations of letters are
different), when a monolingual driver is presented with a bilingual sign, they search the sign in order to locate
the appropriate information. This may mean that, in effect, a driver has to read, or at least perceive, the whole
sign in order to determine the relevant text. In Japan, where most signs on major roads are bilingual
(Japanese/English), this is less of a problem as the two character sets are very different. To minimise the
overall reading time where the character sets are the same, two approaches could be adopted. Firstly, the
sequence of the languages could be optimised such that drivers learn that, for example, the English text is
always uppermost. Secondly, attempts to increase the differentiation between the two languages could be
made by using different colours or fonts.
There are a number of fundamental questions to be addressed in order to evaluate and optimise bilingual
sign configuration. First, how do drivers respond to bilingual signs and does this have implications for traffic
safety? If so, can reading times for bilingual signs be reduced by using one of the methods described above?
The study presented here aimed to answer these research questions using a driving simulator and a number of
bilingual VMS, of varying complexities.
1.4. The current study
There are a number of methodological issues relating to the evaluation of sign readability. Luoma (1991) and
Martens (2000) provide an overview of a number of techniques that could be used and their limitations. Such
techniques include the recording of eye movements, the use of verbal reports, behavioural responses and sign
recall. The review suggests that each of these methodologies has its drawbacks and that, if possible, more
than one method should be used. For example, whilst eye movements can pinpoint the direction of gaze, they
do not necessarily measure attention. Verbal reports could be used as a supplementary measure although
under high workload conditions these may be dispensed with by the driver. Behavioural responses are
considered to be the most useful measure of sign perception and comprehension; however these rely on using
signs that are instructional in nature in order to provide measurable responses. Overall, a behavioural response
is arguably the most ecologically valid measure for road safety evaluations and this provides the focus of this
study.
The study was designed to evaluate if drivers were able to maintain their normal level of driving
performance whilst encountering different types of mono and bilingual VMS. A driving simulator was used to
ensure repeatability of scenarios and the collection of detailed driver behaviour. While driving a simulated
section of rural motorway, participants were presented with a range of VMS signs and instructed to read aloud
the text in their preferred language as soon as they were able. They were also asked to carry out any
instructions indicated on the signs. The complexity of the signs was varied by increasing the amount of
information available to the driver. This allowed the direct comparison of, for example, a four-line bilingual
sign with a four-line monolingual one. The VMS were presented to drivers under varying workload conditions
and their behaviour compared using measures of speed, lateral position and headway to a lead vehicle. In
addition to measuring driver performance, it was thought useful to try and quantify the actual amount of time
it would take a driver to process the VMS. This was achieved by including a number of signs that gave drivers
an instruction. The time taken for them to respond to this instruction was taken as a measure of response time.
Both monolingual and bilingual participants were recruited for the study, as little is known about how
bilingual drivers use bilingual signs. For example, as they are able, do they read both languages - or can they
'limit' themselves to reading just one of them? Linguistics research suggests that bilinguals never fully
disengage from the other language (Green 1998), thus creating cross-language competition (Kroll et al.
2001).
2. Method
2.1. Participants
Twenty-four participants took part in the experiment. The sample comprised of an equal number of
monolinguals (English) and bilinguals (English and Welsh). Monolingual participants were recruited using an
existing database, whilst bilingual drivers were recruited via local Welsh clubs and social gatherings. There
were no difference found, however, between these two groups, and therefore this variable is not considered in
the results section. The sample was balanced for gender, and had a mean age of 28 years (SD 5.25).
Participants had held their driving licences for a mean of 10.2 years (SD= 8.3) and had an average annual
mileage of 8200 miles (SD= 3040). They were paid for their participation. Participants were instructed to read
aloud the text on the VMS in their preferred language as soon as they were able to read it. They were also
asked to carry out any instructions indicated on the signs.
2.2. Experimental environment
The study was carried out in the Leeds Advanced Driving Simulator, allowing the controlled and repeatable
testing of sign configurations. The simulator is based on a complete Rover 216GTi, with all of its driver
controls and dashboard instrumentation still fully operational. A real-time, fully textured and anti-aliased, 3-D
graphical scene of the virtual world is projected on a 2.5 m radius cylindrical screen in front of the driver.
