Soundscape attributes in Spanish: A comparison with the English version of the protocol proposed in Method A of the ISO/TS 12913–2

Soundscape attributes in Spanish: A comparison with the English version of the protocol proposed in Method A of the ISO/TS 12913–2 Jerónimo Vida a, ⇑ , José Antonio Almagro b , Rafael García-Quesada c , Francesco Aletta d , Tin Oberman d , Andrew Mitchell d , Jian Kang d a Applied Physics Department, Faculty of Sciences, University of Granada, Granada 18071, Spain b Scientific Instrumentation Centre, CITIC, University of Granada, Granada 18071, Spain c Building Construction Department, School of Architecture, University of Granada, Granada 18071, Spain d Institute for Environmental Design and Engineering, University College London, London WC1E 6BT, UK article info Article history: Received 22 February 2023 Received in revised form 18 June 2023 Accepted 26 June 2023 Keywords: ISO 12913 Soundscape attributes Soundscape assessment Circumplex model Spanish language abstract In soundscape studies there is a need to validate the ISO 12913 framework and questionnaire protocols in non-English speaking regions, which is being addressed via an international translation effort carried out within the Soundscape Attributes Translation Project (SATP). The objective of the SATP initiative is to promote the widespread use of the soundscape circumplex model attributes by validating proposed translations using standardized listening experiments in different languages and geographical regions. This study compares the Spanish (ISO 639:spa) translation of the ISO soundscape circumplex model with the original English version and introduces the Structural Summary Method (SSM) for analysis. The research findings suggest a reasonable accordance in listening experiments carried out by native English and Spanish speakers, but also highlight the need for revision of the ‘‘vibrant/monotonous” and ‘‘eventful/uneventful” dimensions to harmonize the interpretation of each dimension by the affected population. These revisions not only improve the translation of the model into Spanish, but also raise questions about the adequacy of the original English formulation of the model. The study underscores the cultural differences underlying the meaning of these dimensions and the challenges of objectifying the soundscape circumplex model within the ISO 12913 framework. Ó2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction The Soundscape Attributes Translation Project (SATP) started in 2019, initiated by the Acoustic Group at University College London (UCL), The Bartlett, Institute for Environmental Design and Engineering [1], after the Technical Specifications for soundscape data collection (Part 2) [2] and analysis (Part 3) [3] in the ISO 12913 standards series were released [4]. This standard came to assist a new approach to urban environmental noise management based on citizens’ perception, a new concept that demanded standardized evaluation methods of urban soundscapes. In this context, Part 2 of the ISO 12913 series [2] provided researchers and practitioners around the world with a method and a reliable questionnaire for soundscape characterization. Collected subjective data on the perception of the acoustic environment in context should, afterwards, be analysed according to Part 3 of ISO 12913 [3]. The Soundscape approach as defined by ISO 12913–1:2014 is a relatively new but consolidated framework, leading the way as the reference method to address people’s experiences of acoustic environments [5]. It also suggests how such information should be integrated with outputs from traditional engineering methods for urban noise management, based on strategic noise mapping and noise action plans design. Part 4 is still an unpublished draft [6] and will cover more soundscape design-oriented elements. Overall, the ISO 12913 protocols refer to soundscape literature for the proposal of a Perceived Affective Quality (PAQ) model, sometimes defined as the ‘‘soundscape circumplex” [7]. The main drivers of the SATP initiative were the search for standardized tools to gather and report individual responses on the perception of urban acoustic environments and to provide comparable datasets and soundscape characterizations across different populations’ samples, times, and locations, as well as allowing for replicability studies. These were also the main arguments for the establishment of the original Working Group (WG54) at the https://doi.org/10.1016/j.apacoust.2023.109516 0003-682X/Ó2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ⇑ Corresponding author. E-mail addresses: [email protected] (J. Vida), [email protected] (J. Antonio Almagro), [email protected] (R. García-Quesada), [email protected] (F. Aletta), t.oberman@ucl. ac.uk (T. Oberman), [email protected] (A. Mitchell), [email protected] (J. Kang). Applied Acoustics 211 (2023) 109516 Contents lists available at ScienceDirect Applied Acoustics journal homepage: www.elsevier.com/locate/apacoust International Organization for Standardization (ISO) back in 2008 under the ISO/TC43 [8]. The group has grown and gained international representation, while research and technical interest in urban soundscape has also increased since then, and is currently acknowledged also in policy documents [9,10]. However, while the ISO 12913 framework provides a theoretical model and a reliable questionnaire for soundscape assessments, it