Fire Alarm Signal Recognition
Proulx, G.; Laroche, C.; Jaspers-Fayer, F.;
Lavallée, R.
IRC-IR-828
www.nrc.ca/irc/ircpubs
NRC-CNRC
Fire Alarm Signal Recognition
Guylène Proulx, Chantal Laroche, Fern Jaspers-Fayer and
Rosanne Lavallée
Internal Report No. 828
Date of issue: June 2001
ii
Fire Alarm Signal Recognition
Authors affiliations:
Guylène Proulx, Ph.D. and Fern Jaspers-Fayer
Fire Risk Management Program
Institute for Research in Construction
National Research Council Canada
and
Chantal Laroche, Ph.D. and Rosanne Lavallée
Audiology and Speech-Language Pathology
Faculty of Health Sciences
University of Ottawa
NRC Contract: B4521
iii
Fire Alarm Signal Recognition
G. Proulx, C. Laroche, F. Jaspers-Fayer, R. Lavallée
TABLE OF CONTENTS
ACKNOWLEDGMENTS ………………………………………………………………………
v
EXECUTIVE SUMMARY ……………………………………………………………………..
vi
RÉSUMÉ ..……………………………………………………………………………………… viii
1.0 INTRODUCTION ..…………………………………………………………………………
1
2.0 LITERATURE REVIEW ………………………………………………………………….
2
3.0 RESEARCH OBJECTIVES ..…………………………………………………………….
5
4.0 METHODOLOGY …………………………………………………………………………
6
4.1 Participants ………………………………………………………………………..
6
4.2 Signals Tested …………………………………………………………………….
7
4.3 Materials …………………………………………………………………………..
9
4.4 Procedure ………………………………………………………………………..
10
5.0 STUDY RESULTS ……………………………………………………………………….
11
5.1 Respondent Profile ………………………………………………………………
11
5.2 Merging Data from the Three CDs …………………………………………….
13
5.3 Signal Recollection ………………………………………………………………
14
5.4 Signal Identification ……………………………………………………………… 15
5.5 Signal Recollection and Identification ………………………………………….
17
5.6 Signal Perceived Urgency ………………………………………………………
20
5.7 Signal Identification and Perceived Urgency …………………………………
21
6.0 CONCLUSIONS ………………………………………………………………………….
22
7.0 RECOMMENDATIONS………………………………………………………………….
23
8.0 FUTURE WORK …………………………………………………………………………
26
9.0 REFERENCES …………………………………………………………………………..
28
APPENDIX 1: Spectrum Analysis of the Signals Tested …………………………………
30
NRC Contract: B4521
iv
LIST OF TABLES
Table 1: Signal Presentation Order on Each CD………………………………………….…
8
Table 2: Population Distribution of Canada and Study Sample…………………………… 12
Table 3: Results to the T-3 Signal for Each CD ……………………………………………. 13
Table 4: Number of Occupants who Confirmed Having Heard a Signal Before………… 14
Table 5: Number of Occupants who Correctly Identified Each Signal…………………… 16
Table 6: Wrong Identification of the Temporal-Three ……………………………………… 19
Table 7: Difference in Perceived Urgency between T-3 and Other Signals……………… 20
Table 8: Identification of any Signal as a Fire Alarm and Perceived Urgency …………. 21
LIST OF FIGURES
Figure 1: Gender and Age Distribution of the Study Sample………………….…………
12
Figure 2: Recall Percentage of Each Signal ……………………………….………………
15
Figure 3: Percentage of Signals Identified .……..…………………………….……………
17
Figure 4: Recollection and Identification of the Temporal-Three…………………………
18
Figure 5: Percent of Each Signal Recalled and Identified Correctly…………………….
18
NRC Contract: B4521
v
Acknowledgements
This project could not have been possible without the support of Mr. Arnold Garson, Vice
President of Industry Affairs, Siemens Building Technologies Limited. Mr. Garson readily gave
his support to conducting a study of the Temporal-Three signal and his enthusiasm for
furthering the knowledge of fire alarms and life-safety was greatly appreciated. We would like to
thank both him and his organization for their support and hope that we will collaborate on other
productive projects in the future.
We would also like to thank Mr. Mark Faubert and Mr. Bill Thistle, responsible for fire
safety and security at the Ottawa International Airport, for allowing us to gather data at the
airport. After passing the security check, the waiting area of the airport was an ideal location to
ask people to participate in our study.
As well, we would like to thank Stéphane Denis, Groupe d’Acoustique de l’Université de
Montréal (GAUM), who created the three CDs that present the warning signals that were tested
during the experiment. These tools were an essential part of our study. Mr. Denis was also
responsible for the analysis of the acoustical characteristics of the warning sounds.
We would also like to thank Dr. Russell Thomas, Director of the Fire Risk Management
Program at the National Research Council Canada, for his guidance in the statistical analysis of
our data.
Finally we wish to thank our participants. These busy people graciously agreed to provide
not only fascinating data by answering our survey, but also useful insights into our research
through their astute comments.
This work was jointly funded by Siemens Building Technologies Limited and the National
Research Council Canada (NRC).
NRC Contract: B4521
vi
Executive Summary
The 1995 National Building Code of Canada requires that fire alarm signals sound the
Temporal-Three (T-3) pattern, as defined by the ISO 8201 “Acoustics – Audible Emergency
Evacuation Signal”. This sound pattern has also been required by NFPA 72 since July 1996. It
is intended that the T-3 pattern will become the standardized alarm signal heard around the
world that will unequivocally mean “evacuate the building immediately”. Although new and
refurbished buildings have, for the past 5 years, been equipped with this new signal, no formal
public education has taken place to inform building users about the meaning and response
expected from them when it sounds. In North America, discussions are ongoing regarding the
necessity to develop a public education campaign on the subject of this new evacuation signal,
and whether an automatic recorded message should follow the signal to prompt the public to
evacuate. As a first step, we need to ascertain if the public already recognizes this sound as an
evacuation signal.
The objectives of this project were to assess the public’s recollection, identification and
perceived urgency of the T-3. Data was collected through a field study. Six alarm signals were
recorded on a CD: the T-3, a Car Horn, a Reverse or backup alarm, a fire alarm Bell, the Slow
Whoop alarm, and an industrial warning Buzzer. The presentation orders of the signals were
different on each of the three CDs used to collect data. Members of the public were
approached in buildings such as shopping centers, office buildings, libraries and airports; they
were then asked to listen through headphones to the different sounds. After each signal, the
interviewer asked three questions: “Have you heard this sound before?”, “What do you think this
sound means?” and finally, “How urgent do you feel this sound is on a scale from 1 to 10? 1
means the sound is not urgent at all and 10 means it is extremely urgent.” The first question
tested recollection of the signal, the second tested the ability of the participant to correctly
identify the signal, and the last question rated the perceived urgency of the signal. It was
heavily emphasized to participants that the sounds heard on the CD were from in and around
large buildings, such as the one where they were in at the time of the interview.
In total, 307 participants, representative of the Canadian population in terms of gender
and age distribution, were interviewed for the study. Results showed a significant difference
between how often participants said they could recall the various signals, Q(5,307)=288.15,
p<.001. The Car Horn was recalled the most often, by 97% of the participants, followed by the
NRC Contract: B4521
vii
Reverse alarm by 91%, the Buzzer by 81%, the T-3 by 71%, the Bell by 58% and the Slow
Whoop by 52% of the participants. Respondents who said they had heard a signal before,
however, did not necessarily identify the signal correctly. In fact, the T-3 was identified correctly
only 6% of the time as a fire alarm. The Car Horn was identified correctly 98% of the time, the
Reverse alarm 71%, the Bell 50%, the Slow Whoop 23%, and the Buzzer 2% of the time. There
is a significant difference between the identification of all the signals; x2(5, 307)=555.52, p <
0.001. Consequently, although the participants often reported that they had heard the T-3
before, they could rarely correctly identify it as a fire alarm or evacuation signal. In fact, the T-3
was usually associated with domestic signals such as a busy phone signal or the sound of an
alarm clock.
When asked to rate the urgency of each of the signals, participants rated the T-3 as the
least urgent of all signals. The T-3 received an overall rating of 3.97 on a scale of 1 to 10 (10
being extremely urgent). The Buzzer, Car Horn and Reverse signals received urgency ratings
between 4.91 and 5.60, while the Slow Whoop obtained a rating of 6.01 and the Bell 7.17. It
should be stressed that generally, when a participant identified a signal as a fire alarm, the
urgency rating significantly increased. This finding demonstrates a clear link between the
meaning attached to a signal and the perceived urgency of that signal.
