Improving Intelligibility of Airport Terminal Public Address Systems (2017)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

128 Acoustics and Speech Intelligibility Festen, Joost M., and Reiner Plomp. 1990. “Effects of Fluctuating Noise and Interfering Speech on the Speech-Reception Threshold for Impaired and Normal Hearing.” Journal of the Acoustical Society of America 88 (4): 1725–1736. This paper examines the effects of a high background level (80 dBA) for 20 “normal” hearing listeners and 20 hearing-impaired listeners to understand speech. The background noise is pre- sented as a steady-state modulated noise or a single voice. The speech-reception threshold (SRT) is the sound level at which a steady-state-noise (speech) achieves a 50% score. As discussed by others (S. J. van Wijngaarden 2001), this metric can be directly used to indicate the required increase in signal-to-noise ratio (SNR) to achieve the same level of intelligibility as someone with normal hearing. Interfering modulated noise (fluctuating speech) requires an SNR increase of 4 to 6 dB over a steady-state signal; a single competing voice requires an increase of the signal of 6 to 8 dB. For hearing-impaired listeners, an additional 4 dB change is required for modulated noise and 10 dB for a competing voice. For the project, this information is useful to evaluate how to address passengers with hearing loss. For example, this information would suggest that, in an emergency or urgent condition, announcements could be raised in level by about 8 to 10 dB. Kang, Jian. 1998. “Scale Modeling for Improving the Speech Intelligibility from Multiple Loudspeakers in Long Enclosures by Architectural Acoustic Treatments.” ACUSTICA (S. Herzel Verlag) 84: 689–700. Although this paper pre-dates new computer modeling methods that adequately evaluate long enclosures (such as an airport concourse or ticketing hall), the research and discussion of effec- tive acoustical treatments is still relevant. In such spaces, multiple speakers are often used, and their sound fields can interact destructively to degrade the STI. The paper investigates the effec- tiveness of different treatments; all improved the STI, but only three provided substantial results: (1) highly absorptive treatments at the end walls (effectively extending the enclosure and pre- venting long-delay reflections); (2) membrane absorbers (e.g., acoustical ceilings) can be effec- tive, but their effect can be limited by the room geometry; and (3) strategic obstructions to break the room into smaller ADSs. The other two treatments evaluated—ribbed diffusers and porous absorbers—showed small improvements. This information will be useful for architectural design guidance. Kim, Yong Hee, and Yoshiharu Soeta. 2013. “Effects of Reverberation and Spatial Diffuseness on the Speech Intelligibility of Public Address Sounds in Subway Platform for Young and Aged People.” International Congress on Acoustics. Montreal: Acoustical Society of America. This paper explores listening difficulty, which is a relatively new subjective measure of speech intelligibility and compares those results with STI requirements. Twenty people of about 22 years Annotated Bibliography A p p e n d i x A

Annotated Bibliography 129 average age, and 20 people of about 69 years average age were exposed to 12 simulated sound fields for different reverberant and acoustical conditions. On average, the hearing ability of the older group was almost 18 dB lower than the young group. The listening difficulty evaluation was compared to another subjective evaluation, and it was found that only listening difficulty was strongly correlated for both age groups with the STI and reverberation time. Reverberation time was shown to have a high correlation to STI. Reducing the reverberation time below 1.9 seconds showed the strongest effect on increasing STI. This information will be useful for architectural design guidance. Lundin, F. J. 1986. “A Study of Speech Intelligibility of a Public Address System.” (KTH Computer Science and Communication) 27 (1). This paper is one of the few airport-specific studies we found, and it describes research done after the completion of the Arlanda Airport in Stockholm to compare different predictive models for speech intelligibility. One of the departure halls was modeled; it had a ceiling height of 8.7 m (28.5 ft.), with 39 ceiling speakers distributed on a staggered grid pattern at 9.2 m (30 ft). Two dif- ferent signal-to-noise ratios (SNR) were used with two different background noise levels. Speech intelligibility was evaluated using the articulation index (AI), and the study pre-dates the use of STI as an objective measurement, so there is very little that can be directly applied to the current research. However, one interesting observation was that the models were better at predicting the 10 dB SNR case where the speech level was 85 dBA and the background was 75 dBA, than the case where the SNR was 20 dB and the speech level was 70 dBA and the background level was 50 dBA. While a 20 dB SNR is generally better than 10 dB SNR, it is possible that the lower signal level was inadequate for the size of the room. Morimoto, Masayuki, Hiroshi Sato, and Masaaki Kobayashi. 