only includes an English version of the instrument. Therefore, the reliability of the model’s attributes in non-English speaking regions was in 2019 (and still is nowadays) an open question. The SATP initiative comes to help on this issue and contributes to an international translation effort of the ISO soundscape assessment method in as many languages as possible [1]. The long-term objective of this project is to validate the proposed translations using standardized listening experiments in different languages and geographical regions. This would promote a widespread use of the soundscape circumplex model attributes and, with time, its inclusion in policy documents for a practical use of evaluations in sustainable urban design and noise action planning, which has already started to happen in Wales or Italy [9,11]. After the first translations were proposed by the different institutions taking part in the project, named SATP Stage 1, listening experiments were carried out separately by all participants in the local language with native speakers, using a common set of auditory stimuli and standardized equipment and procedure for calibration. This process, called SATP Stage 2, aimed at validating the translations in the different languages [12,13]. But as new partners keep joining the SATP network and more languages are included in the inventory, an open research question remains: are the national languages translations from the English version of the ISO protocol, ultimately correct and usable? Would citizens and members of the public in different countries around the world assign the same meaning to attributes? This work proposes some answers for the Spanish translation of the soundscape circumplex model, as seen by native Spanish speakers in Spain and compared with data from native English speakers in the United Kingdom. Spanish (ISO 693:spa) is one of the world’s most widely spoken languages, it is estimated to have 595 million speakers globally and is the official language in 21 countries [14]. Despite its widespread use, the availability of soundscape data collection protocols in Spanish is limited in the context of soundscape literature. This is evident from the limited examples of protocols used directly in Spanish, as referenced in sources [15–17]. This scarcity of resources highlights the need for further research and development in this area, particularly in Spanish-speaking countries, to ensure that the language is represented in the soundscape research community. It is also necessary to continue with the work of harmonizing initiatives, to coordinate other translation proposals that do not originate within the SATP project [18]. In view of the above, the aim of this paper is to investigate the adequacy and convergence of the translation into Spanish of the ISO protocol for soundscape evaluations, only formulated in English in the standard. The purpose is to carry out a comparison (Spanish-English) to see whether the Spanish interpretation of the circumplex model conforms to the principles that inspired its original formulation in English and contribute to the standardization of the soundscape evaluation process all over the world. 2. Methodology 2.1. Preliminary translation into Spanish (Stage 1) The translation of perceptual attributes into Spanish was carried out through an iterative process between project partners and colleagues from Chile and Spain [1]. Three researchers, from University of Chile (UChile, Dr. Pablo Kogan), Austral University of Chile (UACh, Dr. Enrique Suárez), and University of Granada (UGR, Dr. Jerónimo Vida), first exchanged files and information concerning their own work and experience on soundscape research involving public surveys. Beginning with the original translation from Swedish given in Axelsson et al. [7] and the slightly altered version in ISO 12913– 2:2018, an iterative translation process was undertaken lasting approximately a month and a half. The team kept in contact and shared information throughout this process, paying special attention to the different meanings and interpretations of words in Spain and Chile, as well as other Spanish speaking countries in Latin America. The primary concern was to choose words with clear and unambiguous meanings, striving to avoid possible misunderstandings or misinterpretation of words which might be considered profane or vulgar. This kind of possible interpretation appears in translations of words like ’pleasant’ or ’exciting’ where we found quite a wide variety of different interpretations from one side of the Atlantic to the other. For this process, field experience of using soundscape questionnaires in Spanish in Chile and Spain was considered. In addition, researchers interviewed colleagues, students, members of the public, and shared preliminary findings as they came out. After a period of reflection and discussion, a consensus was reached, resulting in a Spanish translation with two proposed words for each component. The first set of words, as shown in Table 1, is considered the most preferred. Environmental degree third year students at University of Granada taking an Acoustic course with Dr. J. Vida took part in the exploration of words and synonymous of the model components. They were all native Spanish speaking volunteers from Spain, around 20–25 years old. Applied Physics department colleagues and friends were also asked to think about the Spanish two-word