This scene is generated by a SGI Onyx2 Infinite Reality2 graphical workstation. A Roland digital sound
sampler creates realistic sounds of engine and other noises via two speakers mounted close to each forward
road wheel. The projection system consists of five forward channels, the front three at a resolution of 1280 x
1024 pixels. The images are edge-blended to provide a near seamless total image, and along with two
peripheral channels (640 x 480 each), the total horizontal field of view is 230°. The vertical field of view is
39°. A rear view (60°) is back projected onto a screen behind the car to provide an image seen through the
vehicle's rear view mirror. For this study, the frame rate was fixed to a constant 30Hz. Although the simulator
is fixed-base, torque feedback at the steering wheel is provided via a motor fixed at the end of the steering
column and a vacuum motor provides the brake pedal booster assistance. Data are collected at the frame rate.
A two-lane road of UK motorway design standard was used for the experimental road, with a speed limit
of 112kph (70mph). The road was comprised of 500m straight sections separated by 'filler' pieces of road.
The VMS display was triggered 50m into the straight section and located 250m later. The road was
approximately 32 km in length, with a light volume of surrounding traffic. In the outside lane, occasional
vehicles travelling at 130kph (80 mph) passed the driver. A snapshot of the scene is shown in figure 1.
Figure 1. Four-line VMS sign within the simulated environment
Drivers were asked to follow a lead car until they reached the end of the route. They were asked not to
overtake this car and to remain at all times in the left lane. Participants were required to drive the
experimental road twice, with the order counterbalanced. The difference between the two roads lay in the
behaviour of the lead car. On one of the roads the lead car remained at a constant speed (approximately
110kph). On the second road, the lead car varied its speed from 96-120kph on the road section where the signs
were positioned, table 1.
Table 1. Speed profile of lead car in the high workload condition
Location from start
of section (m)
0 100 200 300 400 500
Lead car speed
(kph)
112 120 108 96 104 112
These two roads provided two workload conditions, with the second requiring more effort from the driver
to maintain a safe headway to the vehicle in front. Participants underwent a familiarisation procedure before
taking part in the main experiment. The simulator offers an ideal testing environment, as the readability of
signs can be tested dynamically, i.e. whilst driving. This has obvious advantages over laboratory testing.
2.3. Sign configuration
Real-world VMS signs use capital letters only, with a case height equivalent to a 320 mm font (Highways
Agency 1994). In order to ensure validity, the signs were designed such that they allowed the driver the same
reading time as in the real world; this meant using a larger sign due to the reduced image clarity in the driving
simulator. The characters were designed according to the Design Manual for Roads and Bridges (Highways
Agency 1994) that specifies the pixels used in each character. The VMS signs were activated when the
simulator passed a trigger point at the appropriate reading distance to ensure drivers were exposed to the sign
for the same amount of time. Five VMS were presented to drivers as shown in figure 2.
[Insert Figure 2 about here]
English
English
En
g
lish
English
English
English
English
English
English
Welsh
Welsh
English
Welsh
Sign 1 Sign 2 Sign 3 Sign 4 Sign 5
Figure 2. Sign types
The signs allowed the investigation of the effects of increasing lines and of increasing languages. For
example, by comparing Signs 1,2 and 3, it was possible to discover if drivers performed differently as the
number of relevant lines of text increased. Alternatively the comparison of Signs 1 and 4 and that of Signs 2
and 5, allowed the analysis of the effect of increasing the number of languages on the sign (whilst the number
of relevant lines of text was held constant). The actual text used for these signs was in the form of
information, for example 'FOR MIDLANDS, USE M4'.
A number of signs were also included which instructed the driver to use various vehicle controls. These
four-line signs included three lines of road traffic related text with the fourth line being the instruction (e.g.
flash your headlights). If the driver did not perform the action it was concluded that the fourth line could not
be read in the available time.
A priori, it was hypothesised that four-line signs would be the most likely candidate to affect driver
performance. Therefore a number of four-line signs were included to allow the testing of one of the most
promising demarcation techniques - a separation line between the two languages (Jamson 2004). Driving
performance when reading this sign was compared to that when reading the same sign, but without a
separation line.