2004. “Listening Difficulty as a Subjective Measure for Evaluation of Speech Transmission Performance in Public Spaces.” Journal of the Acoustical Society of America (Acoustical Society of America) 116 (3): 1607–1613. This paper evaluates the suitability of a new subjective test of “listening difficulty.” Given that subjective intelligibility test results can be highly dependent on the familiarity of the words to the listeners, the authors propose a test based on familiar words to determine the “listening difficulty” under different reverberant field conditions in the laboratory. Despite its focus on subjective evaluation rather than objective measures, this paper provides recent and relevant information on the challenges of speech intelligibility in spaces with long reverberation times consistent with the typical large public spaces found in airport concourses and ticketing halls. The authors found that with a speech signal that is 15 dBA or more above the background noise (signal-to-noise ratio >15 dBA), the speech intelligibility is very high, regardless of reverberant conditions. On the other hand, listening difficulty was markedly affected by SNR and reverberant conditions. A low reverberation time of 0.5 seconds and an SNR of 15 dB provided adequate results, but, at a reverberation time of 2 seconds, the listening difficulty was ranked high, regardless of SNR. This paper provides more support for the necessity of limiting the reverberation time where possible, in particular in areas where critical announcements are made. Morimoto, Masayuki, Hiroshi Sato, and Megumi Wada. 2012. “Relationship Between Listening Difficulty Rating and Objective Measures in Reverberant and Noisy Sound Fields for Young Adults and Elderly Persons.” Journal of the Acoustical Society of America (Acoustical Society of America) 131 (6): 4596–4605. This paper provides more results from the earlier research (Morimoto 2004), in which subjec- tive tests about listening difficulty were performed for a group of young adults and a group of older adults. In the current paper, listening difficulty was directly compared with STI, with the result that the group of older adults required an STI increase of 0.12 points to match the results

130 improving intelligibility of Airport Terminal public Address Systems of the young adults for listening difficulty. This information is useful to evaluate how to reach passengers with hearing loss. Sato, Hiroshi, and Masayuki Morimoto. 2009. “Effect of Noise and Reverberation on Sound Localization of Acoustic Guide Signal for Visually Impaired Persons in Public Spaces.” Ottawa: International Institute of Noise Control Engineering. This paper discusses ways to improve acoustic guide signals as they have been used in Japan. Such signals have been used to guide visually impaired individuals through a complex space. To our knowledge these are not yet used in the United States. If used, acoustical design and guide signal speaker design needs to consider sound localization and SNR, initial delay and reverberation energy (less so reverberation time) as the temporal pattern is important in a reverberant field. Smith, Howard G. 1981. “Acoustic Design Considerations for Speech Intelligibility.” Journal of the Audio Engineering Society 29 (No. 6): 408–415. This paper provides a good overview of the basic concepts underlying this issue, including discussion of the modulation transfer function (MTF), which was the precursor to the STI. The conundrum raised by Lundin above (Lundin 1986) about the signal-to-noise ratio (SNR) and speech intelligibility is discussed. The key issue seems to be reflections. While some researchers advocate that all reflections degrade intelligibility, Houtgast and Steeneken (1972) and Lochner and Berger (1964) indicate that some reflections are helpful. The MTF developed by Houtgast and Steeneken, which takes into account key findings from their research, shows that a high SNR creates reflections that reduce articulation (and intelligibility) and a low SNR allows for the discreet reflections that will improve articulation. This concept is important—speech intel- ligibility issues cannot be solved by increasing the announcement level (and increasing the SNR). Steeneken, Herman J. M. 2014. “Keynote Lecture: Forty Years of Speech Intelligibility Assess- ment (and Some History).” Proceedings of the Institute of Acoustics. Birmingham. This paper, a comprehensive overview of speech intelligibility by one of the pre-eminent researchers in the field, (1) provides a timeline of the development of assessment techniques since 1974, with informative comparisons of differences between the early use of subjective techniques and subjective results between different countries and (2) describes the development and standardization of objective techniques. This paper includes an excellent reference list of foundational documents in speech intelligibility. This information will be useful as supple- mental reading. Tachibana, Hideki. 