Int Arch Occup Environ Health DOI 10.1007/s00420-014-0927-8

ORIGINAL ARTICLE

A study of classroom acoustics and school teachers’ noise exposure, voice load and speaking time during teaching, and the effects on vocal and mental fatigue development Jesper Kristiansen • Søren Peter Lund Roger Persson • Hitomi Shibuya • Per Møberg Nielsen • Matthias Scholz

•

Received: 26 June 2013 / Accepted: 16 January 2014  Springer-Verlag Berlin Heidelberg 2014

Abstract Objectives The study investigated the noise exposure in a group of Danish school teachers. The aims were to investigate if noise posed a risk of impairment of hearing and to study the association between classroom acoustical conditions, noise exposure, vocal symptoms, and cognitive fatigue. Methods Background noise levels, vocal load and speaking time were measured on 35 teachers during actual classroom teaching. The classrooms were characterized acoustically by measurements of reverberation time. Before and after the workday, the teachers answered a questionnaire on fatigue symptoms and carried out two cognitive test tasks sensitive to mental fatigue. Results The average noise level during the lessons was 72 dB(A), but during indoor sports activities the average noise level increased 6.6 dB(A). Room reverberation time (range 0.39–0.83 s) had no significant effect on the noise level. The teachers were talking with a raised voice in 61 % of the time, and the vocal load increased 0.65 dB(A) per dB(A) increase in the average lesson noise level. An increase in voice symptoms during the workday correlated significantly with individual average noise exposure, and a J. Kristiansen (&)
S. P. Lund
R. Persson
H. Shibuya The National Research Centre for the Working Environment, Lersø Parkalle 105, 2100 Copenhagen, Denmark e-mail: [email protected] R. Persson Department of Psychology, Lund University, Lund, Sweden P. M. Nielsen Akustik Aps, 2500 Valby, Denmark M. Scholz Bru¨el & Kjær Sound and Vibration A/S, 2850 Nærum, Denmark

decrease in performance in the two-back test correlated significantly with individual average vocal load. Conclusions Noise exposure in general classrooms posed no risk of noise-induced hearing impairment in school teachers. However, the results provide evidence for an association between noise exposure and vocal load and development of vocal symptoms and cognitive fatigue after work. Keywords Reverberation time
Classroom noise
Noise effects
Voice symptoms
Cognitive fatigue

Introduction In non-industrial professions like teaching, noise is presumed to be primarily a nuisance rather than a risk factor for noise-induced hearing loss. There is indications that noise exposure, even of low intensity, is associated with increased sickness absence (Clausen et al. 2009, 2013). Noise in classrooms has the potential to interrupt ongoing activities and to disturb the perception of speech (Picard and Bradley 2001; Hodgson and Nosal 2002; Shield and Dockrell 2003) and cognitive processes (Kjellberg et al. 2008; Ljung et al. 2009). Indeed, when compared to other occupational groups including industrial professions, Danish school teachers report a high level of exposure to ‘‘noise that disturb their work activities’’ (Kristiansen 2010). In the most recent national survey, 59 % of the teachers reported to be exposed to disturbing noise in at least  of the working time. The average across all occupational groups was 42 %. With regard to teachers, this represents an increase from the 51 % that was observed in the national survey in 2005. The fact that teachers report that the primary source of annoying noise is classroom activities (Kristiansen et al.

123

Int Arch Occup Environ Health

2011) sets an interesting dilemma. On the one hand, noise from classroom activities can be viewed as both inevitable and desirable because it reflects active participation from the pupils.1 Moreover, this ‘‘activity noise’’ is in principle controllable by teacher interventions, (e.g., pedagogical methods, enforcement of rules for behavior in the classroom, etc.). On the other hand, the relatively high proportion of school teachers who report that classroom noise is a nuisance indicates that the working conditions for teaching are not always controllable and optimal as regards sound levels. In line with this, it should be emphasized that satisfactory acoustical condition is a prerequisite for teaching because the room acoustics should not work against the teaching activities. Since very little is known about noise levels and the consequences of noise exposure among school teachers, the purpose of this study was to characterize the noise exposure, vocal loading and speaking time of Danish school teachers during teaching and to investigate to what extent these measures were related to the classroom acoustical conditions. Secondly, we wanted to explore whether the observed noise exposure, vocal loading or speaking time was associated with development of voice symptoms and mental fatigue during the workday. The model to guide our analyses is as follows: First, room acoustical conditions influences the classroom noise levels because reflected sounds will mask sound signals from speakers and therefore impair speech intelligibility and make speakers speak louder (Nijs et al. 2008; Whitlock et al. 2006). The next step in the model is the contention that teachers increase their vocal loads proportionally with the noise. This so-called Lombard effect has been demonstrated in both laboratory setting and in real teaching situations (Whitlock et al. 2006; Sato and Bradley 2004), and we expect that the measurements in the present study will confirm this association. Finally, being exposed to noise and a high vocal load are physically and mentally taxing (Hughes and Jones 2003; Banbury et al. 2001; Sandrock et al. 2009), and we therefore expect that the measured noise and vocal load are positively associated with subjective feelings of fatigue, more stress, feeling of lower energy, poorer performance in cognitive tasks and an increase in voice symptoms during the workday.