These findings suggest that considerable public education is necessary to improve the
public’s identification of the T-3 signal.
It is further suggested that it is unrealistic to believe that building occupants will start to
evacuate a building as soon as they hear an alarm signal, even if the signal is recalled and
identified correctly as a fire alarm. Supplementary information conveyed through a voice
communication system or by staff will always be needed to confirm to occupants that an
evacuation or relocation is needed. Finally, it is advocated to use the T-3 uniformly as the fire
alarm signal in all buildings, whether the egress strategy in place calls for evacuation or not, to
limit the number of warning signals and therefore improve recognition. With proper training and
sufficient exposure, the T-3 could become the standard fire emergency warning signal, which
occupants will recognize. To receive such a warning is in itself an essential piece of information
to convey to the public before they are instructed to apply specific behaviour during a fire
emergency.
NRC Contract: B4521
viii
RÉSUMÉ
L’édition 1995 du Code National du Bâtiment du Canada (CNBC) exige que les alarmes
incendie émettent le signal du Temporal-Trois (T-3), tel que défini par la norme ISO 8201
"Acoustics-Audible Emergency Evacuation Signal". Ce patron sonore est également exigé par
le NFPA 72 depuis juillet 1996. Il est prévu que le patron sonore du T-3 qui signifie "évacuer
l’édifice immédiatement", devienne le signal sonore standard pour l’évacuation au niveau
international. Malgré l’installation depuis 5 ans de ce nouveau signal à l’intérieur de tous les
édifices neufs ou rénovés qui ont pour exigence, selon le CNBC, d’avoir un système d’alarme
central, aucune campagne officielle de sensibilisation n’a été entreprise pour informer le public
de la présence de ce signal sonore et du comportement à adopter suite à son déclenchement.
En Amérique du Nord, les discussions se poursuivent concernant, d’une part, la nécessité d’une
campagne de sensibilisation pour ce nouveau signal et, d’autre part, la nécessité de faire suivre
systématiquement ce signal par un message enregistré encourageant l’évacuation. Dans un
premier temps, il s’avère justifier de déterminer si ce patron sonore est présentement reconnu
comme un signal d’évacuation par le public en général.
Les objectifs de la présente étude étaient d’évaluer la reconnaissance, l’identification et la
perception d’urgence du T-3. Des données ont été recueillies dans le cadre d’une étude sur le
terrain. Six signaux avertisseurs ont été enregistrés sur un disque compact : le T-3, un klaxon,
une alarme de recul, une cloche d’incendie, le Slow Whoop, et un ronfleur industriel. L’ordre de
présentation des signaux sonores était différent sur chacun des trois disques compacts utilisés
pour la cueillette de données. Des membres du public ont été approchés dans des édifices
telles que des centres d’achats, des édifices à bureaux, des bibliothèques et des aéroports et
ont été invités à écouter les différents signaux à l’aide d’écouteurs. Après chaque signal, trois
questions étaient posées au participant : "Avez-vous déjà endendu ce signal auparavant? ",
"D’après vous, quelle est la signification de ce signal? " et finalement, "Quel degré d’urgence
associez-vous à ce signal sur une échelle de 1 à 10 où 1 représente pas du tout urgent et 10
extrêmement urgent?" La première question vérifiait la reconnaissance du signal, la seconde
vérifiait la capacité du participant à correctement identifier le signal, puis la dernière question
évaluait le degré d’urgence perçu par le participant à l’écoute du signal. L’expérimentateur
insistait auprès de chaque participant sur le fait que le disque compact contenait des signaux
pouvant être perçus à l’intérieur et autour de grands édifices, tels que celui où le participant se
trouvait au moment de la cueillette de données.
NRC Contract: B4521
ix
Un total de 307 individus a participé à cette étude. Ceux-ci sont représentatifs de la
population canadienne en genre et en distribution d’âge. Les résultats révèlent une différence
significative entre la fréquence à laquelle les participants rapportaient reconnaître les différents
signaux sonores, Q(5,307)=288,15, p < .001. Le klaxon a été reconnu par 97% des
participants, soit le taux de reconnaissance le plus important, suivi par la reconnaissance de
l’alarme de recul par 91% des participants, le ronfleur industriel par 81%, le T-3 par 71%, la
cloche d’incendie par 58% et le Slow Whoop par 52% des participants. Toutefois, les
participants ayant rapporté reconnaître un signal n’ont pas nécessairement identifié celui-ci
correctement. En effet, le T-3 n’a été correctement identifié comme alarme d’incendie que dans
6% des cas. Le klaxon a été correctement identifié dans 98% des cas, l’alarme de recul 71%,
la cloche d’incendie 50%, le Slow Whoop 23%, et le ronfleur industriel dans 2% des cas. Une
différence significative entre l’identification des différents signaux a été démontrée; x2(5,
307)=555,52, p < 0.001. Par conséquent, malgré la reconnaissance courante du T-3 par les
participants, celui-ci était rarement correctement identifié comme une alarme d’incendie ou un
signal d’évacuation. En effet, le T-3 était plutôt associé à un signal domestique, tel que le signal
d’un téléphone occupé ou le signal d’un réveil matin.
Lorsqu’on leur demandait d’évaluer l’urgence de chacun des signaux, les participants ont
jugé le T-3 comme le signal le moins urgent parmi l’ensemble des signaux. Le T-3 a reçu un
score moyen de 3,97 sur une échelle de 1 à 10 (10 étant extrêmement urgent). Le ronfleur
industriel, le klaxon et l’alarme de recul ont obtenus des scores d’urgence entre 4,91 et 5,60,
tandis que le Slow Whoop a reçu un score de 6,01 et la cloche d’incendie, 7,17. Il importe de
mentionner qu’en général, si le participant identifiait un signal comme une alarme d’incendie, le
degré d’urgence coté pour celui-ci augmentait significativement. Ce résultat démontre
clairement le lien entre la signification attribuée à un signal et la perception d’urgence que celuici provoque.
L’ensemble de ces résultats suggère qu’une sensibilisation du public est essentielle pour
augmenter le taux d’identification correcte du T-3.
De plus, il semble utopique de croire que la perception d’un signal sonore par les
occupants sera immédiatement suivie par une évacuation de l’édifice, et ce, même si le signal
sonore est reconnu et correctement identifié comme étant une alarme d’incendie. Des
informations supplémentaires transmises par un message provenant du système de
NRC Contract: B4521
x
communication phonique ou par un employé en uniforme seront toujours nécessaires, afin de
confirmer aux occupants le besoin d’une évacuation ou d’une relocalisation dans un autre
secteur de l’édifice. Enfin, il est suggéré que le T-3 soit uniformément utilisé comme le signal
d’urgence incendie dans tous les édifices, peu importe si la stratégie en cas d’incendie en est
une d’évacuation ou non de l’édifice. De cette manière le nombre et la diversité des signaux
d’urgence seront réduits et, par le fait même, la reconnaissance sera ameliorée. Le T-3 peut
devenir le signal d’urgence standard reconnu par le public si les gens sont eduqués en ce sens
et s’ils sont exposés régulièrement au signal. Percevoir un signal d’avertissement est en soi
une information essentielle qu’il est important de transmettre au public, avant que ceux-ci soient
informés sur le comportement à adopter en cas d’incendie.
NRC Contract: B4521
1
Fire Alarm Signal Recognition
G. Proulx, C. Laroche, F. Jaspers-Fayer, R. Lavallée
1.0 INTRODUCTION
The 1995 National Building Code of Canada requires that any building equipped with a
fire alarm system should sound the Temporal-Three (T-3) pattern as defined by ISO 8201
“Acoustics – Audible Emergency Evacuation Signal” (1987). This sound pattern has also been
required for fire alarm systems and smoke alarms in buildings by NFPA 72 since July 1996.
Before these requirements came into effect, the fire alarm signal in buildings could sound a
large variety of continuous or temporal signals delivered through various devices such as bells,
horns, chimes or electronic apparatuses. This variety resulted in occupants experiencing
difficulty in recognizing the sound of a fire alarm signal among other signals. Interviews by Tong
and Canter (1985) showed that over 45% of a small sample of building occupants were unable
to distinguish fire alarms from other types of alarms. This identification problem can partially
explain why occupants tend to ignore and disregard fire alarms as genuine emergency warnings
(Proulx, 1994; Edworthy, 1994). Another reason to discredit fire alarm signals is the large
number of nuisance alarms investigated by Karter (1998) and the problem of audibility studied in
previous projects (Laroche et al. 1991; Proulx, Laroche & Latour, 1995).
