A field experiment on the effect of sound absorption installed to a highly reverberant kindergarten classroom Keiji Kawai 1 Midori Ishizawa Yuuki Matsufuji Kumamoto University 2-39-1 Kurokami Kumamoto-city Japan ABSTRACT Japanese national or local governments have not published any standard for acoustic performance of classrooms and this is reflected by the rare consideration of sound absorption in classrooms. A kindergarten classroom, targeted in this study, was found to be highly reverberant (V = 200 m 3 , RT = 1.0 s at 1 kHz) because the room was originally an outdoor space of ground floor pilotis and converted into a classroom by constructing concrete walls between the columns with the fire-resistant hard ceiling panels remained. The author proposed a temporary installation of sound absorbing ma- terial and to conduct a field experiment to examine and experience the effect of sound absorption. Sound absorbers were attached on the ceiling and the walls to reduce the RT to 0.5 s. By this, the slope of doubled distance attenuation increased from 1.5 dB to 2.3 dB, which should mitigate the confusion of speeches by children in the room. The result of sound level monitoring indicated 6-7 dB decrease at 1-2 kHz in the absorbed condition, which is greater than the 3.8 dB decrease measured in the unoccupied condition using a loudspeaker source. Also, the children’s utterances of lunch prayer were more synchronized in the absorbed condition. 1. INTRODUCTION

Reverberation is a factor that can greatly hinder the vocal communication of young children, as stated in [1] as an example: For Early Childhood Facilities (children from birth- age 5) the reverberant acoustic conditions should be optimal, as children are involved in listening and language skill devel- opment during this stage of their life and need the best acoustic conditions possible . However, at present in Japan, national or local governments have not published any standard for reverberation time of classrooms and this is reflected by the rare consideration of sound absorption in classrooms of school and pre-school facilities. During our site inspection survey on the acoustics of kindergarten buildings in Kumamoto city, a highly reverberant classroom happened to be found. Among classrooms in the kindergarten, only this room is a refurbished room from the original outdoor piloti, with highly reflective surfaces such as the original fireproof cement board ceiling remained and concrete walls additionally constructed be- tween columns. The teachers recognized that the room tend to be noisier than others but did not recognized that the bustling was due to the excessive reverberation. At the time of visit, 3-year-old children were freely playing in the classroom and everyone seemed to be screaming. Thus, the authors

1 kkawai@kumamoto-u.ac.jp

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

proposed a temporary installation of sound absorbing material and to conduct a field experiment to examine and experience the effect of sound absorption. 2. OVERVIEW OF EXPERIMENT

The room for the experiment is for one of the two 3-year-old classes with 25-30 children for each. The dimensions of the room are 6.2 m x 10.8 m x 2.5 m (S = 250m 2 , V = 200m 3 ) (Figure 1). The interior materials of the room are concrete for the walls, glass windows, cement boards for the ceiling, and timber floor, which are totally highly reflective. The furniture is plastic chairs and tables and a piano. A curtain attached to a glass window is only sound absorbing material in the room. The rever- beration time measured in this room and the average sound absorption coefficient obtained by Eyring’s equation are shown in Table 1. The absorption coefficient, 0.12, can be regarded as highly reverberant as a classroom with furniture.

10800

h= 2.5 m V= 200 m³ S= 250 m² S1(h=1.2m)

6200

S2(h=0.4m)

R1(h=2.2m) R2(h=2.2m)

absorber

Figure 1: Room dimensions and source/receiving positions

Figure 2: Sound absorber installation

Table 1: Reverberation time and average absorption coefficient

Reverberation time [s] Absorption coefficient Oct. band [Hz] Original Absorbed Original Absorbed 250 1.0 0.9 0.12 0.14 500 1.0 0.6 0.12 0.18 1 k 1.0 0.5 0.12 0.23 2 k 1.0 0.5 0.12 0.23 4 k 0.9 0.5 0.13 0.24

Table 2: Outline of experimental period

Period Date Absorption Recording day N of children N1 Dec.6-Dec.10 no 5 27-30 A1 Dec.13-Jan.21 yes 6 26-30 Jan.24-Feb.14 no - 16-19 A2 Feb.15-Mar.4 yes 6 23-26 N2 Mar.7-Mar.22 no 7 24-28