Study population Teachers and schools In total, 35 teachers from seven schools participated. The group comprised 28 women (age 44 years, range 25–61) 1

Noise is generally defined as unwanted sound. In this study, all sounds from the teacher’s surroundings are defined as noise, although it is clear that not all activity sounds are noise from the teacher’s perspective.

123

and 7 men (age 42 years, range 30–57 years). The study population from which teachers and schools were drawn has been described in two previous paper (Kristiansen et al. 2011, 2013). In brief, ten schools employing 419 teachers in the Copenhagen area were invited to participate in a questionnaire survey. The respondents (283 teachers) were subsequently invited to participate in a study at the National Research Center for the Working Environment that included assessment of the social climate at work, audiometric measurements and cognitive testing. The participants in this study (a total of 104 teachers) were next invited to take part in the field study reported here. They were informed about the purpose of the study, and a total of 40 teachers consented to participate, and noise measurements were made for 37 of these (for practical reasons, at least two teachers from a school was required to take part in the measurements. Three teachers were not included because of this limitation). Technical problems in the noise exposure measurements reduced the number of subjects to 35. Ethical considerations We notified and registered this study with the Danish Data Protection Agency. According to Danish law, only research using biological material needs approval from the Danish National Committee on Biomedical Research Ethics.

Methods Study design and procedure All measurements took place on schools in Copenhagen during normal work days. The teachers reported to the researchers approximately at 8:00 in the morning on the day of measurement. They filled in a short questionnaire on stress and energy levels, as well as on fatigue symptoms (see below). Next, the teacher performed computerized tasks aimed at testing attention and short-term memory functioning (described below). Six teachers on one school were invited to participate only with noise measurements and questionnaire and did therefore not perform the computerized tasks. Noise measurement equipment was mounted after completion of the cognitive tasks (described below). To avoid interference with the normal teaching routines, the teachers were mostly left unsupervised, but checked visually several times through window panes in the classroom door. After teaching, each teacher was interviewed in order to identify any deviations from a normal workday. After this debriefing and removal of the measurement equipment, the teacher again filled in a questionnaire on fatigue symptoms and repeated the computerized cognitive work tasks. Debriefing and testing took place when the teachers had finished teaching and other

Int Arch Occup Environ Health

activities (such as meetings) at the school, which differed between teachers. Twenty-seven teachers finished between 12:30 and 14:10, and six teachers finished between 11.15 and 11:25. On the day of the measurements, the teachers were delivering an average of 4.1 lessons (range 2–6, standard deviation 1.0). One teacher delivered two lessons, eight teachers delivered three lessons, and 26 teachers delivered four or more lessons.

activity. In addition, for each lecture, three equivalent sound pressure level values were determined as follows: 1.

2.

Measurement of reverberation time Reverberation time was measured by impulse excitation and reverse integration of the impulse response and was based on assessments of T30 from impulse decay, as described in ISO 3382-2 under the integrated impulse response method (International Organization for Standardization, 2008). The impulse was generated by the sudden release of pressurized air, and the response was measured by a Bru¨el & Kjær 2260 sound level meter and analyzed in octave bands from 125 to 8,000 Hz using a Bru¨el & Kjær 7830 Qualifier. The measurements were based on 2 different sound source positions at three different microphone positions in classrooms without pupils. Noise exposure measurements Noise exposure assessment was made according to the principles in ISO 9612, following the requirements for measurements with personal noise dose meters. The dose meters, Bru¨el & Kjær Type 4445, were set up to measure the A-weighted equivalent sound pressure level LAeq and to store results in profiles at 1-s intervals. The instruments were calibrated daily before and after the measurements. The microphone was mounted on the teacher’s shoulder approximately 10 cm from the ear and 4 cm above the shoulder. To register voice activity, a throat microphone was attached to the subject’s neck, close to the larynx. The signal from this sensor was continuously recorded as wave files using Bru¨el & Kjær’s SoNoScout recorder. To discriminate between periods with and without voice activity, the signal was processed as follows: The energy in the high and low band-pass filtered signal was determined within time blocks of 100 ms length. Prior to the field measurements, an experiment had been conducted to determine a threshold value for voice activity. Each second, where the energy in at least one of these blocks exceeded this threshold, was marked to contain ‘‘voice activity.’’ The first result of this analysis was a measure for the speaking time for each lesson (ST) expressed as the percentage of voice

3.