2.4. Data collection and analysis
Driver performance was measured using speed, lateral position and headway to a lead vehicle. The data were
collected on the 500m straight sections of road and divided into ten sections of 50m each. This enabled the
analysis of data as the driver approached the sign to discover, for example, which signs produced changes in
driver behaviour, where these changes occurred and how long they lasted for. Driver performance on these
sections was compared to that collected on a baseline section where no sign was present. The data were
analysed separately for the two workload conditions, as the behaviour of the lead vehicle would directly affect
the participant’s behaviour. For example when the lead car slowed, the participant might also reduce their
speed to maintain their desired headway – not as a direct result of them attempting to read the VMS.
As a measure of global subjective mental workload, the NASA TLX (Byers et al. 1989) was administered
following each drive. This required drivers to rate their driving in terms of mental demand, physical demand,
time pressure, performance, effort and frustration level. A bipolar scale represents each of these items;
participants place a line on the scale between the two extremes of the item to indicate the strength of the
attribute. This questionnaire was used to evaluate perceived differences in the two workload conditions.
3. Results
The main focus of this experiment was the detection of changes in speed and headway in the vicinity of the
signs, and establishing which signs have the most impact. For each of the datasets (Low Workload and High
Workload), analyses of variances were performed using Sign Type and Road Section as repeated measures.
In the High Workload condition, only the minimum time-headway data was considered because, as explained
above, the lead car’s behaviour would directly affect absolute values of speed and headway. Driver Type
(monolingual or bilingual) was also included as a between-subjects variable. However, no significant effects
of this variable were found in any of the analyses.
First, paired t-tests on the separate components of the NASA TLX confirmed that participants perceived
their mental workload to be higher when they were required to monitor the behaviour of the lead car whilst
reading the VMS, figure 3. Workload scores for mental demand (t(23)= -6.18, p<.001), performance (t(23)=
-5.99, p<.001) and effort (t(23)= -4.83, p<.001) were all higher in the High Workload condition.
0
20
40
60
80
100
mental
demand
physical
demand
time pressure performance effort frustration
Workload score
Low workload
High workload
Figure 3. Mental workload scores (±SE)
3.1. Driver performance
Variables of speed, lateral position and headway were analysed using repeated measures ANOVA with two
factors. The first factor, Road Section, corresponded to the ten road segments in the proximity of the VMS.
The second factor, Sign type, had 6 levels corresponding to the 5 signs described in section 2.3 and a baseline
road section (no sign present). Post-hoc contrasts were carried out where appropriate using the Bonferroni
correction to minimise familywise error.
3.1.1 Speed. Mean speeds were calculated across each of the ten sections. In the Low Workload condition,
there were main effects of Road Section with drivers reducing their speed in the vicinity of VMS signs,
(F9,207 = 32.38, p<.001). Post-hoc pairwise comparisons showed that mean speeds in sections 3-7 (i.e. the
period for which the driver was engaged in reading the signs) were lower than in all other sections, see figure
4. Mean speeds in the first and last sections were comparable; thus drivers increased their speed after having
read the sign such that they returned to their observed speed before the sign was activated.
100
102
104
106
108
110
112
114
116
118
120
0-50m
5
1
-
10
0m
101-15
0m
151-20
0
m
201-250m
251-300m
301
-
350m
351
-
400m
40
1-
450m
45
1-
5
0
0m
Distance along road section (m)
Mean speed (kph)
Baseline
VMS
inactive
VMS
active
VMS
position
VMS out of
sight
Figure 4. Mean speed (±SE) through the road sections (Low Workload)
In addition, there was a main effect of Sign type (F
5,115
= 162.52, p<.001) and an interaction between Sign
Type and Road Section (F
45,1035
= 27.82, p<.001). With regards to the main effect, pairwise comparisons
showed that mean speeds were lower for both the monolingual and bilingual four-line signs (in comparison to
the baseline condition). Across all road sections, the average speed reduction was in the order of 3kph (2mph).