proposal. They were all native Spanish speaking persons from Spain, 35–60 years old. Finally, some native Spanish speaking mid age (30–40) technicians from Granada City Council Environmental Department were also consulted about this task, but differences from scientific staff at University of Granada were minimum. The experience from soundscape research by means of surveys containing the Swedish Soundscape-Quality Protocol (SSQP) [7] applied in Spanish in Argentina and Chile from 2013 to 2019, was the principal source of information that fed the translation criterion as reported by Dr. P. Kogan. Besides this background and specifically for the SATP Spanish work, Dr. P. Kogan established communication with colleagues and other professional from different countries of Latin America. The six collaborators were a soundscape researcher (7 years active in soundscape) native Spanish speaker from Colombia (male, mid age); an environmental acoustics researcher native Spanish speaker from Ecuador (male, mid age); an acoustics engineer native Spanish speaker from Bolivia (male, mid age); a language teacher native Spanish speaker from Chile (female, mid age); a senior psychologist native Spanish speaker from Argentina (female) and a senior architect native Spanish speaker from Argentina (male). Table 1 SATP Spanish Stage 1 translation. ISO 12913–2: 2018 Spanish - Word 1 (preferred) Spanish - Word 2 (synonymous) pleasant (p) agradable placentero chaotic (ch) caótico confuse vibrant (v) estimulante vibrante uneventful (u) sin actividad estático calm (ca) calmado tranquilo annoying (a) desagradable molesto eventful (e) con actividad dinámico monotonous (m) monótono aburrido J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 2 Finally, researchers from the Institute of Acoustics of the Universidad Austral de Chile collaborated in the selection of the words for the Spanish language (native Spanish speakers from 35 to 55 years old) with Dr. E. Suárez. In addition, a final year student of Acoustic Civil Engineering worked on his final master thesis research on this topic of soundscape. In their research they applied the proposed words in surveys to 30 students (20–25 years old), who evaluated 12 soundscapes. This process is briefly described in Fig. 1. 2.2. Listening experiment in Spain (Stage 2) The recordings used in the listening experiment in Spain (and in other countries by other SATP partners) were made at different locations in London (UK), at various times of day by an operator wearing a binaural kit consisting of BHS II microphones and a SQobold (HEAD acoustics) device. Recordings were then exported to WAV via the ArtemiS SUITE software, using the original dynamic range from HDF. The listening experiment and the calibration procedure were intended for a headphone playback system (Sennheiser HD650 or similar open-back headphones recommended). The final set of recordings was selected from an initial pool of 80 recordings through a pilot study conducted at UCL to ensure the test set had an even coverage of the soundscape circumplex model. Audio stimuli were selected with the aim to cover a broad enough range of sound environments that people could reasonably expect in any urban context in a global city (i.e., not posing risks of sounding ‘‘unfamiliar” for semantics or cultural reasons). The Audio stimuli included sound environments with varied sound sources compositions (e.g., both natural and mechanical sound sources), which had the goal to elicit the full spectrum of assessments on the soundscape circumplex as a whole (i.e., on either of the soundscape attributes scales scored by the sample) [19]. Stage 2 was performed according to SATP common instructions given to all partners in the project: the listening tests were done at UGR facilities with negligible background noise levels (Anechoic Acoustic Chamber and experimental labs at Faculty of Sciences) using Sennheiser HD600 headphones connected to the computer with a sound card. To experience the binaurally recorded soundscapes with the same loudness as found in-situ during recording, the headphones were previously calibrated following the aforementioned common SATP protocol. Stimuli were randomly sorted and presented to each participant, asking them to listen to the 30-second Audio first and then rate the scales. Participants were allowed to re-listen to any stimulus as many times as they wish. The question for each stimulus followed ISO recommendation: ‘‘For each of the 8 scales below, to what extent do you agree or disagree that the present surrounding Fig. 1. Brief Stage 1 process description and timeline. Table 2 Selected audios because of their apparent distance Spanish-English in ISO circumplex diagram; the colour scheme is the same as in Fig. 5. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 3 sound environment is...” followed by the 8 PAQ model dimensions in Spanish using the translation performed in Stage 1 and shown in Table 1. Response data were collected with printed sheets in which a 10 cm segment ranging from 0 (label ‘‘strongly disagree”) to 100 (label ‘‘strongly agree”) let participants mark their answers. These data were afterwards converted into the (1–5) Likert scale according to ISO and the SATP protocol. The previously described methodology took place in Spain during May, June and July 2021, but is still under development in Chile. So, a comparison of the Spanish translation of the PAQ model from Spanish native speakers’ results was decided, pending new data from non-European Spanish speakers that will arise after completion of Stage 2 in Chile and any other Spanish speaking country that would join STAP project in the future. 