2013. “Plenary Lecture: Public Space Acoustics for Information and Safety.” International Congress on Acoustics. Montreal: Acoustical Society of America. In this paper, Tachibana summarizes the research efforts of his group and provides informa- tion about field studies to document the ambient conditions at many large, interior, public spaces, including air terminals, railway stations, and shopping centers. The ambient sound environments in these spaces ranged from 60 to over 90 dBA. The reverberation time at five of these spaces is charted, indicating potentially challenging conditions for speech intelligibility with all spaces mea- suring over 1 second up to 2.4 seconds. The research group noted that one air terminal building with “excellent” acoustics measured 2.4 seconds in the middle frequency range. These ambi- ent environments were re-broadcast in a controlled laboratory setting to evaluate the subjects’ responses to these sound pressure levels: environments just over 60 dBA were considered “A little noisy,” whereas environments of 70 dBA were ranked “Moderately noisy.” In ranking the diffi- culty of speaking and listening with someone within 1 m (3 ft), environments greater than about 67 dBA were “A little disturbing,” and environments over 70 dBA were “Moderately disturbing.” The research group also used subjective evaluations of PA system speech intelligibility with a mix of native language (Japanese) and non-native (mostly Asian language) subjects for different

Annotated Bibliography 131 conditions of room reverberation time and background noise. With an air conditioning equip- ment background noise level of 65 dBA, native listeners rated the conditions “a little difficult” for reverberation time of about 4 seconds, with “fairly difficult” conditions for 5 seconds or longer. Non-native listeners had a “fairly difficult” experience with reverberation time exceeding 1 second. This information is consistent with other research and will be useful for architectural design guidance. van Wijngaarden, Sander J. 2001. “Intelligibility of Native and Non-Native Dutch Speech.” Speech Communication (NH Elsevier) 35: 103–113. This paper, developed from van Wijngaarden’s doctoral research, compares how speakers of Dutch were able to understand the language under different acoustical conditions. The non- native speakers were all Dutch-speaking Americans who had lived in the Netherlands for 1 to 3 years. Two different, subjective tests were applied, but the one that was most directly linked to objective differences was the speech reception threshold (SRT) evaluation, which provides a solid measure for the speech intelligibility of whole sentences at a speech-to-noise (SNR) ratio that corresponds to 50% understanding for short, everyday sentences; the SRT listening test with non-native speakers required an additional boost in the SNR of 3 to 4 dB compared to native speakers. The paper also includes results on the effectiveness of Dutch spoken by non- native speakers. This information is consistent with other research and will be useful for design guidance—for example, this information would suggest that in an emergency or urgent con- dition, announcements should be raised in level by about 3 to 4 dB for international terminals where a higher percentage of travelers are non-native speakers. van Wijngaarden, S. J., et al. 2004. “Using the Speech Transmission Index for Predicting Non- Native Speech Intelligibility.” Journal of the Acoustical Society of America 115 (3): 1281–1291. This paper (which extends van Wijngaarden’s 2001 paper to relate standard STI label catego- ries to corresponding results for non-native speakers) presents the results for five characteriza- tions of non-native speakers by earlier researchers, depending on the relative age at which the listener learned the test language (early or late) and the proficiency of the listener at understand- ing the test language (high and low). These previous studies used different subjective tests, which were corrected to the corresponding STI values. Four of these earlier tests were conducted in English and one was conducted in German. Non-native listeners’ corresponding STI values for fair speech (0.45 points) ranged from 0.50 for early learners to 0.74 for late learners; low- proficiency subjects required an STI 0.60 for fair speech, while high-proficiency subjects also tested to 0.50. The paper includes (1) results of speech reception threshold (SRT) to investigate the effects of bandwidth limiting and reverberation time for non-native listeners and (2) results for Dutch spoken by non-native speakers. This information will be useful for design guidance— for example, this information would suggest that in comparison with domestic terminals, a higher STI target should be considered for international terminals where a higher percentage of travelers are non-native speakers. van Wijngaarden, S. J., H. J. M. Steeneken, and J. A. Verhave. 2011. “The Future Is Bright for the Speech Transmission Index; Dealing with New Challenges after Four Decades of Development.” Proceedings of the Institute of Acoustics. Seattle. A comprehensive overview of the electronic instrumentation involved in evaluating speech intelligibility, current challenges with STI applications, and an outlook on upcoming develop- ments. Compared with a similar overview approach by Steeneken (2014), this paper is more focused on the development of electronics and measurements standards that enable the objec- tive assessment of SI. The paper includes a useful reference list. This information will be useful as supplemental reading.