The need to devise a unique fire alarm signal that could be used and recognised
universally was acknowledged many years ago. Since the 1970s, numerous discussions to
develop a standard signal have taken place (Mande, 1975; CHABA, 1975). In the end, experts
finally agreed not to limit the fire alarm signal to any one sound but instead to support
recognition of the signal through the use of a specific sound pattern. The T-3 pattern, described
in ISO 8201, is expected to become the universal standard evacuation signal. It is intended to
be used around the world and to unequivocally mean, “evacuate the building immediately”.
Major fire alarm system distributors in Canada and the United States have been
systematically installing the T-3, as the fire alarm signal, in all new and refurbished public,
institutional, and office buildings for the past 3 years. According to the National Fire Protection
Association’s Director of Public Education, Judy Comoletti, no formal public education has taken
place to inform building users of this new signal, its meaning or the response expected from
occupants upon hearing this signal. In North America, discussions are ongoing regarding the
NRC Contract: B4521
2
necessity to develop a public education campaign on the subject of this new evacuation signal.
It is also discussed, whether an automatic recorded message should follow the signal to prompt
the public to evacuate. Some people are suggesting that the temporal content of this alarm
might be sufficient to trigger an evacuation response since urgency can be indicated by the
acoustical structure of an alarm signal (Edworthy et al. 1995; Haas & Casali, 1995). As a first
step, we need to ascertain if the public already recognizes this sound as the evacuation signal
and if people associate some level of urgency to this signal that would prompt them to move
during an emergency.
2.0 REVIEW OF LITERATURE
Warning signals are commonly used to convey information, both urgent and
inconsequential. The relationship existing between warning signals and their significance has
been of interest for many researchers. Over the past decade, the works of Edworthy and
collaborators have led to a better understanding of how human beings interpret audible warning
signals presented in their daily environments (Edworthy, Loxley & Dennis, 1991; Edworthy &
Adams, 1996). Design criteria for audible warning signals have been proposed for the purpose
of improving their appropriateness and suitability which will ultimately govern whether a set of
warnings are actually effective (Edworthy & Stanton, 1995; Edworthy & Adams, 1996). The
aforementioned authors believe it is essential to consider two vital elements. First, when an
audible signal is perceived, it must be recognized as a warning signal. Second, the signal’s
significance, and ideally, the response expected, must also be understood.
In order to ensure the recognition of audible warning signals, many researchers have
proposed design criteria to standardize their intensity, spectral content, length, temporal aspects
etc. (Edworthy, Loxley and Dennis, 1991, Patterson, 1982; Tran Quoc et Hétu, 1996;
Momtahan, 1990). An international standard (ISO 7731, 1986) is currently available for audible
warning signals in the workplace. The international standard for evacuation signal (ISO 8201,
1987), commonly referred to as the T-3, also addresses some of the acoustic parameters
necessary to ensure the recognition of this type of signal. It does not, however, specify the
signal’s spectral content. It exclusively addresses the signal’s length, sound pressure level and
temporal aspects.
The ISO 8201 standard stems from the works of the National Academy of Sciences,
Committee on Hearing, Bioacoustics, and Biomechanics (CHABA, 1975). The members were
NRC Contract: B4521
3
set the objective of suggesting an audible evacuation signal that would be, ”specific and simple
so that it would be universally recognized and easily distinguished from other alarm signals”.
Two specific considerations guided their choice of audible signal. First, the signal had to be
evident and easily perceived by the occupants of a building. In fact, the signal had to be
detectable in any ambient noise. The second consideration was that the distinctive
characteristics of the signal had to make the signal clearly different from a variety of other alarm
signals. It was important to limit the amount of confusion occupants felt when the alarm
sounded. The evacuation alarm must be instantly recognizable.
Acknowledging that the alarm would be activated in a variety of environments, where the
level and spectral content of the ambient noise could vary tremendously, the members of the
committee opted for a standard that addressed the temporal aspects of the signal, as opposed
to one that addressed its spectral content. According to the members, this approach has the
following advantages. First, the signal could be designed for any environment by matching the
signal to the existing background noise to optimize its perception. Second, having specific
temporal characteristics would avoid any confusion of the signal with other audible warning
signals, particularly those used outside. Third, the distinct temporal pattern allowed for easy
adaptation of the signal to sight and touch alarms. Finally, the problems surrounding the
adaptation of existing evacuation alarm systems to this temporal signal are minimized because
it is much easier to install a temporal switch into the current circuitry then it is to replace the
entire system with new signals. Consequently, the time and money involved in updating
existing systems with the temporal signal would be limited.
The temporal characteristics proposed by the CHABA committee differ somewhat from
that of the ISO 8201 standard but they are closely related. Nonetheless, it was the committee’s
belief that the audible signal would undoubtedly be recognized by its temporal characteristics. It
is questionable, however, whether the temporal characteristics proposed are sufficient enough
to exclusively guarantee the association of the alarm to a specific meaning and adequate
response, assuming a well-adjusted intensity level.
In general, the concepts of perceived hazard and auditory affordance are referred to by
the same studies that address the significance and the adequate response to audible signals
(Edworthy & Stanton, 1995; Stanton & Edworthy, 1998; Hellier, Wright & Edworthy, 2000). It is
evident that a warning signal should not only communicate the presence of a danger, but also
NRC Contract: B4521
4
the level of hazard related to that danger. When warning signals vary in their level of urgency
and are adequately associated with the level of danger involved in a situation, they gain a
behavioural advantage (Haas & Casali, 1995; Finley & Cohen, 1991). This process, referred to
as “hazard matching”, is the object of the recommendations that are made for implementation of
warning signals (Edworthy & Adams, 1996; Hellier, Edworthy & Dennis, 1995; Edworthy, Loxley
& Dennis, 1991; Momtahan & Tansley, 1989). It is known, for example, that the fundamental
frequency, harmonic series, amplitude envelope shape, delayed harmonics, speed, rhythm,
pitch range, and melodic structure all have clear and consistent effects on perceived urgency. A
signal’s perceived urgency is therefore related to several acoustic parameters.
Nonetheless, an individual’s interpretation of a warning signal cannot be limited to
acoustic parameters. There is also the associative meaning dimension that comes into play.
Gibson (1979) claimed that people do not solely attend to the purely physical aspects of objects
in the world; rather, we seek to interpret their meaning. This very theory is what brought
Stanton & Edworthy (1998) to propose the theory of auditory affordances. They argued that an
auditory warning is perceived in terms of its potential for action. They upheld their theory by
means of 4 main propositions: 1) we are surrounded by sounds; 2) we are introduced to sounds;
3) we learn sounds through seeing other people respond to them; and 4) sounds have a definite
function.
With fire alarms as an example, Edworthy and Stanton (1995) tried to explain this
concept. The audible signal associated with a fire alarm can be anything from a bell to a siren,
neither extremes having any direct relationship with the sound produced by an actual fire.
However, for most people, these signals have been the object of a learned association. Would
it be better to use a sound that produces an image of burning flames? It would seem that a
signal’s suitability is a function of its association to the fire and its ability to incite individuals to
move away from the source of danger. Consequently, it appears that certain situations would
be more appropriately signalled by an audible warning signal that has a significant acoustic
relation to the situation, a representational sound, while other situations will be better
represented by sound that has a learned association.
In their 1998 study in an intensive treatment unit (ITU), Stanton & Edworthy proposed
two hypotheses: A) the function of representative, environmental sounds will be identified more
readily than abstract, non-representative sounds, and; B) the interpretation of the function of a
NRC Contract: B4521
5
sound will be dependant upon the experience of the individual. The study was comprised of two
participant groups, ITU staff and non-ITU staff. The signals used throughout the study were
either signals typically used in the ITU (abstract sounds) or new signals designed to be
acoustically related to the meaning of each sound signal; in other words sounds that were
representational. For example, representational signals included CD recordings of a heart beat
for an EEG monitor, a nursery chime for an infant warmer, and bubbles for the syringe pump.
Preliminary results revealed difficulties in recognition of audible signals as well as poor
performances with abstract sounds, particularly during initial performances. To learn the
significance of abstract sounds requires considerable effort, whereas representational sounds
are more intuitive. Moreover, the authors noted an interaction between the type of signal and
the experience. In fact, hypothesis A) arguing that individuals would better identify
representative sounds, was verified for the group of participants that were unfamiliar with the
typical ITU audible signals, but was not retained for the individuals used to ITU warning signals.