Speed reductions for these four-line signs were similar. The Road Section*Sign Type interaction reveals more
about the speed reductions in particular road sections for the different signs. As can be seen in figure 5, the
one and two-line signs had no effect on drivers' speed. This was confirmed with repeated contrasts carried out
separately for each sign type: on the whole there were no significant differences between the road sections.
90
94
98
102
106
110
114
118
0
-
5
0m
51
-
1
00
m
10
1-15
0m
151-20
0
m
201
-
250m
25
1-
3
0
0m
3
01-35
0
m
351-400m
40
1-
4
5
0m
45
1-50
0m
Distance along road section (m)
Mean speed (kph)
Baseline
1 line
2 lines, Monolingual
4 lines, Monolingual
2 lines, Bilingual
4 lines, Bilingual
VMS
inactive
VMS
active
VMS
position
VMS out of
sight
Figure 5. Mean speeds for different sign types (Low Workload)
For each of the four-line signs, changes in speed are apparent as drivers approach and pass the VMS. For
each, a maximum reduction of approximately 11kph (7mph) was observed by the time drivers had reached the
sign. This represents a deceleration rate of approximately 0.37ms
-2
and is the result of engine braking, i.e.
drivers simply released the accelerator (as opposed to using the foot brake). Slight differences were observed
between these two signs in that when confronted with the bilingual sign, drivers reduced their speed earlier
(p<.01) and then did not attain their baseline speed by the time they had reached the last road section
(p<.001). Therefore, there seems to be evidence of a 'hang-over' effect, perhaps due to the fact that drivers
were still trying to process the four-line bilingual sign.
The High Workload condition produced similar results in terms of mean speeds. However this measure is
of less interest than that of headway, as this provides us with a measure of the safety margin that drivers are
willing to adopt whilst reading the VMS.
3.1.2. Car following. A car following task was used to evaluate drivers' ability to adapt their driving
behaviour whilst reading the VMS. It measures the ability of drivers to maintain a safe headway to a lead
vehicle and has been used as a measure of driving safety in numerous studies of impairment (e.g. Brookhuis et
al. 1987, Ramaekers et al. 1994). Only headways of less than 6 seconds were used as higher values assume
that drivers are not intending to follow the lead vehicle (Vogel, 2002). Minimum time headways were
calculated to reflect the smallest safety margin that drivers were willing to adopt. The minimum headways for
the Low Workload condition are shown in figure 6. Main effects of both Sign Type (F5,115 = 52.50, p<.001)
and Road Section (F9,207 = 5.21, p<.001) were found. Pairwise comparisons indicated that drivers adopted
longer minimum headways when they encountered four-line signs compared to the baseline condition. This is
simply an effect of the speed reductions found in the previous section - when drivers were reading the four-
line signs, they reduced their speed. This would result in a longer headway to the vehicle in front.
1.5
1.7
1.9
2.1
2.3
2.5
2.7
2.9
3.1
3.3
3.5
0-50m
51-100m
1
01
-15
0
m
1
51
-20
0
m
2
01
-2
50
m
25
1
-30
0
m
30
1
-35
0
m
35
1
-40
0
m
40
1
-45
0
m
45
1
-50
0
m
Distance along road section (m)
Minimum headway (s)
Baseline
1 line
2 lines, Monolingual
4 lines, Monolingual
2 lines, Bilingual
4 lines, Bilingual
VMS
inactive
VMS
active
VMS
position
VMS out of
sight
Figure 6. Minimum headways for the Low Workload condition
Of more interest is what happens when drivers are faced with a higher workload. Are they able to adapt
their speed to that of the vehicle in front in order to maintain the same headway margin? A Sign Type*Road
Section interaction, (F
45,1035
= 5.54, p<.001) in the High Workload headway data suggests not, figure 7. It
appears that drivers are sufficiently loaded with the additional task of reading the four-line signs such that
they are willing to accept a smaller lead time to the vehicle in front. On average, their headway reduced by
one second, whilst reading the signs.