132 improving intelligibility of Airport Terminal public Address Systems Yokoyama, Sakae, and Hideki Tachibana. 2008. “Study on the Acoustical Environment in Public Spaces.” Shanghai: International Institute of Noise Control Engineering. This paper provides some of the background alluded to in Tachibana’s 2013 presentation. Specifically, the airport had an acceptable background noise level (57 dBA) with high 19 m (63 ft.) ceilings and acoustical treatment on the ceilings and walls. In the observations of the researchers, the acoustical conditions were well-designed for subjective speech intelligibility. This information will be useful for evaluating the field measurements. Human Factors Alm, M., & Behne, D. 2015. “Do Gender Differences in Audio-Visual Benefit and Visual Influ- ence in Audio-Visual Speech Perception Emerge with Age?” Frontiers in Psychology, 6, 1014+. Similar to the Amano-Kusumoto research, this study supported the findings that females are typically more intelligible speech readers than males. This information will be useful for design guidance to define ways to improve message intelligibility. Amano-Kusumoto, A., & Hosom, J.-P. 2011. A Review of Research on Speech Intelligibility and Correlations with Acoustic Features. Oregon Health & Science University (OHSU), Depart- ment of Biomedical Engineering. Beaverton: Center for Spoken Language Understanding (CSLU). This study reviewed current research on speech intelligibility. Noteworthy was that gender plays a part in intelligibility—cited were two studies in which female speakers were more intelli- gible than males. The researchers hypothesized that this could be because female speakers tend to have larger vowel spacing and more precise inter-segmental timing than male speakers, although it was not clear whether other factors such as frequency could have been in play here. This infor- mation will be useful for design guidance to define ways to improve message intelligibility. Bor, R. 2007. “Psychological Factors in Airline Passenger and Crew Behavior: A Clinical Overview.” Travel Medicine and Infectious Disease, 5, 207–216. Air travel can induce considerable stress in individuals who are simply outside their natural environments and comfort zones. It has been suggested that air travel especially “can induce depression, anxiety, panic attacks or even psychosis in vulnerable individuals.” This paper further suggests that cultural background, gender, and age may mediate how passengers deal with stress during traveling. For this project, this research suggests that stress may be a strong determining factor in how well messages are attended to. Cherry, E. C. 1953. “Some Experiments on the Recognition of Speech, with One and Two Ears.” Journal of the Acoustical Society of America, 25(5), 975–79. Research in attention and perception suggests that even relaxed participants who are pre- primed with an expectation of a message will, on some level, be “ready to attend” to the mes- sage. This ties in with the “Cocktail Party Effect” found by Cherry and dichotic listening tests. For instance, one may be attending to a particular conversation, but will pick up one’s name in another conversation. In application to this project, the suggestion is therefore twofold: (1) even when relaxed, regular travelers will expect updates/travel information in the form of auditory announcements and so it is suggested therefore that they are more attuned to them than novice travelers; and (2) the attention of regular travelers will be drawn more to specific messages than novice travelers who may be overwhelmed with new stimuli. Festinger, L. 1957. A Theory of Cognitive Dissonance. Stanford: Stanford University Press. “Cognitive dissonance” occurs when an individual holds two conflicting attitudes or beliefs. The dissonance refers to the sense of discomfort felt by the individual. Festinger proposed the

Annotated Bibliography 133 theory of cognitive dissonance and suggested that we all have an inner drive to hold all of our inner beliefs and attitudes in harmony and, therefore, seek to avoid or reduce disharmony. In an airport environment, this could be as simple as a passenger told at the check-in desk that they would be boarding at Gate 6 and an announcement that calls passengers to Gate 8. Unless the call to Gate 8 was announced as a specific gate change in the message, the passenger might find the two pieces of information in conflict and assume the auditory message was in error. An oral message delivered in person by an airport employee (e.g., at a check-in desk) would exert more influence over a decision-making conflict than a PA message—unless that message was announced as a change. Forrester, A. M. (2007). Auditory Perception and Sound as Event: Theorising Sound Imagery In Psychology. Retrieved January 21, 2016, from T[H]E [EAR] OF THE DUCK: https:// theearoftheduck.wordpress.com/2012/10/15/auditory-perception-and-sound-as-event-theorising- sound-imagery-in-psychology/ Forrester notes in his paper that speech is “sound first and ‘text’ second.” He notes that work by Rodaway (1994) highlights a gap between sound as a “perceptual experience” and the actual recognition of the meaning of the sound. Forrester also highlights the evolutionary perspective of perceiving sound. In much the same way as we are evolutionarily pre-programmed to associ- ate red with danger, we will react to loud, disruptive noises or noises that we have, over time, come to associate with a particular warning (e.g., we may associate sirens with fires). This evolu- tionary reaction will (1) precede any understanding of a message informing us that there is a fire and where we should proceed to evacuate to and (2) will serve to cue us both to pay attention to the following instructions and to the basic meaning of the following message. Fritz, J. B., et al. 2007. “Auditory Attention—Focusing the Searchlight on Sound.” Current Opinion in Neurobiology, 17, 1–19. This research discusses how our processing for auditory attention can be bottom up or top down. Bottom-up processing begins with the stimulus and the stimulus influences what we per- ceive. For example, one starts with no preconceived idea of what one is hearing and the stimu- lus itself influences the perception of what one hears. It is data driven and the perception of the message itself directs people’s cognitive awareness of what they hear. Top-down processing uses background knowledge, learning, expectations, and current goals to influence perception. Behavior and processing are influenced by expectations. With top-down processing we use what we know to understand what we are perceiving—it is goal driven (voluntary or task-dependent). In bottom-up processing, we use the auditory stimulus itself to drive our perception (sound- based salience). Top-down processing focuses on the expected features of an auditory target. This aligns with an experienced passenger’s response to flight information messages. An experienced passenger is likely to use top-down processing to seek information from auditory messages and will have expectations about the information and format of those messages based on experience. It is assumed that the experienced passenger would be an effective listener, actively seeking the information from the message and able to understand, through experience, what is required from them and how to act. Experienced passengers employing top-down processing would know what they are searching for and employ a template-based search. Relevant to this project, inexperienced passengers, having no expectation of the flight infor- mation messages, would use bottom-up processing—they would seek information from the stimulus and identify salient points in the message in order to understand it. Bottom-up saliency is the process of identifying salient points using features extracted from the sound (e.g., names and dates) and comparing them with its neighbors. Bottom-up salient detection includes detect- ing parts of the auditory signal that attract people’s attention in terms of contrast or characteristic

134 improving intelligibility of Airport Terminal public Address Systems features. Given that novice passengers may not be actively listening, they will be more likely to employ bottom-up processing. Hodoshila, N., Arai, T., & Kurisu, K. 2008. “Effects of Training, Style and Rate of Speaking on Speech Perception of Young People in Reverberation.” Presented at Acoustics ’08. Paris. Retrieved from www.acoustics08-paris.org This paper suggests a benefit to conversational speech over clear speech. The authors looked at the effects of training, style and rate on speech perception in simulated reverberant environ- ments (to replicate conditions for spoken messages over loudspeakers in public places such as railway stations). They did not find a difference in slowed speaking rates in reverberant conditions. In contrast to the other two similar studies, they found conversational speech had a higher correct rate than clear speech (82% compared to 78.6%). They hypothesized that this could be due to the fact that the reverberant conditions masked features of the clear speech. They noted that it was possible for the characteristics of clear speech to be varied in environments where clear speech was recorded. From this they suggested that recording a clear speech auditory mes- sage in reverberation conditions would produce a much higher correct rate (rate at which the message was correctly perceived/understood) as the speakers may adjust their style to be more intelligible in that particular environment, mimicking the Lombard Effect, where the involun- tary tendency of speakers is to increase their vocal effort when speaking in loud noise to enhance the audibility of their voice. Pitch, duration of syllables, and rate are also implicated with loud- ness in this phenomenon. Most research supports clear speech, and it is possible that these authors have identified a key point to clear speech – that reverberant conditions may mask some elements of it. This informa- tion will be useful for design guidance to define ways to improve message intelligibility. Iwamiya, S.