The ITU staff showed a significantly superior success rate for the recognition of typical ITU
sound signals, whereas non-ITU staff were significantly better in the identification of the new
representative sound signals. The authors consequently proposed that the targeted population
be involved in the design of warning signals. Moreover, the theory of auditory affordances
suggests that individuals comprehend the sounds that surround them in terms of potential for
action. This implies that it is people’s physical reaction to the alarms that must be considered,
rather than what the alarms’ equipment is monitoring. According to the authors, this outcome is
subtle, but substantial in the design of new audible warning signals. The referents are the
actions required by people. In conclusion, the authors insisted that the development of new
audible signals would most likely rely on a combination of traditional and representational
warning signals.
Thus, based on the literature, it can be predicted that individuals having to recognize a
new audible signal, such as the T-3, will perform poorly according to the hypothesis that these
individuals will not have learned the significance associated with the signal prior to their
exposure. This hypothesis is further substantiated by the acoustic characteristics advocated by
the ISO 8210 standard, which convey little information related to a fire or an evacuation.
3.0 RESEARCH OBJECTIVES
This research project endeavours to answer three fundamental questions about the T-3
signal. These questions are associated with recollection, identification and urgency.
NRC Contract: B4521
6
The first objective of this project is to assess whether people can recall the T-3 signal. It is
essential to know if people believe that they have previously heard the T-3 to ascertain their
previous exposure to the signal.
The second objective is to evaluate if people can correctly identify the meaning of the
signal. It is one thing to say that you have heard a signal before, and something entirely
different to correctly identify the meaning of that signal.
The third objective is to measure the degree of urgency that people associate with the T-3.
Building occupants may not be able to identify the T-3 or they may not have even been exposed
to the signal previously. It is expected, however, that if they attach a high level of urgency to the
signal they will feel compelled to act on it if they unexpectedly hear it in a public building. The
perceived urgency is particularly important to assess whether the signal will prompt action in a
fire situation where no other occupants or fire cues are there to indicate a fire situation.
Two hypotheses have been developed to study the effectiveness of the T-3. The first
hypothesis is that the recollection of the T-3 will be lower than other warning signals. The
second hypothesis is that the identification of the T-3 would be lower than that of other signals.
This is because there are a limited number of buildings equipped with the T-3 and the potential
for the public to have been exposed to the signal is limited. Finally, no specific hypothesis was
formed regarding the perceived urgency of the T-3 signal.
4.0 METHODOLOGY
To investigate the recollection, identification, and perceived urgency of the T-3 an
experimental study was conducted. Early in the study it was determined that it would be
impractical to activate the T-3 in a number of public buildings to observe occupant’s response.
Instead, it was considered more appropriate to approach a sample of individuals and conduct
interviews in the field to obtain the data that would answer the study objectives.
4.1 Participants
A sample of participants, representative of the Canadian population in terms of gender
and age distribution, was approached in the field. It is a recognized methodology when
evaluating auditory warning signals to focus on the real users, measuring their capacity to make
NRC Contract: B4521
7
an association between sound and meaning (Edworthy & Stanton, 1995). The general public
constitutes the real users of the T-3 and it is essential that the research and subsequent
development of the T-3 be based on actual building occupants.
People were approached in public buildings such as shopping centers, libraries, and
airports. All of the buildings were places where the T-3 signal could be heard if the fire alarm
system had been installed or upgraded in the last few years.
4.2 Signals Tested
Although the objective of this project was to study the T-3, it was decided to test it among
5 other warning signals. It would have been an acceptable methodology to go around and test
only theT-3 without any other signal. This approach would however necessitate, meeting with a
large group of people to test the T-3 then to use another signal and test it with another group of
people to compare the recollection, identification and perceived urgency between the two
groups. This strategy was not used for fear of finding large discrepancies between groups
which could have been difficult to interpret if not enough was known about the participants
experience, background, prior exposure, etc. There was also the worry that exposing a person
to the T-3 “out of the blue” could be such a surprise to the participant that a large number of
wrong answers would be obtained which might not be representative of the “true” recollection,
identification and perceived urgency of the T-3.
It was decided to use a total of 6 signals which seems sufficient for the participants to
understand the context of the task while keeping the whole test under 5 minutes for each
participant. Experience in previous field studies in public buildings suggests that 5 minutes is
the maximum time you could expect participants to be prepared to answer questions. To
consider a case, it was necessary that the whole test be completed, therefore it was important
to ensure that participants would not give up before the end.
The signals selected were also used to validate the task. It was expected that the Car
Horn should obtain an overall good recollection and identification from the participants.
Consequently, if the findings would show a large number of participants not accurately
recollecting or identifying the Car Horn, that would indicate that the task was not well
understood by participants or the methodology not appropriate for the research objectives.
NRC Contract: B4521
8
The 6 signals selected were all warning signals. The signal selection was carefully made
to include signals that were expected to obtain a range of recollection and identification from
very good such as for the Car Horn or the Alarm bell and very poor performance such as for the
industrial Buzzer. It was expected that the T-3 would perform within that range. It was also
interesting to compare the performance of the T-3 with two other fire alarm signals: the Slow
Whoop which has many advocates in the United States and the alarm Bell which is used
extensively in institutional buildings in Canada.
Three CDs were prepared which presented six signals in different orders. The T-3 signal
was presented first, third and last among five other signals. The exact presentation order of the
signals on each CD can be seen in Table 1. To limit the number of CDs, and simplify the
analysis, it is considered acceptable in the warning literature to restrict the number of
presentation orders (Momtahan et al. 1993).
Table 1: Signal Presentation Order on Each CD
CD
Signals Order
1
T-3
Car Horn
Reverse
Slow Whoop
Buzzer
Bell
2
Slow Whoop
Car Horn
T-3
Bell
Reverse
Buzzer
3
Bell
Car Horn
Buzzer
Reverse
Slow Whoop
T-3
The signals were presented in different orders on each CD so that order effects could be
counterbalanced. Presenting one signal before another could affect the opinion a participant
would have had about following signals. For instance, having heard the Bell first, and identifying
it as a fire alarm, participants later interpretation of the T-3 signal may be coloured. By
changing the order of the signals so that the Bell is not always heard before the T-3 signal, this
colouring can be accounted for. This is why the T-3 was presented first, third and last, to
determine whether hearing the signal “out of the blue” as the first signal would provide
comparable recollection, identification and degree of urgency as listening to the T-3 after
hearing other signals.
An analysis of the acoustic characteristics of each of the signals has been conducted in
order to document the main parameters, such as the spectra and temporal patterns. All signals
NRC Contract: B4521
9
had an overall duration of 12 seconds. The Car Horn had a fundamental frequency of 350 Hz
with harmonics up to 2 kHz. It was sounded few times during the 12-second interval. The Slow
Whoop is characterized by a frequency sweep between 875 and 4000 Hz, and each sweep
lasted approximately 2 seconds. The Buzzer had a wide spectrum between 500 Hz and more
than 8 kHz. The Reverse alarm was an intermittent signal with a 400 msec ON period followed
by a 350 msec OFF period. The fundamental frequency was 1302 Hz with harmonics
significantly lower in level. The Bell is composed of three main frequencies (1244, 1866 and
2527 Hz). The T-3 is a 500 msec ON-500 msec OFF signal repeated three times followed by
an OFF period of 1.5 second. The fundamental frequency is 505 Hz with the odd harmonics
(3rd, 5th, 7th, etc.). The T-3 used in the study was recorded from Simplex 1996, 4100 Fire Alarm
Audio Demonstration CD. Appendix 1, at the end of this report, shows the spectrum analysis of
the 6 signals.
4.3 Materials
The signals were recorded so that they played at a maximum of 90 dBA when the CD
player was at maximum volume. This volume level is not considered harmful to the hearing of
the participant if the duration of the exposure is limited to 15 minutes or less and is not repeated
day after day over a long period of time (WHO, 1999). The interviewer adjusted the volume of
the CD player to a comfortable level for the environment they were in at the time of the study
before beginning the test (this volume was usually at 4 out of a maximum volume of 10). Each
participant, however, had the opportunity to adjust the volume during a 10-second period of
music, prior to listening to the test signals.
The signals were played on a CD Walkman (Sony Sports model D-SJ01) and both the
experimenter and participant had a set of headphones to listen to the signals (the experimenter
had Sony MDR-GO51 headphones). The participant listened to the signals through Sony Noise
Cancelling headphones (Sony MDR-NC20) so that the ambient noise of the environment did not
overly interfere with the quality of the test signals.