1.1
1.3
1.5
1.7
1.9
2.1
2.3
2.5
2.7
2.9
3.1
3.3
3.5
0-50m
51-100m
1
01
-15
0
m
1
51
-20
0
m
2
01
-2
50
m
25
1
-30
0
m
30
1
-35
0
m
35
1
-40
0
m
40
1
-45
0
m
45
1
-50
0
m
Distance along road section (m)
Minimum headway (s)
Baseline
1 line
2 lines, Monolingual
4 lines, Monolingual
2 lines, Bilingual
4 lines, Bilingual
VMS
inactive
VMS
active
VMS
position
VMS out of
sight
Figure 7. Minimum headways for the High Workload condition
3.1.3. Lateral deviations Standard deviation of lateral position (SDLP) was used as an indicator of drivers'
ability to control their vehicle and prevent it wandering out of lane. There were no significant main or
interaction effects of Sign Type or Road Section on SDLP. This is likely to be an effect of the reductions in
speed noted above, such that drivers were able to control their steering whilst reading the signs because they
compensated by reducing their speed.
3.1.4.
Response accuracy The data presented above suggest that drivers have difficulty in reading the four-
line signs without adapting their driving behaviour. Whether or not they are able to actually comprehend all
four lines of the sign was tested by the inclusion of a number of one, two and four-line signs that required
them to perform an action. The one-line signs simply displayed the instruction e.g. 'Flash your headlights'.
The two-line signs contained one line of information, followed by the instruction. The four-line signs
contained three lines of instructional text, with the last line being an instruction. Drivers' ability to perform
the instruction was used as a measure of whether or not drivers had enough time to read all four lines.
Response time was also recorded. Different types of instructions were provided to minimise learning effects.
Ninety-nine % of drivers succeeded in carrying out the instruction when faced with the one and two-line
signs. This fell to 86% for four-line signs. Thus even when drivers compensated by lowering their speed, they
still performed worse when encountering these longer signs. The second element of this assessment concerned
the time drivers took to respond to the instruction on the sign. A significant effect for the number of lines of
relevant text was found (F
[2,392]
=168.84, p<0.001), table 2.
Table 2. Response times
No. of lines
of relevant
text
Mean RT
(s)
Std. Error
(s)
1 2.81 0.26
2 3.27 0.11
4 6.43 0.24
An incremental increase in response time was observed as the number of lines of relevant text increased.
Post-hoc testing showed that there was a reliable difference between two and four-line signs only. In other
words, whilst there was a relatively small increase in response time when the number of lines doubled from
one to two (16%), response times doubled when lines increased from two to four lines.
3.1.5 Demarcation A separation line was introduced on some signs to investigate whether this type of
demarcation technique could improve the readability of the four-line bilingual signs. No significant
differences were found across any of the variables when such a sign was compared to an identical one without
a separation line.
4. Conclusions
Participants in this study were required to read and react to a number of monolingual and bilingual signs
under both high and low workload driving conditions. Subjective mental workload scores suggested that
drivers were able to differentiate between these two conditions, in that they reported that their performance
deteriorated and they felt they had to increase their effort and as a consequence experienced higher mental
demand. Whilst both monolingual and bilingual drivers were recruited for this study, no differencess were
found between the two groups.
Drivers' ability to maintain their baseline driving performance whilst reading a selection of VMS was
investigated. This driving performance consisted of both longitudinal and lateral measurements and was
monitored on their approach to and passing of the signs. It was of interest to see if and how drivers adapted
their behaviour and also whether there was a period of 'recovery', after having passed the sign. The most
interesting analyses were that of the interaction between Sign Type and Road Section. This provided a
temporal analysis of behaviour making it possible to pinpoint where any changes in behaviour occurred,
depending on the type of sign drivers encountered.
Drivers were able to read one and two-line monolingual signs and two-line bilingual signs without
disruption to their driving behaviour. However, both four-line monolingual and four-line bilingual signs
impacted on driver performance, when compared to the baseline condition. Drivers reduced their speed by
approximately 11 kph (7 mph), over a distance of 250m, in order to read these types of signs, achieved by
engine braking. The fact that this speed reduction was the same for each of the four-line signs, suggests that
drivers are perhaps reading, or at least searching, the irrelevant text on the bilingual sign. In addition, whilst
drivers were able to recover from the monolingual sign (i.e. they returned to their previous speed), drivers
were still travelling slowly after they had passed a four-line bilingual sign. This suggests that the drivers were
perhaps still trying to process the information and work out whether they had missed something vital. This
type of effect has been observed in a number of studies, whereby increases in cognitive load have led to speed
reductions, e.g. whilst using a mobile phone (Burns et al. 2002, Haigney et al. 2000, Reed et al. 1999) or a
navigation system (Srinivasan et al. 1997). Researchers have interpreted this in different ways; some believe it
is a compensatory process that drivers engage in to reduce their accident risk, whilst others purport that
drivers are simply paying less attention to the task of speed regulation.