2.3. Data analysis In order to carry out the comparison, data from the listening experiments carried out in UK and Spain were used. English data consists in answers from 32 participants and Spanish data in answers from 29 participants. As listening experiments involved the same 27 selected audios, a total of 864 coordinates for each of the PAQ 8 dimensions conform the English data set and 783 coordinates the Spanish data set. Average age for Spanishspeaking participants was 24.4 years, a population composed by 10 females (34.5%) and 19 males (65.5%) giving a sex ratio M/F of 1.9. The average age for English-speaking participants was 29.7 years with 13 females (40,6%) and 19 males (59,4%) and a sex ratio of 1.5. While for both samples the sex ratio is not particularly well-balanced, previous research in soundscape studies suggests that when participants are asked to score specific ISO-related soundscape dimensions, male/female differences actually account for a relatively small amount of variance [20], so this concern was disregarded in this case. Following generally accepted criteria and ISO 12913–3 recommendations [3] for data analysis, descriptive statistics was computed for each of the 8 PAQ model dimensions of the 27 audios from the answers of participants in Spain and in UK. This calculus included mean,median,mode,standard error of the mean (SEM), standard deviation (SDEV), variance (VAR), kurtosis (KUR), skewness (SKW) and 95% confidence interval (CI95), though discussion will only consider the statistic that turned out to be more relevant. Additionally, ISO pleasantness and ISO eventfulness were computed from 8 PAQ dimensions values as suggested by ISO 12913– 3[3] and graphed by means of the soundscapy tool [5] for each one of the 27 audios in both languages. After a previous exploration of results, main differences and similitudes were analyzed, focusing on dimensions that showed more similarity in original English and translated Spanish and, contrarily, those dimensions that were more affected by different interpretations. In that case, listening to audios also helped the analysis and drawing of conclusions. In order to strengthen the analysis, the Structural Summary Method (SSM) has been used to test the difference between English and Spanish for how much the relations between the PAQs match what would be expected. The SSM approach for circumplex data, described by J. Zimmermann et al. in [21] and sometimes called the Cosine Curve Model, has not yet been introduced into the soundscape literature and is used here for the first time. When the attributes of a circumplex model are arranged according to their order around the circumplex, the values for the attributes should follow the shape of a sinusoidal curve. The values plotted can be the mean response for a group or can be a correlation value with some external variable. Here, the attributes are ordered from 0°to 315°or, in the case of the soundscape circumplex, in the order ‘pleasant’, ‘vibrant’, ‘eventful’, ‘chaotic’, and so on, moving counterclockwise around the circumplex (see Fig. 2). A cosine curve can then be fit to the data, following the equation: D i ¼eþax cosðh i dÞ in which ‘‘D i ” is the PAQ dimension, ‘‘h i ” the corresponding angle in the circumplex as shown in Fig. 13 and ‘‘e”, ‘‘a” and ‘‘d” are model parameters. From this cosine model fit, the model parameters would describe the profile of PAQ responses: elevation ‘‘e”, amplitude ‘‘a” and displacement ‘‘d” (more information can be found in [20–21]). The goodness of fit to the expected pattern is given by the coefficient of determination R 2 . According to Girard et al. [22], if fit is less than 0.7 it is considered ‘‘unacceptable”, if fit is between 0.7 and 0.8 it is considered ‘‘adequate” and if it is>0.8 it is considered ‘‘good”. Under this new analysis method, it can be expected two primary ways to get a poor R 2 : either strong responses which do not fit the expected pattern of the circumplex (strong, contrasting, and unexpected mixed feelings), or neutral responses which effectively vary randomly around the neutral response level (that is, 3). One example of point one might be a response which is neutral to all attributes except one (e.g., calm = 5). Clearly this is a strong opinion, however it violates the expectation of the circumplex, where we would expect the opposite attribute (chaotic) to be strongly negative and the related attributes (pleasant and uneventful) to be slightly raised. Although this would clearly be a strong opinion, it would likely have a low R 2 since the shape would not match a cosine curve. An example of point two might be a sound which is generally neutral that one participant may rate calm slightly higher (=4) while another may rate only uneventful slightly higher, and another participant may rate only pleasant slightly higher and still another participant may rate monotonous slightly higher. For a mostly neutral sound, all of these are valid, but in the final analysis which of them dominates is effectively up to chance and the overall profile is basically random. There is a chance it could outline a cosine profile, but it’s unlikely. Fig. 2. The 8 PAQ coordinates and its corresponding angles for SSM analysis. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 4 3. Results 3.1. ISO 12913 Pleasantness-Eventfulness analysis Pleasantness (ISOPleasant) and Eventfulness (ISOEventful) coordinates have been computed for the Spanish and English datasets following the recommendation by ISO 12913–3 [3], as explained in [5]. This procedure converts the Likert scale responses to the 8 Perceived Affective Quality (PAQ) model dimensions into two dimensional circumplex coordinates, as shown in Fig. 3 for audio CG01 (audio ID = 1) as an example. The scatter plot of soundscape circumplex transformed according to ISO 12913 part 3 is hereafter named ‘‘ISO space”. A similar procedure carried out with the rest of audios gives a total Spanish dataset of n = 783 circumplex coordinates (783 = 27x29, as a result of 27 audios evaluated by 29 volunteers) and an English total dataset of n = 864 coordinates (864 = 27x32, as a result of 27 audios evaluated by 32 volunteers). The resulting global scatter plots for the 27 audios are shown in Fig. 4, for Spanish and English. After that, a 27-point dataset is obtained for each language by computing mean ISOPleasant and ISOEventful coordinates for each audio as shown in Fig. 5. Even though mean responses distribute in a similar way throughout the two-dimensional ISO space for the two languages, relevant differences occur for some audios. Taking into consideration absolute distances in the [P-E] space>0.2 (which represents a 10% in the range of ISOPleasant and ISOEventful coordinates variation), Fig. 5 also highlights audios for which most distant responses (Dg) are obtained. After Fig. 5, the following audios (Table 2) are highlighted because of their significative English vs Spanish differences in the perceptual responses according to the absolute distance in the diagram (Dg). Having this in mind, in order to deeper analyse and compare responses in both languages, a descriptive statistic test has been carried for each audio including most relevant parameters. If we take into consideration mean values for the 8 PAQ dimensions and associated SEM (standard error of the mean) and SDEV (standard deviation), a summary of results is shown in Table A1 in the Appendix. In this table, the 8 PAQ dimensions are named as follows: p (pleasant), ch (chaotic), v (vibrant), u (uneventful), ca (calm), a (annoying), e (eventful) and m (monotonous). From Table A1, taking into consideration differences (Dm) >0.5 in mean PAQ model dimension values (which represent a 10% in the range of variation of model dimensions) together with SEM/ SDEV values and information on which dimensions are repeatedly affected from one sound to another, a summary of selection results is shown in Table 3. From Table 3, it comes out that the audio recordings ID 2, 3, 6, 11 and 12 feature perceptual ratings with important differences in more than four PAQ model dimensions (see in Appendix detailed extracted information in Table A2). Contrarily, the audio recordings ID 1, 7, 10, 13, 15, 16, 17, 21, 23 and 26 feature the least differences in the listening experiments as only one model dimension appears as significatively different (see in Appendix detailed extracted information in Table A3). Finally, putting all these results together it is found that out of the 8 PAQ model dimensions, the most affected dimensions are concentrated around the upper-right-tolower-left part of the original Swedish Axelson’s emotional diagram [7]. That is, the greatest number of mean values differences concentrates in the [eventful – uneventful] and [vibrant – monotonous] axes of the emotional model, adding a total of 12 audios sharing uneventful (u) as one of the model dimensions affected, 11 sharing eventful (e), 9 sharing monotonous (m) and 8 sharing vibrant (v), as indicated in Table A4–A7 shown in Appendix. In order to visualise these audios in the context of the 8 PAQ model dimensions diagram, Fig. 6 shows the 12 audios selected in Table A4 (see Appendix) sharing differences in the uneventful (u) dimension, showing a clear displacement towards the eventful side of the diagram for Spanish. Similarly, Fig. 7 shows the 9 audios selected in Table A6 (see Appendix) sharing differences in the monotonous (m) dimension, showing again a clear displacement towards the eventful side of the diagram for Spanish. Finally, considering the most affected audios in this comparison as shown in Table A2 in Appendix, that is, audios ID 2, 3, 6, 11 and 22, Fig. 8 shows these audios in a similar way as before. Again, an evident displacement towards eventful and noticeable differences in the monotonous-vibrant axis is found in Spanish language with respect judgment in English auditions. 