Three experimenters collected the data. All experimenters followed a rehearsed script so
that the participants experienced a similar situation with each experimenter. The experimenters
recorded the participants’ answers on a data sheet that the participant could view at the time of
the survey.
NRC Contract: B4521
10
The answers to the perceived urgency rating were recorded on a simple 1-10 scale,
although the psycho-acoustic literature lately suggests using a free-modulus magnitude
estimation scale or the cross modality matching (Hellier et al., 1995). It was decided that these
methods would not work well for the present study. These methods need to be explained very
carefully to participants before a study can begin, often taking long moments so that there is no
confusion, therefore the interviews would not have been under 5 minutes. Finally, the perceived
urgency rating was not the main objective of this study which was more concerned with
recognition. In our view, recognition of the T-3 could be better studied during a field study with a
large number of participants being involved for a short period of time. Consequently, the simple
1- 10 scale was used by participants to assess perceived urgency for the sake of simplicity
within the context of this field study.
4.4 Procedure
Initially, experimenters attempted to select participants as randomly as possible, however
under-represented groups were targeted near the end of the study so that a representative
sampling of the general Canadian population was possible. Participants were approached
when alone and inactive, resting or waiting for somebody on a bench or in the food court.
After an introduction, the experimenter asked the respondent if they were willing to
participate in a five-minute sound recognition study. Participants were told that they would listen
through headphones to six 12-second sounds that were being tested. It was explained that the
first sound, which was a music sample, was not part of the study, it was being played only to
check that the equipment was working and that the sound level was comfortable. The
participants were then told that after this music sample, the six test sounds would be played and
after each sound, the CD player would be stopped so that they could be asked three questions.
The questions were stated at this point so that the respondents would know exactly what they
would be asked.
The questions were “Have you heard this sound before?” “What do you think this sound
means?” and finally, “How urgent do you feel this sound is on a scale from 1 to 10? 1 being not
urgent at all and 10 extremely urgent.” The first question tested the recollection of the signal,
the second tested the ability to correctly identify the signal and the last question rated the
perceived urgency of the signal.
NRC Contract: B4521
11
It was heavily emphasized that the sounds heard on the CD were from in and around
large buildings, such as the one they were in at the time of the interview.
After the participant had listened and responded to each of the signals, they were asked
for their age and occupation. These characteristics were only obtained to explain individual
differences during analysis, not to discriminate against the participation of any person or groups
of people. When all of the data had been collected, the participants were carefully debriefed.
The experimenters explained that the study was on the recollection and identification of fire
alarm signals and that the T-3 signal was the new evacuation signal to be installed in public
buildings.
5.0 STUDY RESULTS
The raw data from this study was entered into SPSS 6.0 so that a statistical analysis could
be performed. Nominal and ordinal data were collected during the study and nonparametric
tests were used in the analysis. Nonparametric tests were selected because they are more
robust and better suited to deal with nominal and ordinal data. The results, therefore, were
achieved through the use of such tests as the Sign test, the Kruskal-Wallis one-way analysis of
variance, and the Wilcox matched pairs signed-ranks test. For the purpose of the analysis, an
alpha level of .05 was used as the critical value for determining a significant difference.
Therefore, if p was less then 0.05 (p < .05) the results were considered significant.
5.1 Respondent Profile
The sample contained 307 participants ranging in age from 15 to 85 years old
(mean = 38.45, SD = 17.22). The majority of the respondents were recruited from in and
around shopping centers and public buildings in the Ottawa area. The rest were recruited at the
Ottawa International Airport. The age distribution of the sample was similar to the age
distribution of Canada, as reported by Statistics Canada (2000), see Table 2 below.
NRC Contract: B4521
12
Table 2: Population Distribution of Canada and the Study Sample
Age Group
Canada
Study Sample
0-14
19%
0%
15-19
7%
15%
20-29
14%
23%
30-49
32%
34%
50-64
16%
20%
65- 85
11%
8%
86 and over
1%
0%
As Figure 1 shows, there was a higher number of young women sampled. It is speculated
that this occurred because young woman may frequent the malls, where many of the interviews
took place, more often than other age or gender groups.
60
Frequency
50
40
Females
30
Males
20
10
0
15-19
20-29
30-49
50-64
65-85
Age
Figure 1: Gender and Age Distribution of the Study Sample
The total sample contained 164 women (53%) and 143 men (47%). According to Statistics
Canada (2000), the Canadian population is also slightly biased towards women (50.5%
Females and 49.5% Males). The sample was composed of 173 people who worked in large
buildings (56%), 69 students (22%), 45 people who were unemployed at that time (15%), and
20 people who worked outside or in small buildings (7%).
NRC Contract: B4521
13
5.2 Merging Data from the Three CDs
The three CDs that were used to gather data were designed to present the signals in
three different orders so that if an order effect occurred, it could be taken into account during the
analysis. Consequently, the T-3 was presented as the first signal to one third of the sample, as
the third signal to another third of the sample, and as the last signal to the last third of the
sample. In total, 101 participants listened to CD1 and 103 listened to CD2 and CD3 for a grand
total of 307 participants in the study.
There were significant differences between the three CDs for the T-3 in recollection
2
x (2,307)=9.87, p < .01, and urgency x2(2,307)=11.14, p < 0.01. Both of these statistics were
computed using the Kruskal-Wallis one-way analysis of variance. As presented in Table 3, CD3
where the T-3 was presented last was the one that obtained the higher recollection and
urgency. Interestingly, there were no significant differences found in recollection or urgency for
the alarm Bell, which was placed in the first, fourth and last positions, x2 (2, 307)= 1.9242,
p<.05.
Table 3: Results to the T-3 Signal for Each CD
CD
Presentation
location of the T-3
Recollection
% of “Yes”
Identification
% of correct answer
Perceived
Urgency rating
1
1st signal on CD
59.4%
3.0%
3.47
2
3rd signal on CD
73.1%
5.8%
3.80
3
6th signal on CD
78.6%
7.8%
4.64
An order effect was discounted because biasing did not occur with the Bell. Also,
basically only one experimenter used each CD during the experiment, and so the difference in
response could be as dependent on the experimenter as the signal order presentation. In fact,
it is believed that the difference in perceived urgency of the T-3 occurred because of a
procedural difference between experimenters. For example, when a subject hesitated between
two urgency ratings, the experimenters prompted the participant differently for a definite
response. One experimenter had a tendency to systematically prompt for a higher rating.
When the participant hesitated between “a 3 or a 4”, the experimenter would say, “so that would
be a 4?” while the other two experimenters did not prompt the participant. This is believed to be
why a significant difference in urgency for the T-3 occurred between experimenters when a
NRC Contract: B4521
14
Kruskal-Wallis one-way analysis of variance was performed, x2(2,307)=7.05, p < .05 . Despite
the biasing, the shape of the distribution remained the same for each experimenter. For this
reason, the procedural difference between the experimenters was ignored for the purpose of
analysis.
All of the participants who were interviewed were used in the analysis: no cases were
discarded. Of the ten respondents who failed to recall the Car Horn, the original manipulation
check, six were over 50 years old. Apart from failing to recall the Car Horn, their other
responses were not overly different from the responses of the average participant, so their
cases were kept in the final analysis.
5.3 Signal Recollection
The first question participants were asked after listening to each signal was “Have you
heard this signal before?” The answer to this question on recognition was either “yes” or “no”.
Table 4 presents the number of positive answers to each of the signals.
Table 4: Number of Occupants who Confirmed Having Heard a Signal Before
Signal
Participants (N=307)
Percentage
Car Horn
297
97%
Reverse alarm
280
91%
Buzzer
249
81%
T-3
219
71%
Bell
177
58%
Slow Whoop
159
52%
There was an overall significant difference found between how often participants recalled
each of the signals, Q(5, 307)=28.15, p< .001. This statistic was computed using the
Cochran Q test which is used when looking for significant differences between three or more
matched sets of frequencies or proportions. The Car Horn was recalled the most often, by 297
out of 307 participants (97%), followed by the Reverse alarm, which was recalled by 280 people
(91%). The Buzzer was recalled by 249 people (81%), the T-3 by 219 people (71%), the Bell by
177 people (58%). Finally the Slow Whoop was recalled by 159 people (52%). Participants
were significantly more likely to say that they had heard the T-3 then the Slow Whoop, x2 (1,
NRC Contract: B4521
15
307)= 27.80, p < .001 and significantly more likely to say that they had heard the T-3 then the
Bell, x2(1, 307)= 13.91, p< .001. Participants who said they had heard a signal before, however,
could not necessarily identify the signal correctly. Figure 2 below shows the percentage of ‘yes’
answers compared to ‘no’ answers.