With regards to response times, these increased with the number of lines of relevant text. Drivers were able
to read and respond to signs that were one or two lines in length within a relatively similar timescale.
However, when drivers were required to read four lines of text, their response time was significantly higher.
This suggests that drivers had to 'chunk' the information, using several glances to the sign, in order to
complete the task - although this could only be confirmed using additional data collection techniques (eye-
movement recording). This chunking technique is a reasonable one to adopt, as long as the sign is within the
driver's sight distance for long enough. For example, based on the response time of 6.43s, driver would
require a four-line sign to be in legible view for approximately 200m (at motorway speed).
5. Practical implications
The results of this experiment suggest that in the High Workload condition, where drivers were required to
adapt their headway to the vehicle in front, observed minimum headways decreased. Drivers appeared either
to accept a shorter safety margin or not notice that the vehicle in front was decelerating. Either situation could
impact on both the drivers' ability to avoid a collision and the severity of one should it occur.
The practical implications of this speed reduction and reduction in following headway are decreased traffic
flow stability and an associated increase in accident risk. Drivers reading long (four-line) messages reduce
their speed. Following drivers are reading the same message and while also beginning to reduce speed might
not respond appropriately to the slowing of the lead vehicle. As inter-vehicle headways decrease, the margins
available for drivers to take avoidance action are reduced and accident risk increases. Avoidance actions,
when required, become more severe and 'shockwaves' develop in the traffic flow. The shockwave
phenomenon is most often observed on motorways, where decreases in speed occur not only at the site of the
problem (or sign) but also further back in the traffic stream where the reason for the speed reduction would
not be visible to drivers. Such speed variation has been linked to increased accident risk (Garber et al. 1989,
Aljanahi et al. 1999). These studies suggest that reducing speed variance and 85th percentile speed (the speed
exceeded by the fastest 15% of drivers) are the most important aspects of speed to target for improved road
safety.
This study has attempted to benchmark the possible effects that bilingual VMS might have on driver
performance by comparing them with monolingual signs of varying lengths. The results suggest that drivers
encountering four-line bilingual signs attempt to read the whole sign and that this process takes in the order of
6s. This is probably split into smaller chunks of time, separated by glances back to the road ahead. Highway
engineers should take this into account when locating four-line VMS so that drivers have adequate sight
distance.
With drivers reacting to such signs by slowing down, it is sensible to try and limit the visual distraction by
the use of separation techniques. One such technique was tested in this study (a separation line) but it was not
found helpful to the drivers. This could be a novelty effect in that drivers need to learn such techniques in
order for them to be useful. Thus, multiple presentations may have been needed in order to draw out these
advantages and further studies should concentrate on developing effective demarcation techniques.
References
ALJANAHI, A.A.M., RHODES, A.H. and METCALFE, A.V. 1999, Speed, speed limits and road traffic accidents
under free flow conditions. Accident Analysis and Prevention, 31, pp.161-168.
A
NDERSON, J.R. 1990, Cognitive psychology and its implications, 3
rd
edition. (New York: W.H.Freeman and
Company).
A
NTTILA, V., LUOMA, J. and RÄMÄ, P. 2000, Visual demand of bilingual message signs displaying alternating
test messages, Transportation Research Part F, 3, pp.65-74.
BONSALL, P.W. and PALMER, I.A. 1999, Route choice in response to Variable Message Signs: factors
affecting compliance, in R.
Emmerink and P. Nijkamp (eds.), Behavioural and network impacts of
driver information systems. (Aldershot
: Ashgate), pp.181-214.