-i., et al. 2004. “Design Specifications of Audio-guidance Systems for the Blind in Public Spaces.” Journal of Physiological Anthropology and Applied Human Sciences, 23(6), 267–271. These researchers gained feedback on message content of PA announcements during a study that tested an audio-guidance system for the blind in public spaces, with particular ref- erence to travel facilities. The study was conducted at Tojinmachi Station on the Fukuoka City Subway. The blind participants noted the cognitive load involved when they had to both rec- ognize an auditory signal and understand any information contained in any message associated with the signal while walking through an environment, a task that carried a high load by itself. For this reason, these participants believed that some PA announcements were unnecessary or overly long. For example, they considered welcome greetings and polite expressions unnecessary and stated that the messages should be short and simple and just contain the key information that the passenger requires to act on. Labiale, G. 1990. In-Car Road Information: Comparisons of Auditory and Visual Presenta- tions. Proceedings of the Human Factors Society 34th Annual Meeting. This presentation considered different information presentation forms (visual/auditory/ repeated auditory) about in-car information. Although difficult to generalize between in-car and airport environments, the study did provide support for accurate recall of auditory messages of 7 to 9 information units by a significant number of tested drivers (93.6%). Processing of audi- tory messages requires the retention of this information to allow time for recognition of words and comprehension to take place. This is known as Short-Term Auditory Memory and works in much the same way as short-term memory itself works. Labiale’s paper provides support for “Miller’s Magic Number 7 (plus or minus 2)” (Miller 1956).

Annotated Bibliography 135 Lotto, A., & Holt, L. 2011. “Psychology of Auditory Perception.” Wiley Interdisciplinary Reviews: Cognitive Science, 2(5), 479–489. This paper describes that limited research in the field of complex auditory perception makes it difficult to categorically define how complex sounds–such as conversational speech–are affected by preceding and following sounds. Studies frequently focus on simple tones and signals pre- sented in isolation. Airports are aurally complex environments to begin with, given that they have background music, machine sounds, aircraft noise, retail outlet information or music, and conversational background from traveling passengers. Mense, B., Debney, S., & and Druce, T. (2006). “Classroom Listening and Remembering.” In Ready, Set, Remember: Short-Term Auditory Memory Activities. Camberwell: ACER Press. Motivation to listen to a message will strongly affect the amount of information understood or effectively processed from an auditory signal. Listeners may be directing their attention to other auditory inputs from mobile devices and so exhibit a lack of interest in auditory signals provided by the airport. This is similar to research by Umera-Okeke. Miller, G. 1956. “The Magical Number Seven, Plus Or Minus Two: Some Limits on Our Capacity for Processing Information.” The Psychological Review, 63, 81–97. A noted psychological theory on short-term memory states that most adults can store between 5 and 9 items in their short-term memory. This theory has been expanded to state that, if infor- mation can be grouped or chunked together, more information can be stored. The relevance of this with regard to PA announcements is support for keeping messages short and simple—5 to 9 items are ideal, a sentiment also expressed in research by others. An example of putting key information first and keeping the message short and concise follows: “Chicago, Chicago. Flight AA6754 now boarding at Gate 6.” Moran, M. 2012. “Designing for Intelligibility vs. Audibility.” Eaton, Cooper Notification Solutions. Long Branch: white paper. In highly reverberant spaces it may be prudent to identify areas in which intelligibility can be obtained and to highlight these locations by design features within the environment. This white paper refers to these as “rally points” and may be viewed in the same way as passengers clustering around visual information boards. This information will be useful for design guidance to define ways to resolve problems with highly reverberant spaces. Payton, K. L., Uchanski, R. M., & and Braida, L. D. 1994. “Intelligibility of Conversational and Clear Speech in Noise and Reverberation for Listeners with Normal and Impaired Hearing.” Journal of the Acoustical Society of America, 95(3), 1581+. These researchers considered the intelligibility of clear speech and conversational speech with both “normal” and hearing-impaired participants. Clear speech is defined as having a slower speaking rate, greater speech intensity and emphasis, increased emphasis on consonants com- pared to adjacent vowels, and increased word duration. They found that clear speech was more intelligible across participant types and across degraded listening