100
90
80
70
60
50
40
30
20
10
0
yes
no
T3
Car
Ho rn
Reverse
Slo w
Who o p
B uzzer
B ell
S igna ls
Figure 2: Recall Percentage of each Signal
5.4 Signal Identification
The second question that the participants were asked was if they knew what the signal
meant. In many cases the subject, although asked, “what does this sound mean”, gave
responses that would have been expected had they been asked, “what is this sound”. For the
purpose of analysis the answers were placed into three simple categories, “right”, “wrong” and
“I don’t know”. The definition of “right” used for the T-3, Slow Whoop and Bell, all of which were
fire alarms, was that the answer had to be related to fire alarms or evacuation to be considered
“right”. Answers given in response to the industrial Buzzer were defined as “right” if they were
related to a factory or mechanical setting, although it is possible that participants could have
heard this signal in a different context. The Reverse alarm was operationally defined as correct
if the participant identified it as a Reverse alarm. The answers “snow removal” and “garbage
truck” were also accepted because in such situations the signal could be heard as the vehicle
reverses. The Car Horn was considered correct when participants identified it as a horn. The
answers accepted as correct for this signal, included car horn, car alarm, truck horn and boat
horn.
NRC Contract: B4521
16
Table 5 presents the number of occupants who correctly identified the different signals.
The T-3 was identified correctly only 6% of the time (17 out of 307 said it was a fire alarm). The
Car Horn was identified correctly 98% of the time (302 people identified it correctly), the
Reverse alarm was correctly identified 71% of the time (by 218 people), the Slow Whoop 23%
of the time (by 71 people), the Buzzer 2% of the time (6 people) and the Bell 50% of the time
(by 155 people).
Table 5: Number of Occupants who Correctly Identified Each Signal
Signal
Participants (N=307)
Percentage
Car Horn
302
98%
Reverse alarm
218
71%
Bell
155
50%
Slow Whoop
71
23%
T-3
17
6%
Buzzer
6
2%
The seventeen people who correctly identified the T-3 as a fire or evacuation alarm were
aged 21 through 84 years old (M = 41.20, SD = 16.52), and were equally likely to have been
male or female (53% female and 47% male). Among these seventeen participants there was,
for example, a mechanic, a student, a janitor, an accountant, a person who was looking for work
at the time, and a Ph.D. fellow.
There is a very significant difference between how often each of the signals was correctly
identified, x2(5, 307)=555.52, p < 0.001. The T-3 was identified correctly less often than any
other fire alarm signal (Bell and Slow Whoop). Figure 3 gives a visual representation of these
findings.
NRC Contract: B4521
17
100
90
80
Percentage
70
60
Right
50
Wrong
40
Don't know
30
20
10
0
T3
Car Horn Reverse
Slow
Whoop
Buzzer
Bell
Signal
Figure 3: Percentage of Signals Identified
It is also interesting to notice that about the same proportion of participants answered that
they “didn’t know” the three fire alarm signals (T-3, Slow Whoop and Bell). For the T-3, 69 of
307 or 22% of the participants said they could not identify the sound, for the Slow Whoop, 64 of
307 participants or 21%, and for the Bell, 70 of 307 people or 23% said they didn’t know the
signal. Eight participants in the sample (3%) could not identify any of the fire alarm signals.
5.5 Signal Recollection and Identification
It is informative to know how often participants could correctly identify a signal after stating
that they “had heard the signal before”. Although the T-3 was identified correctly by 6% of the
population, only 4% of the participants said that they had heard the T-3 before and then
identified it as a fire alarm. Participants who said that they recalled the T-3 but then went on to
state that they could not identify the signal accounted for 10% of the population. An astounding
57% of the participants felt that they recalled the T-3 but then erroneously decided that the
signal was something other than a fire alarm. Five participants (2% of the total population)
correctly identified the T-3 as a fire alarm but did not state that they had heard the T-3 first. It is
assumed that they “guessed” the right answer. Three of them were male, two were female and
they ranged in age from 16 to 59 years old (M= 33.20, SD= 16.08). These results are displayed
in Figure 4.
NRC Contract: B4521
18
4% Yes, recall it, right answ er
57% Yes, recall it, w rong answ er
4%
12%
2% No, do not recall it, right answ er
15% No, do not recall it, w rong answ er
10%
10% Yes recall, don't know
12% No, do not recall it, don't know
15%
57%
2%
Figure 4: Recollection and Identification of the Temporal-Three
As can be seen in Figure 5, the Car Horn was recalled and identified correctly 96% of the
time whereas the T-3 was recalled and identified correctly only 4% of the time. The Reverse
alarm was recalled and identified correctly 69% of the time, the Slow Whoop 14%, the Buzzer
1%, and the Bell 38% of the time.
100
90
80
Percent
70
60
50
40
30
20
10
0
ll
Be
op
ho
W
er
zz
Bu
ow
Sl
e
n
or
s
er
ev
R
h
ar
C
T3
Signal
Figure 5: Percent of Each Signal Recalled and Identified Correctly
NRC Contract: B4521
19
When the T-3 was identified as something other than a fire alarm, 61% of the time it was
because the participants imagined the signal in the context of a domestic environment instead
of a public building. This occurred despite the conscious attempt by all experimenters to
carefully state that all of the signals heard on the CD were from in and around large buildings.
As presented in Table 6, most often the T-3 was identified as a signal related to the telephone
with answers that included busy signal, telephone, phone that is on hold and answering
machine.
Table 6: Wrong Identification of the Temporal-Three
Identification
Participants (N=174)
Percentage
Busy phone signal
55
32%
Phone beep
20
12%
PA pre-announcement
18
10%
Reverse alarm
17
10%
Time signal
14
8%
Alarm clock
13
7%
TV message coming
10
6%
Hospital alarm
6
3%
Bank machine
4
2%
Other identification
17
11%
The Car Horn was perceived correctly to be a horn for a vehicle by almost all of the
participants (98%). Five people gave odd answers that included, reverse and pool horn. The
Reverse alarm was identified correctly by the majority of respondents (71%). Wrong answers
were generally associated with a domestic environment and included busy phone signal, time
signal and alarm clock (39%). The Slow Whoop was correctly identified as a fire alarm by 23%
of the sample. Among the wrong answers were emergency vehicle or siren (62%) bomb
warning (14%) and security alarms (6%). The industrial Buzzer was incorrectly identified as part
of a domestic environment 50% of the time. Wrong answers in this case included alarm clock,
stove timer, and door buzzer. The Bell was perceived as a fire alarm by half of the participants.
Wrong answers included security alarm (6%), radios and alarm clock (6%), and train sounds
(3%).
NRC Contract: B4521
20
5.6 Signal Perceived Urgency
Each participant was asked to rate, on an ordinal scale from 1 to 10, the degree of
urgency they felt for each signal, one was not at all urgent and 10 meant they felt the signal was
extremely urgent. The differences in perceived urgency between all of the signals, were
significant, x2(5,307)= 311.87, p< .001.
Table 7: Difference in Perceived Urgency between T-3 and Other Signals
Signal
Perceived Standard
Urgency Deviation
Probability
T-3
3.97
2.42
Buzzer
4.91
2.74
p< .001
Car Horn
4.93
2.46
p< .001
Reverse
5.60
2.78
p< .001
Slow Whoop
6.01
2.50
p< .001
Bell
7.17
2.74
p< .001
As can be seen in Table 7, the T-3 was considered the least urgent of all the signals.
When the T-3 was compared to each of the signals separately it was significantly less urgent
than each of them. The T-3 received an overall rating of 3.97, which could be assessed as a
low urgency rating on a scale from 1 to 10. The Buzzer, Car Horn and Reverse signal all
received urgency ratings between 4.91 and 5.60, which could be judged as medium urgency
ratings. Finally the Slow Whoop and Bell received ratings of receptively 6.01 and 7.17, which
could both be considered high urgency ratings.
On each of the three CDs one of the alarm signals was played first. Although in the initial
analysis an order effect was discounted, it was considered prudent to analyze the average
perceived urgency of each of these signals when they were played first. The T-3, when no
other signal had been heard beforehand, had a mean perceived urgency rating of 3.58, the
Slow Whoop had an urgency rating of 6.40 and the Bell had an urgency rating of 7.13. In other
words they remained unchanged, which further confirmed the lack of an order effect caused by
signal presentation.