BROOKHUIS, K.A., VOLKERTS, E.R. and O’HANLON, J.F. 1987, The effects of some anxiolytics on car
following performance in real traffic, in P.C Noordzij and R. Roszbach (eds), Alcohol, Drugs and
Traffic Safety, (Excerpta Medica: Amsterdam), pp.223-226.
BURNS, P.C., PARKES, A., BURTON, S., SMITH, R.K. and BURCH, D. 2002, How dangerous is driving with a
mobile phone? Benchmarking the impairment to alcohol. TRL Report 547. TRL Limited, Crowthorne,
UK.
BYERS, J.C., BITTNER, A.C. and HILL, S.G. 1989, Traditional and raw task load index (TLX) correlations: Are
paired comparisons necessary? in A. Mital (ed.), Advances in Industrial Ergonomics and Safety I.
(Taylor and Francis: London), pp.481-485.
COLOMB, M. and HUBERT, R. 1991, Legibility criteria of Variable Message Signs, in A.G. Gale, I. Brown,
C.M. Haslegrave, I. Moorhead and S.P. Taylor (eds.), Vision in Vehicles III. (Amsterdam: North-
Holland), pp.309-316.
DUDEK, C.L. and HUCHINGSON, R.D. 1986, Manual on real-time motorist information displays. Report No.
FHWA-IP-86-16. FHWA, US Department of Transportation, Washington DC.
FORBES, T.W., SNYDER, T.E. and PAIN, R.F. 1965, Traffic sign requirements. A review of factors involved,
previous studies and research needed, Highway Research Record, 70, pp.48-56.
G
ARBER, N.J. and GADIRAJU, R. 1989, Factors affecting speed variance and its influence on accidents,
Transportation Research Record, 1213, pp.64-71.
GARBER, N.J. and PATEL, S.T. 1995, Control of vehicle speeds in temporary traffic control zones (work zones)
using changeable message signs with radar, Transportation Research Record, 1509, pp.73-81.
GREEN, D. 1998, Mental control of the bilingual lexico-semantic system, Bilingualism: Language and
Cognition, 1, pp.67-81.
H
AIGNEY, D.E., TAYLOR, R.G. and WESTERMAN, S.J. 2000, Concurrent mobile (cellular) phone use and
driving performance: task demand characteristics and compensatory processes. Transportation Research
Part F, 3, pp.113-121.
H
ALL, R.D., MCDONALD, M. and RUTLEY, K.S. 1991, An experiment to assess the reading time of direction
signs, in A.G. Gale, I. Brown, C.M. Haslegrave, I. Moorhead and S.P. Taylor (eds.), Vision in Vehicles
III. (Amsterdam: North-Holland), pp.333-350.
HIGHWAYS AGENCY 1994, Design Manual for Roads and Bridges, Volume 8 - Traffic Signs and Lighting.
http://www.official-documents.co.uk/document/deps/ha/dmrb/index.htm
INTERNATIONAL ORGANISATION FOR STANDARDISATION 1997, Road vehicles - Transport information and
control systems - Man machine interface - Definitions and metrics related to the measurement of driver
visual behaviour. ISO/TC22/SC13/WG8/N115. Geneva: International Organisation for Standardisation.
J
AMSON, S.L (2004), Evaluation of techniques to improve the legibility of bilingual Variable Message Signs.
Advances in Transportation Studies, IV, pp.71-88.
J
OHANSSON, G. and BACKLUND, F. 1970, Drivers and road signs. Ergonomics, 13, pp.749-759.
KERR, J. SNELGAR, R. JORDAN, T. EMMERSON, P. and LINFIELD, P. 1987, Optimum display factors for light
emitting variable message road signs, in A.G. Gale, I. Brown, C.M. Haslegrave, I. Moorhead and S.P.
Taylor (eds.), Vision in Vehicles III. (Amsterdam: North-Holland), pp.277-288.
KINNEY, G.C., MARSETTA, M. and SHOWMAN, D.J. 1966, Studies in display symbol legibility, part XXI. The
legibility of alphanumeric symbols for digitised television (ESD-TR-66-117). (Bedford, MA: The Mitre
Corporation).