3.2. Analysis of the eight individual dimensions of the soundscape circumplex Results from ISO 12913 Eventfulness-Pleasantness circumplex analysis can be strengthened and further informed by the analysis of the eight PAQ model dimensions. To do so, mean model dimensions in Spanish and English are represented in Fig. 9 for the same audios as in Fig. 9. In these graphs, standard deviation (SDEV) and 95% confidence interval (CI 95%) are given as error bars for the mean estimations. SDEV gives information on the dispersion of data from its mean and CI95% the range of plausible means. Both confidence estimations give relevant and complementary information as to interpret how close answers and averages values are in both languages. A sinusoidal pattern can be observed in some graphs in Fig. 9 and the Structural Summary Method (SSM), previously introduced, can give some more information on this. As an example of application of this method, mean PAQ response scores are show together with SSM fitting for both languages in Fig. 10 for audio ID 17 (OS01d). SSM fitting results (model parameters and R 2 values) for audio ID 17 can be taken as a reference patter of good agreement between Spanish-English interpretation of perceived attributes, as this audio is one of the least affected audios in the comparison Fig. 3. The 8 PAQ model responses from Spanish (n = 29 volunteers) and English auditions (n = 32 volunteers) projected onto the bidimensional circumplex model for audio CG01. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 5 (see Table A3 in Appendix) As it is known, the coefficient of determination R 2 determines the variance explained by the model, that is, how well the data fit the model and R 2 values>0.8 are considered ‘‘good” [22]. Audio ID 17 SSM analysis results in of 0.97 and 0.94 R 2 values for English and Spanish fitting respectively. Taking this into consideration, SSM fitting results for the 27 audios in both languages are show in Table 4. In this table, ‘‘d” parameter is given in degrees and R 2 under 0.8 are marked in red. 4. Discussion Previous analysis has shown that soundscape responses to the 8-PAQ model can present very surprising results when comparing some of its dimensions’ Spanish-English coordinates. The main question that arises is in line with the research question itself, that is, to know if the observed differences come from a model translation into Spanish that ‘‘can be improved” or from a model originally formulated in English that ‘‘can be improved”. That is, questioning the English attributes resulting from a free translation from original Swedish after a first approximation by Axelsson et al [7]. If one of the main objectives of the SATP project is to standardize the use of the emotional model, and consequently soundscape evaluations, in as many languages as possible, these results suggest a bidirectional path for improvement in order to achieve such standardization objectives. A route that starts with the revision of the model formulation in English and, for now, continues with the Fig. 4. The 8 PAQ dimensions projected onto the bi-dimensional circumplex model according to ISO 12913–3:2019 for Spanish (left, n = 783) and English (right, n = 864) auditions. Fig. 5. Mean bidimensional circumplex model coordinates for the 27 audios. Number inside marks refers to audio ID number (from 1 to 27) Distance between responses marked with a red line (>0.4), orange line (>0.3) and green line (>0.2). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 6 revision of the model translation into Spanish. The correspondence between PAQ model main four axes in English and correspondent translation into Spanish is shown in Table 5. Having in mind audios shown in Tables A3 and A5 (see Appendix), greatest discrepancies in auditions are most evident along axes n°1 (eventful/uneventful) and n°4 (vibrant/- monotonous) of the model. A fact that can also be visualized in Fig. 9 for audios in Table A2 in Appendix, where discrepancies along axes n°1 (e/u) and n°4 (v/m) are also more significantly evident for the selected audios. If we focus on axis n°1, ‘‘eventful / uneventful”, these two dimensions translates into Spanish as ‘‘con actividad / sin actividad” literally meaning ‘‘with activity/without activity”. However, ‘‘eventful”andits opposite refer to some kind of incident within that event, some kind of significance for the evaluator of whatever is happening. The Spanish term ‘‘con actividad”, on the other hand, means just ‘‘with activity” (as in ‘‘active, dynamic” and not ‘‘stable”), a concept emptier of meaning than correspondent dimension in English. Therefore, the ‘‘eventful” dimension in original English is somehow adding an emotional significance that does not translate into Spanish. But the intention when formulating the PAQ model in this specific dimension was to evaluate whether there was some kind of activity, or not, but not if that possible activity was interpreted as good or bad, if it had any interest to the evaluator, a situation that is better described in Spanish by ‘‘con actividad” and its opposite than in English with ‘‘eventful” and its opposite. On soundtrack ID 22 (W09), for example, we hear a constant engine sound. It is not pleasant, and all respondents (Spanish and English) agree on this, as it can be seen by its ISO coordinates located at left side in Fig. 5. The difference Spanish-English comes from the eventfulness dimension, much greater in Spanish, as the pleasantness dimensions are almost the same in both languages. It seems as if the Spanish perception of eventfulness is much higher than that of the English precisely because of the perception of activity suggested by the engine. Something that is not the meaning of the term ‘‘eventful” in English, a word that adds some kind of interest, a situation that is non-existent in W09 audio so lowering eventfulness coordinates in English. Something similar occurs in audio ID 5 (E01b), with low levels close to silence with a sound similar to a pressurised water hose cleaning a street, which means activity to Spanish participants. In addition to this, in audio ID 7 (E05) distant voices, including laughter, show a state of relaxation without activity, but the sound of seagulls hints at a coastal environment and birds in flight (i.e., activity). In this case (E05) the ‘‘e” of eventful is better punctuated for Spanish-speaking participants than for English-speaking participants and the nuance of this variation again coincides with a perception of activity: in this case of the flight of seagulls. Table 3 Differences in Spanish vs English auditions expressed by means of distance in ISO space (Dg) or by PAQ model dimensions mean values increment (Dm). Marked in green, audios with just one dimension affected; yellow, best Spanish-English coincidence. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 7 As a contrast, in soundtrack ID 6 (E02) there is little difference in the perception of ‘‘eventfulness” but important discrepancy in ‘‘pleasantness”, as shown in centre Fig. 5. The distant sound of the murmur of water, coming in fact from a waterfall generated by a lock being discharged 20 m from the position of the recording, does not indicate activity, but seems to be more interesting to Fig. 6. English vs Spanish comparison of the 12 audios sharing ‘‘uneventful” (u) model dimension affection, presented by means of the soundscapy tool as proposed in [5]. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 8 English evaluators, greater pleasantness, than to Spanish. Therefore, activity is equated with ‘‘interesting” events in English and not so much in Spanish, for which ‘‘monotonous” gains ground over the other dimensions as shown in Fig. 8. Consequently, activity in context seems to be a key concept that is not well represented in the 8 PAQ model, at least not equally represented in English and Spanish terms along axes n°1. There are other cases (soundtracks) where a relationship between event and activity can be found for further analysis of Spanish-English differences. For instance, track ID 27 (W23a) refers to distant voices located, again, some 20 m from the position of the recording: few, legible and at low levels, denoting little activity as an event. Mean attributes differences are only significative (Dm > 0.5) for chaotic and uneventful dimensions but standard deviation SDEV is found similar for all dimensions resulting in a small 95% CI as shown in Fig. 11, in contrast with tracks in Fig. 9. There is an overall good agreement Spanish-English, that can also be seen by W23a coordinates in Fig. 5, indicating that English ‘‘eventful” differences with Spanish ‘‘con actividad” and their opposites almost disappear when there is really no activity and the use of both terms coincide in substance and form. Similar coincidence Spanish-English situation can be found in soundtrack ID 17 (OS01d), where an improvised drum kit playing jungle music suggest to a lot of activity as an event. The apparent activity in a nice, relevant, significative event in everyday life, seems the reason why all the respondents (Spanish and English) answered practically the same as shown by ISO coordinates in Fig. 5. In this case, only vibrant dimension results in a significative difference (Dm > 0.5) as shown in Table A3 in Appendix, being greater in English. This result points out that Spanish-English coincidence is quiet well based on a stronger evaluation of vibrant in English somehow equilibrated by a stronger evaluation as eventful in Spanish as shown in Fig. 12. The activity in the event as evaluated in Spanish is interpreted as more interesting (vibrant) in English, evidencing once again what has been said before. So, concluding this analysis, it seems that original English ‘‘eventful /uneventful” dimensions include somehow a feeling of significance, importance in a person’s life, of whatever is moving Fig. 7. English vs Spanish comparison of the 9 audios sharing ‘‘monotonous” (m) model dimension affection, presented by means of the soundscapy tool as proposed in [5]. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 9 References [1] F. Aletta et al.. ‘‘Soundscape assessment: Towards a validated translation of perceptual attributes in different languages”. Proc. 49 th Int. Congr. Noise Control Eng. INTER-NOISE 2020. Seoul. Korea. 2020. [2] International Organization for Standardization. ISO/TS 12913–2:2018 Acoustics — Soundscape — Part 2: Data collection and reporting requirements. Geneva: ISO; 2018. [3] International Organization for Standardization. ISO/TS 12913–3:2019 Acoustics — Soundscape — Part 3: Data analysis. Geneva: ISO; 2019. [4] International Organization for Standardization. ISO 12913–1:2014 Acoustics — Soundscape — Part 1: Definition and conceptual framework. Geneva: ISO; 2014. [5] Mitchell A, Aletta F, Kang J. How to analyse and represent quantitative soundscape data. JASA Express Lett 2022;2(3):. https://doi.org/10.1121/ 10.0009794037201. [6] International Organization for Standardization. (2020). ISO/PWI TS 12913-4 Acoustics — Soundscape — Part 4: Guidelines for the assessment of soundscape investigation results. ISO/TC43/SC1/WG54 Draft (unpublished). [7] Axelsson Ö, Nilsson ME, Berglund B. A principal components model of soundscape perception. J Acoust Soc Am 2010;128(5):2836–46. https://doi. org/10.1121/1.3493436. [8] ISO. ‘‘ISO/TC 43/SC 1 Noise.” https://www.iso.org/committee/48474.html. [9] Welsh Governemnt. ‘‘Noise and soundscape action plan 2018 to 2023.” p. 127. 