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

Sound absorbers were temporarily installed in this room with an average sound absorption coeffi- cient of 0.25 as a target, that is, 163 polyester fiber boards with a size of 170 x 400 mm and a thickness of 50 mm were attached to the ceiling with magnets, and 12 m 2 of polypropylene non-woven fabric with a thickness of 4 mm was hung on two adjacent walls (Figure 2). The experiment was carried out from December 2021 to March 2022, during which the sound absorbers were attached and detached twice. Table 2 shows the outline of the experimental periods. Due to the spread of infection of COVID-19, a request to refrain from going to the kindergarten was issued after A1 period. The re- quest continued until March, but gradually the number of children coming to kindergarten increased and the recording restarted in the middle of February. Acoustic measurements were performed in the condition with and without the absorbers, and room sounds were recorded for a total of 24 days. 3. IMPROVEMENT IN ROOM ACOUSTICS

The improvement of room acoustic conditions due to the sound absorption was examined by acoustic measurement. In the actual measurement, using an omnidirectional speaker as the sound source, the reverberation time and STI (S/N ratio set at 15 dB in every octave band) were measured by impulse response method, and the distance attenuation from the sound source was measured using stationary pink noise. The reverberation time was reduced to about half around 0.5 s, which was a desirable value for a room with V = 200 m 3 in reference to DIN 18041 (0.55 s). Figure 3(1) shows the relationship between the distance from the sound source and STI. The STI values in the original conditions were 0.65 at 1 m point and 0.60 or less (fair) at 3 m point or farther, and improved to 0.7 or more in the entire room due to sound absorption. Figure 3(2) shows the distance attenuation from the sound source. The sum of two octave band levels of 1 and 2 kHz is selected as the representative voice level of the children that was obtained from an analysis of the recorded sound, and is used in this paper hereafter. In the original condition, the distance attenuation from 1 m point to 8 m point was 4.4 dB and this increased to 7.0 dB due to sound absorption, which will increase the S/N ratio of face-to-face conversation. 4. EFFECT OF SOUND ABSORPTION ON CHILDREN'S ACTIVITIES

4.1. Recording and analysis Based on the acoustic improvements described above, their effect on the children was investigated through daily recordings. Microphones of sound level meter were hung at 30 cm from the ceiling at

0.8

Absorbed

0

Relative SPL 1-2 kHz [dB]

excellent

Original

-2

0.7

STI

good

-4

0.6

-6

fair

Absorbed

Original

-8

0.5

0.5 1 2 4 8

0 2 4 6 8 Distance (m)

Distance (m)

(2) Distance attenuation

(1) STI

Figure 3: Acoustical improvement in STI and distance attenuation

two points (R1 and R2) as indicated in Figure 1, and the room sounds were recorded with a PCM recorder during the daily activity time (9:00 to 14:15). First, the reduction in SPL due to the installation of sound absorbers was measured when the room was unoccupied. Assuming typical positions of the teacher and children (S1 and S2), omni-directional speakers were set at 1.5 m (S1) and 0.6 m (S2) from the floor. Then pink noise was generated and recorded with two microphones in the conditions with and without sound absorption. The results are shown in Table 3. The farther point (R2) from the sources had greater attenuation due to sound ab- sorption than the closer point (R1) because direct sound from the sound source, which is not affected by sound absorption, is distance attenuated. This level difference is the reference value when discuss- ing SPL during activity. In the analysis of the SPL, the L eq , 1sec of the sounds recorded by the two microphones were calcu- lated and the smaller of the two levels of at the same time was taken as the representative value, which was compared with the level attenuation of the microphone farther from the child position (3.8 dB, sum of 1 and 2 kHz octave band levels of S2-R2 position in Table 3). The rationale for this is that the smaller SPLs of the two simultaneous recordings can be considered to indicate that the microphone located at a greater distance recorded it. Another item of analysis was to examine the degree to which the children's utterances of lunch prayer were synchronized. At this kindergarten, before the meal, the children say a prayer to thank God for the meal. Due to the confusion of voices caused by reverberation, it may be difficult for the children to synchronize themselves with the voices of those around them, and we conducted a sub- jective evaluation experiment focusing on this point.

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

Table 3: Reduction of SPL by installing absorbers at recording microphone positions

Source – receiving point Distance [m] Level reduction [dB] for each octave band [Hz]

250 500 1k 2k 1k+2k

S1-R1 2.0 0.8 2.6 3.1 2.1 2.5

85 80 70

S1-R2 6.4 2.0 3.6 3.7 4.0 3.8

S2-R1 2.4 1.0 1.3 1.7 2.3 1.9

S2-R2 4.7 1.8 2.7 3.5 4.2 3.8

90

N1 A1 N2 A2

85

80

SPL 1-2 kHz [dB]

L eq L 5

75

L 50

70

L 95

65

60

Day

Figure 4: Daily SPL in free play time after lunch

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

Table 4: SPL of 1-2 kHz octave band in 10 minutes free play time arithmetically averaged over days of each period

Period L eq [dB] L 5 L 50 L 95 N1 80.0 (0)* 84.9 (0) 78.0 (0) 70.3 (0) A1 73.2 (-6.8) 77.9 (-7.0) 71.5 (-6.5) 64.7 (-5.7)

A2 73.6 (-6.4) 78.8 (-6.1) 70.9 (-7.1) 63.6 (-6.7) N2 77.5 (-2.5) 82.1 (-2.8) 75.7 (-2.3) 69.2 (-1.1) *Values in () are relative levels (  L ) with N1 = 0 dB.  L = -3.8 dB is the expected reduction by sound absorber (see Table 2, S2-R2).