NRC Contract: B4521
21
5.7 Signal Identification and Perceived Urgency
It is interesting to consider the urgency rating of each signal while keeping in mind the
identity of the signal given by participants. When any of the signals were identified as fire
alarms, independent of whether or not they actually were, the urgency generally increased as
can be seen in Table 8. Only the Car Horn was never identified as a fire alarm so this signal is
not included in Table 8.
Table 8: Identification of any Signal as a Fire Alarm and Perceived Urgency
Signal
# of Ss who
Urgency # of SS who Urgency as a
did not
when not a identified
fire alarm
identify the fire alarm
signal as a
signal as a
fire alarm
fire alarm
Significance
T-3
290
3.83
17
6.29
p<0.001
Slow Whoop
236
6.76
71
7.42
p<0.05
Buzzer
265
4.49
42
7.55
p<0.001
Bell
152
5.89
155
8.43
p<0.001
Reverse
298
5.59
9
5.67
N.S. (p=.88)
The average urgency rating changed from 3.83 to 6.29 when participants identified the T3 as a fire alarm signal. The same tendency was observed for the Slow Whoop, the Buzzer and
the Bell which obtained significantly higher urgency ratings when identified as fire alarms.
There was no significant difference for the Revere alarm although results are in the same
direction.
NRC Contract: B4521
22
6. 0 CONCLUSIONS
The objectives of this research project were to evaluate if the T-3 signal was an alarm
signal that the general public could recall and identify. Another objective of this study was to
assess the level of urgency people would attach to the T-3 signal.
Among the signals tested, the Car Horn was recalled the most frequently, followed by the
Reverse signal, the Buzzer, the T-3, the Bell, and finally the Slow Whoop. This order of
recollection, however, proved inconsequential. Often participants recalled the T-3 as something
other than a fire alarm: usually the T-3 was identified as an everyday household signal, such as
a busy telephone signal.
It was found that only 6% of the population surveyed was able to identify the T-3 as a fire
alarm signal. When compared to the Slow Whoop, which was identified as a fire alarm by 23%
of the population, and the Bell, which was identified correctly by 50% of the population, it
becomes obvious that the T-3 is not a well known fire alarm.
It is possible that the procedure used led to the problem of incorrect identification.
Although people were briefed that the signals they were going to hear through the headphones
were from in and around buildings like the one they were in, a considerable number of
participants identified signals as sounds that could be found in domestic or small scale
environments. Since participants were using headphones they may have had difficulty
transferring the unfamiliar signals to the actual surroundings. However, this was not the case
with more familiar signals such as the Car Horn and the Reverse alarm which obtained correct
identification (98% and 71%) when heard through headphones.
Regarding perceived urgency, the T-3 was considered to be the least urgent signal among
all six of the signals tested. The T-3 scored a mean of 3.97 on a scale of 1 to 10 with 10
indicating extreme urgency. In contrast, the Bell scored the highest urgency rating with a mean
of 7.17, on the same scale.
Generally when a signal was identified as a fire alarm the perceived urgency was
significantly increased. This finding applies also to the T-3. It is suggested, therefore, that to
increase the perceived urgency of the T-3, the public should be educated to readily identify the
sound pattern of the T-3 as a fire alarm signal. In light of the temporal characteristics of the T-3
NRC Contract: B4521
23
signal, it is not surprising to obtain a weak degree of perceived urgency associated to the signal.
In fact, the T-3 ’s rhythm does not seem of sufficient speed to convey an adequate degree of
urgency for an evacuation. It is unlikely, however that the T-3 will be modified since it is an ISO
standard and is now a requirement in building codes in Canada and the USA. Consequently,
the acquisition of the T-3’s significance and expected response will depend highly on educating
the public to ensure that the T-3 improves life safety.
7.0 RECOMMENDATIONS
There are a number of approaches that can be considered to educating the public about
the recollection, identification and behaviour that they are expected to perform upon hearing the
T-3. One possibility is to broadcast an advertisement campaign on television or radio. Although
this method would reach the greatest number of people, and could have effective content, such
advertisements are extremely costly. Another strategy would be to distribute leaflets and
brochures. This is a less costly alternative, but it might be difficult to graphically display the
sound of the T-3 signal and ensure that the general public will make the connection between a
graphic display and the auditory signal. An interesting suggestion, made by a member of the
fire alarm industry, would be to display the T-3 signal on the Internet which seems to be a new
means by which an increasing number of people learn new things. Another good idea could be
to make the new evacuation signal the theme for the National Fire Safety week, with the sound
being played by the media who cover the events around Canada and the United States.
Another alternative may be to educate the public while they are visiting buildings by
performing evacuation drills. After the fire alarm is activated, a recorded message would be
played stating: “This is the evacuation signal, please leave the building immediately”. One
drawback of this approach is that this would only be possible in buildings equipped with voice
communication capabilities. Further, these kinds of recorded messages are not very informative
and may not prompt overall evacuation movement. Moreover, this approach is unlikely to
educate the public rapidly because the potential of exposing the general public to such a drill or
alarm activation is limited.
A more promising approach would be to require immediate alarm signal upgrading in all
elementary and high schools. The majority of the population learns fire safety during schooling
years. The lessons learned at this time are applied for the rest of life. Children would rapidly
NRC Contract: B4521
24
learn the sound pattern of the T-3 and would be taught the associated meaning and behaviour,
which they would be able to transfer to other buildings that they would be visiting in the future.
In an effort to educate the public so that they can identify the T-3, the USA has decided to
make all single station smoke alarms, also commonly referred to as, “home smoke detectors”
sound the T-3. Canada has decided to go against this practice because the T-3 signal is
specifically the evacuation signal that is supposed to unequivocally mean, “evacuate the
building immediately.” This is not the common behaviour observed when the smoke alarm
sounds in a person’s apartment or house. Within a dwelling, upon hearing the smoke alarm,
unless substantial fire cues are present, occupants usually investigate the situation in order to
decide on the best course of action. Since most smoke alarm activations in dwellings are due to
non-fire-related problems or minor incidents that can be efficiently controlled by the occupant,
the best response to the smoke alarm is not to “immediately evacuate”. Canadians have a
strong logical rational for their decision requiring that smoke alarms should not sound the T-3
signal. However, it has been demonstrated that it is extremely difficult to attach a behavioural
response to a warning signal (Stanton & Edworthy, 1998) and therefore it is unlikely that people
will evacuate upon hearing the T-3 from a smoke alarm in American homes. In fact, the
American approach has the advantage of exposing a large number of occupants to the T-3
signal since smoke alarms have a high probability of activation. Consequently, the American
public is more likely to be able to recall and identify the T-3, which is not a guarantee that
people will evacuate upon hearing the signal, but does mean that they will most likely correctly
identify the signal. It would be interesting to conduct the signal recollection study in the USA to
assess if the recollection and identification of the T-3 is superior to the results of the Canadian
study, since smoke alarms sold for the last 3-4 years in the USA all sound the T-3.
As already stated, the T-3 was originally intended in the ISO standard 8201 (1987) to be
used as the “International Emergency Evacuation Signal”. Despite this, it is often installed in
buildings as the fire alarm signal. According to the ISO definition, some public buildings should
not sound the T-3 if their Fire Safety Plan calls for occupants to stay-in-place or relocate in case
of a fire. It also means that the T-3, which is currently supposed to be the universal evacuation
signal, would be sounded for non-fire incidents such as for a bomb threat, chemical spill, gas
leak, nuclear radiation, etc. which require an evacuation of the building. From all the research
conducted in the field of human behaviour during building evacuation, it appears unrealistic to
expect that a signal such as the T- 3, by itself, could ever instigate a complete building
NRC Contract: B4521
25
evacuation of all occupants. The results of this study certainly dispute the suggestion that the
temporal content of the T-3 may be sufficiently urgent enough to trigger an evacuation response
as suggested by previous scientists (Edworthy et al. 1995; Haas & Casali, 1995). This
suggestion was also rejected in the 70’s when the T-3 was developed. A signal by itself cannot
convey a sufficient enough message of urgency to people that will let them know readily that
they are expected to evacuate a building.
The field of psycho-acoustic has, however, evolved tremendously. It is now known that
signals are built with a number of characteristics and these characteristics can greatly influence
the perception, interpretation and response of a person to that signal. For example, a high pitch
or high repetition rate will convey a higher degree of urgency then a signal with low pitch or a
low repetition rate. The representational characteristics of a signal are also an important factor.