KROLL, J. and TOKOWICZ, N. 2001, The development of conceptual representation for words in a second
language, in J. Nicol (ed), One mind, two languages: bilingual language processing (Explaining
Linguistics). (Malden, Massachusetts and Oxford: Blackwell), pp.35-52.
LESAGE, P.B. 1981, Design and comprehension of bilingual traffic signs. Report CR 8101. Transport Canada
Research and Development Centre, Road and Motor Vehicle Traffic Safety Branch, Ottawa, Canada.
LUOMA, J. 1991, Perception of highway traffic signs: Interaction of eye movements, recalls and reactions, in,
A.G. Gale, I.D. Brown, C.M. Haslegrave, I. Moorhead and S. Taylor (eds.), Vision in Vehicles-III.
(Amsterdam: North-Holland), pp.325-332.
MARTENS, M. 2000, Assessing road sign perception: a methodological review, Transportation Human
Factors, 2, pp.347-357.
MCCOY, P.T., BONNESON, J.A. and KOLLBAUM, J.A. 1995, Speed reduction effects of speed monitoring
displays with radar in work zones on interstate highways, Transportation Research Record, 1509,
pp.65-72.
NEISSER, U. 1967, Cognitive psychology. (New York: Appleton).
P
ADMOS, P., VAN DEN BRINK, T., ALFERDINCK, J. and FOLLES, E. 1987, Matrix signs for motorways: system
design and optimum light intensity, in A.G. Gale, M.H. Freeman, C.M. Haslegrave, P.A. Smith and S.P.
Taylor (eds.), Vision in Vehicles II, (Amsterdam: North-Holland), pp.269-276.
RÄMÄ, P. and KULMALA, R. 2000. Effects of variable message signs for slippery road conditions on driving
speed and headways. Transportation Research Part F, 3, pp.85-94.
R
AMAEKERS, J.G. and O’HANLON, J.F. 1994, Acrivastine, terfenadine and diphenhydramine effects on
driving performance as a function of dose and time after dosing, European Journal of Clinical
Pharmacology, 47, pp.261-266.
R
EED, M.P. and GREEN, P.A. 1999, Comparison of driving performance on-road and in a low-cost simulator
using a concurrent telephone driving task, Ergonomics, 42, pp.1015-1037.
RUTLEY, K. 1972, An investigation into bilingual (Welsh/English) Traffic signs. Report LR 475. Transport
Research Laboratory, Crowthorne, UK.
SCHNEIDER, W. and SHIFFRIN, R.M. 1977, Controlled and automatic human information processing: I.
Detection, search and attention, Psychological Review, 84, pp.1-66.
SPERLING, G.A. 1960, The information available in brief visual presentation. Psychological monographs, 74.
SRINIVASAN, R. and JOVANIS, P.P. 1997, Effect of in-vehicle route guidance systems on driver workload and
choice of vehicle speed. Findings from a driving simulator experiment, in Y.I. Noy (ed). Ergonomics
and safety of intelligent driver interfaces, (New Jersey: Laurence Erlbaum Associates), pp.97-114.
TULVING, E., MANDLER, G. and BAUMAL, R. 1964, Interaction of two sources of information in tachistoscope
word recognition, Canadian Journal of Psychology, 18, pp.62-71.
ULLMAN, G.L. and DUDEK, C.L. 2001, Effect of roadway geometrics and large trucks on variable message
sign readability, Proceedings of 2001 Annual Meeting of the Transportation Research Board,
(Washington DC:US).
UPCHURCH, J., ARMSTRONG, J.D., BAAJ, M.H. and THOMAS, G.B. 1992, Evaluation of variable message signs:
target value, legibility, and viewing comfort, Transportation Research Record, 1376, pp.35-44.
VOGEL, K. 2002, What characterizes a "free vehicle" in an urban area? Transportation Research, Part F, 5,
pp.313-327.
ZWAHLEN, H.T., ADAMS, C.C. and DEBALD, D.P. 1988, Safety aspects of CRT touch panel controls in
automobiles, in A.G. Gale, M.H. Freeman, C.M. Haslegrave, P. Smith and S.P. Taylor (eds.) Vision in
Vehicles II, (Amsterdam: North-Holland), pp.335-344.