50

Time percentage [%]

40

N1 A1 A2 N2

30

20

10

0

55-60 60-65 65-70 70-75 75-80 80-85 85-90 90-95

L eq, 1-2 kHz, 1sec. [dB]

Figure 5: Comparison of time distribution of SPL by absorption periods

4.2. Changes in Sound Pressure Levels Since the activities in the kindergarten vary depending on the day, the free play time after meals, which is usually the same every day, was targeted, and the time period of 10 minutes after the greeting of the end of lunch was analyzed. The daily SPLs are shown in Figure 4, and the distribution of the L eq , 1sec for the entire observation days of each experimental period is shown in Figure 5. In the original N1 period, L eq was high at around 80 dB and L 95 was around 70 dB, indicating a very noisy condition (Figure 4). When sound absorption was installed in A1, the SPL was immedi- ately reduced by 7-8 dB. When the sound absorber was again installed 25 days after A1, during A2, the SPLs were similar to A1. Then, in N2, where it was removed again, the level returned to the level of N1 on the third day. On the last day of A2, the atmosphere was bustling and there were frequent shouts, resulting in a high L eq . On the last day of N2, the atmosphere was generally gloomy. Despite these variations, as shown in Table 4, an average reduction of 6-7 dB was observed in the sound-absorbing conditions. The actual reduction was greater than the 3.8 dB reduction with the loudspeaker as the source, which indicates that the voices became quieter.

4.3 Synchronicity of prayer voices. The lunch prayer lasts approximately 30 seconds, and the 23-30 children, in total, utter the phrases of the prayer to thank God for the blessings of the meal. Two of these phrases were selected for the analysis: (“ ARIGATO GOZAIMASU ,” meaning “We thank you (thee)...”: Phrase “Thank” ), which is

21-24 AUGUST SCOTTISH EVENT CAMPUS ? 02 ? GLASGOW

50

Phrase: Thank Phrase: Name

40

30

[%]

A1+A2 N1+N2

20

10

0

Not at all

moder- ately

Not at all

moder- ately

Very a little

Very a little

Figure 6: Subjective evaluation of asynchronicity in children’s uttering lunch player

often spoken in daily life, and " KAMISAMA NO NAWO TOSHITE ,” meaning “In Jesus' name,” Phrase “Name” ), which is not often spoken. A simple experiment was conducted in which subjects were asked to judge whether the voices of the children saying the prayer were synchronized. In total, 48 sound pieces, which are the two phrases from recorded prayer voices for a total of 24 days of four periods, were randomly played back. Participants were seven students and they listened to them through headphones and rated their asynchronicity on a 4-point scale. The results are shown in Figure 6. Utterances were rated as more synchronous in the absorbed condition and for the phrase Thank that is familiar to children. In the original condition without sound absorption, more than one-third were rated as moderately or very asynchronous, indicating that re- verberation made them more difficult to synchronize with the other children. An analysis of variance with the 4-point scale assumed as an interval scale showed that the differ- ence was significant for both phrases (p<5%). The level of utterances measured was around 70-80 dB, which did not correlate with the ratings. 5. Concluding Remarks

In this experiment, we examined how children are affected by a poor acoustic environment from two aspects: changes in sound pressure levels and synchronization of children's utterances. The results of high noise levels and reduced synchronization of children's voices in conditions without sound ab- sorption indicated an inappropriate environment for growing and learning of children. In the current situation in Japan, where acoustic design is not widely recognized, it is necessary to verify and promote society's understanding of what a poor acoustic environment could cause on chil- dren and teachers, and hence this is the purpose of this study. We would like to study the effect of a good acoustic environment in the near future after the acoustic design is widely spread. 6. ACKNOWLEDGEMENTS

We wish to thank the children and teachers of Ouei Kindergarten for their cooperation in this exper- iment. This work was supported by JSPS KAKENHI Grant Number 19H01706. 7. REFERENCES

1. Department for Education, Government of South Australia, “Early childhood facilities (birth to age 8) design standards and guidelines”, p.33 (2016).