conditions (e.g., additive noise and reverberation). It was also noted that, as noise levels increased, the difference in scores between the two types of speech also increased. This information will be useful for design guid- ance to define ways to improve message intelligibility. Potter, R., & Choi, J. 2006. “The Effects of Auditory Structural Complexity on Attitudes, Attention, Arousal, and Memory.” Media Psychology, 8, 395–419. Research into radio messages proposed that when structurally complex and structurally simple auditory messages were played, participants showed improved memory for the audio messages which were more structurally complex. Structurally complex for the purpose of this research paper was defined as containing multiple voice changes, sound effects, music onsets, and/or production

136 improving intelligibility of Airport Terminal public Address Systems effects, NOT making the message itself complex. Structurally complex messages resulted in more positive attitudes to messages, greater arousal (reported by galvanic skin response and cardiac monitors), greater memory for the message and larger self-reported attention. Many of these pro- duction effects would be inappropriate in the airport environment and add to auditory clutter. To translate the findings of this paper into use for airports, it is suggested that different announcers (i.e., voices) could be used for different message types: this would provide structural complexity with differing tones. Proctor, R. W., & Zandt, T. V. 2008. Chapter 7, “Hearing, Proprioception, and the Chemi- cal Senses.” In Human Factors in Simple and Complex Systems (pp. 165–185). Boca Raton: CRC Press. This text describes work by Miller and Isard in 1963 in which the researchers presented normal sentences (e.g., bears steal honey from the hive); semantically anomalous but grammatically cor- rect sentences (e.g., bears shoot work on the country); and ungrammatical strings (e.g., across bears eyes work the kill), to listeners. They found the lowest recall rate for the ungrammatical strings, followed by semantically anomalous sentences. The best recall was obtained from the meaningful (normal) sentences. These results indicate that, while perception of an auditory message is helped by grammatically correct sentences, semantic context is also important. This information will be useful for design guidance to define ways to improve message understanding. Spence, C., & Santangelo, V. 2010. “Auditory Attention.” In E. C. Plack, Oxford Handbook of Auditory Science: Hearing (pp. 249–270). Oxford: Oxford University Press. These authors note that many studies now show that people can only effectively attend to one auditory stimulus at a time. They also note that most studies are conducted in silent laboratory conditions with only one or two stimuli presented. They suggest that, in complex auditory envi- ronments, our awareness is much less than we would believe. They state that “in the absence of attention, people have no conscious awareness of most of the auditory stimuli around them.” This suggests limited capacity for attention and that individuals need to focus their attention on a single object or stream within the auditory scene around them in order to process the auditory information correctly. This supports reducing other auditory distractions where possible prior to message presentation. This text also refers to a study by Conway et al. in 2001 in which, having tested a group of par- ticipants on working memory performance, they found that a group of “low span” participants found it harder to filter out irrelevant information. Individual differences in working memory capacity were found to correlate with an individual’s ability to selectively focus their auditory attention to a particular auditory stream. There was evidence to suggest that the low span partici- pants also attended to a background irrelevant message to a better degree than the high span par- ticipants. However, it is noted that the background message contained the participant’s name. It is possible that the presentation of their name caused them to focus on that message rather than the message they were asked to attend to, suggesting a degree of plasticity in their focus. Tavassoli, N., & and Lee, Y. 2003. “The Differential Interaction of Auditory and Visual Advertising Elements with Chinese and English.” Journal of Marketing Research, 468–480. A lack of auditory distraction does not always lead to greater attention to an auditory stimu- lus. This paper discussed work in 1989 by Anand and Sternthal which suggested that an abun- dance of available cognitive resources may lead to the generation of idiosyncratic thoughts which may distract and result in less attention to the auditory message to be played. For exam- ple, when we compare this to a work environment, periods of low workload and stimulation result in lapses in attention. This could be translated in the passenger’s case to periods of wait- ing with little to do leading to daydreaming, low levels of attention, and a likelihood of missing announcements.