Many occupants like the evacuation alarm that emits 120 pulses a minutes because it matches
very well the pace of a moving crowd down a stairwell. Even when the occupants can recall
and correctly identify a warning signal they do not necessarily comply with the expected
behaviour. Thus, it is not because people will recall the T-3 that they will necessarily evacuate a
building. They will still need more information to confirm the situation and obtain precision on
the expected evacuation movement.
Nevertheless, it is believed that the implementation of the T-3 is a step in the right
direction because building occupants need a standard fire alarm signal that they can
immediately identify when the fire alarm is activated. In fact, it is suggested that the T-3 should
be used regardless of the evacuation, or non-evacuation, strategy used in all public buildings in
order to limit the variety of signals that can be found in buildings. It is our recommendation that
all fire alarms and smoke alarms should sound the T-3, to improve people’s recollection and
identification of the signal. The response to the signal will still need to be learned, and most
likely will still need to be prompted by voice communication instructions or by the behaviour and
directions given by staff. It cannot be expected that occupants will immediately start evacuation
or relocation movement upon hearing the T-3 signal; more information will always be needed to
ensure occupant safe response during a fire incident in a public building.
NRC Contract: B4521
26
8.0 FUTURE WORK
This study is the very first attempt at looking at the occupant response toward the T-3.
Although this signal was developed over a long period of time, there was never any work to
study the efficiency of this signal with actual building occupants. The findings from this study
open up a large number of questions that should be researched.
The same test should be repeated with occupants of a building where the T-3 is the fire
alarm signal to verify if recognition is better in those occupancies. The T-3 is the signal used for
the fire alarm in a number of new or refurbished office buildings. In those buildings, occupants
should have been trained during an evacuation drill for example in the first few months of
occupation. It would be interesting to assess if occupants can recognized the T-3 after a one
time exposure to the signal during a fire drill, if they can transfer this recognition to other type of
environments and if they still recognize the T-3 when produce with different sounds.
Also the study could be repeated in the United States where smoke alarms emit the T-3 to
assess if this educational approach is indeed effective for signal recognition. In the States
where NFPA 72 is applied, houses in new neighbourhood should all be equipped with smoke
alarms sounding the T-3. Would these people recognize more readily the T-3 signal as a fire
alarm? Would they be able to transfer this knowledge to other type of occupancies? Do they
think they should evacuate when hearing the T-3 signal? These are very interesting questions
that should be studied.
A number of people who have reviewed this work have suggested that the T-3 used in this
study, which was an electronic sound, might explain the poor urgency rating obtained by the
signal. Many suggested that a study should be conducted to assess the difference in
recollection, identification and perceived urgency with the T-3 emitting the sound of the slow
whoop, the bell, buzzer or some other sounds.
Yet if it is planned to introduce a recorded message to inform the occupant that the
activated signal is the new alarm “evacuation” signal, what should be the content of the
message. Little is known at this time on what message content is most effective, what should
be the wording, should it be a male or a female voice, how many times should it be repeated,
etc.
NRC Contract: B4521
27
Much more work is necessary to better understand the impact of introducing the T-3 in
public buildings and how to ensure that occupants’ response matches the fire safety plan
expectations.
NRC Contract: B4521
28
9.0 REFERENCES
CHABA, 1975, “A proposed standard fire alarm signal”, Fire Journal, Vol. 69, No.4, pp. 24-27.
Edworthy, J., 1994, “ The design and implementation of non-verbal auditory warnings” Applied
Ergonomics, Vol. 25, No. 4, pp. 202-210.
Edworthy, J. and Adams, A., 1996, Warning Design. A Research Prospective. Taylor &
Francis, London, 219 p.
Edworthy, J., Hellier, E., Hards, R., 1995, “The semantic association of acoustic parameters
commonly used in the design of auditory information and warning signals”, Ergonomics, Vol.
38, No. 11 pp. 2341-2361.
Edworthy, J., Loxley, S. and Dennis, I., 1991, “Improving Auditory Warning Design:
Relationship between Warning Sound Parameters and Perceived Urgency”, Human Factors,
Vol. 33, pp. 205-231.
Edworthy, J., Stanton, N., 1995, “A user-centered approach to the design and evaluation of
auditory warning signals: 1 Methodology”, Ergonomics, Vol. 38, No. 11, pp. 2262-2280.
Finley, G.A. & Cohen, A.J., 1991, “Perceived urgency and the anaesthetist : responses to
common operating room monitor alarms”, Canadian Journal of Anaesthesia, Vol. 38, pp.958964.
Gibson, J.J., 1979, The Ecological Approach to Visual Perception. Houghton-Mifflin, New York.
Haas, C. H., Casali, G. J., 1995, “Perceived urgency of and response time to multi-tone and
frequency-modulated warning signals in broadband noise”, Ergonomics, Vol. 38, No. 11, pp.
2312-2326.
Hellier, E., Edworthy, J. & Dennis, I.,1995, “A comparison of different techniques for scaling
perceived urgency”, Ergonomics, Vol. No. 38, pp.659-670.
Hellier, E., Wright, D.B. & Edworthy, J., 2000, “Investigating the perceived hazard of warning
signal words”, Risk Decision and Policy, Vol. 5,pp. 39-48.
ISO 7731, 1986, Danger signals for work places – Auditory danger signals. International
Organization for Standardization, Geneva, Switzerland.
ISO 8201, 1987, Acoustics – Audible emergency evacuation signal. International Organization
for Standardization, Geneva, Switzerland.
Karter, J. M., 1998, “Fire loss in the United States during 1997”, National Fire Protection
Association, Internal report, Quincy, MA.
Laroche, C., Tran Quoc, H., Hétu, R., McDuff, S., 1991, “Detectsound: A computerised model
for predicting the detectability of warning signals in noisy workplaces”, Applied Acoustics,
Vol. 32, pp. 193-214.
NRC Contract: B4521
29
Mande, I., 1975, “A standard fire alarm signal temporal or ‘slow whoop’”, Fire Journal, Vol. 69,
No.6, pp. 25-28.
Momtahan, K.L, Hétu, R., Tansley, B., 1993, “Audibility and identification of auditory alarms in
the operating room and intensive care unit”, Ergonomics, Vol. 36, No. 10, pp. 1159-1176.
Momtahan, K.L., 1990, Mapping of psychoacoustic parameters to the perceived urgency of
auditory warning signals, Unpublished Master’s thesis, University of Carleton, Ottawa,
Ontario, Canada.
Momtaham, K.L & Tansley, B., 1989, “An ergonomic analysis of alarm signals in the
operating and recovery rooms”, Annual Conference of the Canadian Acoustical
Association, Halifax, Nova Scotia, Canada.
National Academy of Sciences, Committee on Hearing, Bioacoustics, and Biomechanics,
1974, A Proposed Standard Fire Alarm Signal. Office of Naval Research, Report of
Working Group 73, Contract No. N00014-67-A-0244-0021, USA.
Patterson, R. D., 1982, Guidelines for auditory warning systems on civil aircraft, Civil Aviation
Authority paper 82017.
Proulx, G., 1994, "The Time Delay to Start Evacuating Upon Hearing a Fire Alarm",
Proceedings of the Human Factors and Ergonomics Society 38th Annual Meeting, Vol. 2,
Human Factors and Ergonomics Society, Santa Monica, CA, pp. 811-815.
Proulx, G., Laroche, C., Latour, J.C., 1995, “Audibility problems with fire alarms in apartment
buildings”, Proceeding of the Human Factors and Ergonomics Society 39th Annual
Meeting, Vol. 2, Human Factors and Ergonomics Society, Santa Monica, CA, pp. 989-993.
Stanton, N. & Edworthy, J., 1998, “Auditory affordances in the intensive treatment unit”,
Applied Ergonomics, Vol. 29, pp. 389-394.
Statistics Canada, 2001 <http://www.statcan.ca/english/Pgdb/People/Population/demo31a.htm>
Tran Quoc, H. & Hétu, R., 1996, “La planification de la signalisation acoustique en milieu
industriel: critères de conception des avertisseurs sonores de danger”, Acoustique
canadienne, Vol. 24, pp. 3-17.
Tong, D., Canter, D., 1985, “The decision to evacuate”, Fire Safety Journal, Vol. 9, no 3, pp.
257-265.
WHO, 1999, “Community noise”, World Health Organization, Geneva.
NRC Contract: B4521
30
Appendix 1
Spectrum Analysis of the Signals Tested
NRC Contract: B4521
31
Car Horn
Slow Whoop
Buzzer
NRC Contract: B4521
32
Reverse alarm
Temporal-3
Bell
NRC Contract: B4521