2018. https://www.gov.wales/noise-and-soundscape-action-plan-2018-20230. [10] M. Mc Vay. ‘‘Noise and soundscape in Welsh planning policy”. Proc. 51 st Int. Congr. Noise Control Eng. INTER-NOISE 2022. Glasgow. UK. 2022. https:// az659834.vo.msecnd.net/eventsairwesteuprod/production-inconferencepublic/9adf5239317d4eb4a95cac2dca13bf1d. [11] D. T. Ecologica and D. T. Ecologica. ‘‘Ministero della Transizione Ecologica.” pp. 6–10. 2022. [12] Watcharasupat KN, Jaratjarungkiat S, Lam B, Jitwiriyanont S, Ooi K, Ong Z-T, et al. Quantitative evaluation approach for translation of perceptual soundscape attributes: initial application to the Thai Language. Appl Acoust 2022;200:108962. [13] Papadakis NM, Aletta F, Kang J, Oberman T, Mitchell A, Stavroulakis GE. Translation and cross-cultural adaptation methodology for soundscape attributes–a study with independent translation groups from English to Greek. Appl Acoust 2022;200:. https://doi.org/10.1016/j. apacoust.2022.109031109031. [14] Instituto Cervantes. ‘‘El Español: Una Lengua Viva. Informe 2022”. 2022. https://cvc.cervantes.es/lengua/espanol_lengua_viva/ [15] Almagro Pastor. J. A.. Vida Manzano. J. & García Quesada. R.. ‘‘Soundscape approach for noise management of conflict urban areas. the PTS case in Granada (Spain)”. Proc. 48 th Int. Congr. Noise Control Eng. INTER-NOISE 2019. Madrid. Spain. 2019. http://www.sea-acustica.es/fileadmin/ INTERNOISE_2019/Fchrs/Proceedings/1414.pdf. [16] Vida J, Almagro JA, García-Quesada R. The importance of changing urban scenery in the assessment of citizens’ soundscape perception. On the need for different time-related points of view. Noise Mapping 2021;8:138–61. https:// doi.org/10.1515/noise-2021-0011. [17] Vida J, Almagro JA, García-Quesada R, Aletta F, Oberman T, Mitchell A, et al. Urban Soundscape Assessment by Visually Impaired People: First Methodological Approach in Granada (Spain). Sustainability 2021;13 (24):13867. https://doi.org/10.3390/su132413867. [18] Useche. J.E.. Jahir Rodíguez. B.. Felipe Guevara. A.. Sebastián Romero. J.. Herrera Martínez. M. and Hermida. L. F.. ‘‘Mobile App for Soundscape evaluation for the Spanish speaking community validated through principal component analysis”. TECCIENCIA. 18. no. 34. 15-15. 2023. 10.18180/tecciencia.2022.34. [19] Oberman. Tin; Mitchell. Andrew; Aletta. Francesco; Almagro Pastor. José Antonio; Jambrošic ´. Kristian; Kang. ‘‘Soundscape Attributes Translation Project (SATP) Dataset.” 2022. https://zenodo.org/record/7143599#.Y6GP8HbMK00. [20] Erfanian M, Mitchell A, Aletta F, Kang J. Psychological well-being and demographic factors can mediate soundscape pleasantness and eventfulness: A large sample study. J Environ Psychol 2021;77:101660. [21] Zimmermann J, Wright AGG. Beyond description in interpersonal construct validation: methodological advances in the circumplex structural summary approach. Assessment 2017;24(1):3–23. https://doi.org/10.1177/ 1073191115621795. Table A4 The 12 audios sharing ‘‘uneventful”(u) model dimension affection. Audio ID Audio name PAQ model dimensions affected (Dm > 0.5) 2 CG04 v, u, ca, e 3 CG07 ch, v, u, ca, e, m 4 CT301 u,e 5 E01b ch, u 8 E09 v, u,a 9 E10 v, u,e 11 E12b ch, u,e,m 19 VP01b p, u,m 20 W01 u,e 22 W09 ch, u,e,m 24 W15 u,e,m 27 W23a ch, u Table A5 The 11 audios sharing ‘‘eventful”(e) model dimension affection. Audio ID Audio name PAQ model dimensions affected (Dm > 0.5) 2 CG04 v, u, ca, e 3 CG07 ch, v, u, ca, e,m 4 CT301 u, e 7 E05 e 9 E10 v, u, e 10 E11b e 11 E12b ch, u, e,m 12 HR01 v, e 20 W01 u, e 22 W09 ch, u, e,m 24 W15 u, e,m Table A6 The 9 audios sharing ‘‘monotonous”(m) model dimension affection. Audio ID Audio name PAQ model dimensions affected (Dm > 0.5) 3 CG07 ch, v, u, ca, e, m 6 E02 p, ch, ca, a, m 11 E12b ch, u, e, m 13 KT01 m 18 RPJ01 v, m 19 VP01b p, u, m 22 W09 ch, u, e, m 24 W15 u, e, m 25 W16 ca, m Table A7 The 8 audios sharing ‘‘vibrant”(v) model dimension affection. Audio ID Audio name PAQ model dimensions affected (Dm > 0.5) 2 CG04 v, u, ca, e 3 CG07 ch, v, u, ca, e, m 8 E09 v,u,a 9 E10 v,u,e 12 HR01 v,e 16 OS01c v 17 OS01d v 18 RPJ01 v,m Table A2 Most affected audios in SPA-ENG comparison according to PAQ model dimensions. Audio ID Audio name PAQ model dimensions affected (Dm > 0.5) 2 CG04 v, u, ca, e 3 CG07 ch, v, u, ca, e, m 6 E02 p, ch, ca, a, m 11 E12b ch, u, e, m 22 W09 ch, u, e, m Table A3 Least affected audios in SPA-ENG comparison according to PAQ model dimensions. Audio ID Audio name PAQ model dimension affected (Dm > 0.5) 1 CG01 ca 7 E05 e 10 E11b e 13 KT01 m 15 N1 ca 16 OS01c v 17 OS01d v 21 W06 p 23 W11a ca 26 W22 a J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 16 [22] Girard J. Zimmermann J. Wright A (2023). circumplex: Analysis and Visualization of Circular Data. https://github.com/jmgirard/circumplex. http://circumplex.jmgirard.com/. [23] Ö. Axelsson. P. Lundén and M. E. Nilsson. ‘‘Sound Cities: Computational modelling of urban soundscape quality”. Proc. 42th Int. Congr. Noise Control Eng. INTER-NOISE 2013. Innsbruck. Austria. 2013. http://www. internoise2013.com/images/onlineversion_internoise_2013.pdf. [24] Herranz-Pascual K.. Iraurgi I.. Aspuru I.. Garcia-Pérez. I.. Eguiguren J.L. and Santander. A.. ‘‘Development of the Acoustic Comfort Assessment Scale (ACAS12): Psychometric properties. validity evidence and back-translation between Spanish and English”. PLoS ONE 18(2): e0281534. 10.1371/journal. pone.0281534. J. Vida, José Antonio Almagro, R. García-Quesada et al. Applied Acoustics 211 (2023) 109516 17