Annotated Bibliography 137 Tsimhoni, O., Green, P., & Lai, J. 2001. “Listening to Natural and Synthesized Speech while Driving: Effects on User Performance.” International Journal of Speech Technology, 4(2), 155–169. Comprehension of text-to-speech synthesized speech messages was compared to that for natural speech messages in a study undertaken while driving. The study found that varying driver workload did not affect comprehension, but that comprehension of synthesized speech in the text-to-speech condition was significantly worse than the natural speech information condi- tion. This is consistent with other research and will be useful for design guidance to define ways to improve message understanding. Umera-Okeke, N. 2008. “Listening Effectively for Results in an ESL/EFL Classroom.” African Research Review, 1(1), 47–54. This researcher notes that individual listening types will always affect attention to messages. These listening types are identified in a teaching environment but may be assumed to be gener- alizable across individuals and, therefore, relevant to passengers. Related to the project, bored, tired, and inattentive listeners may have been subject to delays and have simply “switched off ” from external information. This researcher also refers to controlling listeners and describes them as people who prefer always to talk rather than to listen. This particular type of listener may dis- regard external cues and messages. This information will be useful for design guidance to define ways to improve message understanding. Van Horn, L. 2007. “Disability Travel in the United States: Recent Research and Findings.” 11th International Conference on Mobility and Transport for Elderly and Disabled Persons (TRANSED). Montreal. June 18–22, This study summarizes work in 2005 by the Open Doors Organization, which undertook a travel market study looking at 1,373 adults with disabilities traveling. One of their findings was that 17% of all passengers surveyed said that they had difficulty hearing announcements. While unable to ascertain the participant breakdown for this study, it is suggested that not all of the 17% may have hearing impairment. This information will be useful for design guidance to explore ways to improve message communications. Venkatagiri, Horabail S. 2003. “Segmental Intelligibility of Four Currently Used Text-To- Speech Synthesis Methods.” Journal of the Acoustical Society of America 113 (No. 4, Pt. 1): 2095–2104. It can be difficult to understand artificial voice transcription of electronically stored text, and this paper provides review and discussion of the shortcomings of text-to-speech (TTS) systems available in 2003. This paper studies intelligibility of four TTS systems compared to a control human voice under challenging signal-to-noise conditions. The paper indicates that (1) intel- ligibility is improved for a signal-to-noise ratio (SNR) of 5 decibels (dB) compared to 0 dB, and (2) listeners tend to process artificial voice differently, once they realize that a TTS system is being used. This information will be useful for design guidance. For example, this information could suggest that announcements with artificial voice be repeated to extend the overall duration and allow passengers to adjust to the voice delivery. This work could also imply that synthesized voice may not be suitable for international terminals. Yokoyama, Sakae, and Hideki Tachibana. 2013. “Subjective Experiment on Suitable Speech- Rate of Public Address Announcement in Public Spaces.” International Congress on Acoustics. Montreal: Acoustical Society of America. This is another paper on text-to-speech (TTS), in this case evaluating the suitable speech-rate the TTS system should be set to. The research evaluates the speech rate based on subjective tests that measured listening difficulty and speech intelligibility for Japanese language words. One key find- ing of this research is that reverberation time is the most important factor to consider for speech

138 improving intelligibility of Airport Terminal public Address Systems intelligibility by non-native listeners; native listeners were not much affected by reverberation time or speech rate. However, listening difficulty was affected for both native and non-native listeners when the speech rate and reverberation time were changed. This information will be useful for design guidance. For example, this work could imply that synthesized voice may not be suitable for international terminals. Zhang, Y., et al. 2005. “Effects of Language Experience: Neural Commitment to Language Specific Auditory Patterns.” NeuroImage, 703–720. This paper discusses work by Strange in 1995 in which it is stated that speech perception is affected by an individual’s language experience and that adult non-natives have difficulty in dis- criminating other language contrasts. Using magnetoencephalography, these researchers found that processing non-native speech required a significantly longer period of brain activation. It is suggested that this may lead to frustration and confusion if the whole message was not attended to or if the non-native speaker’s attention had been drawn to the message part way through. This supports an argument for repeating auditory messages. This information will be useful for design guidance to define ways to improve message understanding.