Considering only those sections of the profile where the teacher did not speak gives a measure for the equivalent A-weighted sound pressure level of the average ambient noise created by the class or other sound sources in the room, LAeq,ambient. Considering those sections of the profile where the teacher was speaking gives a measure for the equivalent A-weighted sound pressure level due to the teacher’s voice during voice activity, that is, the vocal load, LAeq,vocal. Considering the full profile for the lecture (i.e., the full 45 min) is the combination of result 1 and 2 and equal to the equivalent A-weighted sound pressure level due to both ambient noise and the teacher’s voice, LAeq,shoulder.

Self-reported stress and energy Stress and energy levels were monitored by the stress and energy inventory (SEI), which is an adjective checklist that assess self-reported arousal in correspondence with the circumplex model of affect (Russell 1980; Posner et al. 2005). SEI has been developed and validated by Kjellberg and Iwanowski (1989), Kjellberg and Wadman (2002) and has been shown sensitive enough to detect systematic variations within workdays, workweeks and across seasons (Persson et al. 2003, 2010). The directions of the items were the higher score the more perceived stress and higher perceived energy (Chronbach’s alpha varied between 0.78 and 0.88). Voice symptoms One question assessed voice symptoms. The question read ‘‘How are you right now? Do you feel hoarseness or fatigue in your voice?’’ The response categories were ‘‘Not at all,’’ ‘‘Very little,’’ ‘‘A little,’’ ‘‘Somewhat,’’ ‘‘Considerably,’’ and ‘‘Very much,’’ which were scored 1–6 in the statistical analysis. Mental fatigue One question assessed mental fatigue. The question read ‘‘How are you right now? Do you feel tired in your head?’’ with the same response categories and scoring as the voice question above. A question with similar wording has previously been able to discriminate between the effects of different cognitive tasks as well as between the effects of working on office background noise versus quiet conditions (Kristiansen et al. 2008). Higher score indicates more intense fatigue symptoms.

123

Int Arch Occup Environ Health

Exertion One question assessed the teachers’ degree of exertion during the workday. This item was responded to in the afternoon and read ‘‘How was the working day?’’ The response categories were ‘‘Not strenuous at all,’’ ‘‘a bit strenuous,’’ ‘‘somewhat strenuous,’’ ‘‘very strenuous,’’ and ‘‘extremely strenuous.’’ The responses were rated 1–6 in the statistical analysis. Cognitive tests Mental fatigue is thought to compromise executive control of behavior. In this context, executive control refers to the ability to regulate other cognitive processes in a goaldirected manner, allowing the individual to respond adaptively to novel or changing task demands (van der Linden et al. 2003). Deterioration of executive control impairs the ability to inhibit learned responses and the ability to update and monitor the working memory content (Miyake et al. 2000). The sustained-attention-to-response test (SART) assess the participants’ ability to inhibit responses (van der Linden et al. 2005). Specifically, the numbers 1–9 (stimulus) are presented 50 times each in pseudorandom order, that is, a total of 450 trials. The task is to press the response key (the keyboard bar) as fast as possible in reaction to any number but avoid doing so when the number 3 is displayed. The ‘‘no-go’’ trials constitute 11 % of the 450 trials. The stimulus lasted 500 ms, and the interstimulus interval was constant 1,000 ms. Outcome measures were the number of inhibition errors (i.e., pressing the response key in a no-go trial). The two-back test (TBT) was used to assess aspects of the participants’ working memory function (Bailey et al. 2007). In the TBT test, letters (consonants) are presented for the participants in pseudorandom order. The task is to press the ‘‘Yes’’ response key when the target letter was preceded by the same letter 2 trials earlier (that is, the trials with the target letter and the letter to compare it with are disjoined by a ‘‘distractor’’ letter). The stimulus was presented for 1,000 ms, and the interstimulus interval was constant 1,500 ms. The total number of trials (target letters) was 66, in which there were 24 matches dispersed randomly among 42 non-matches. The outcome measure of the TBT was the percentage of correct responses.

children’s age and physical room characteristics (volume and reverberation time) on LAeq,ambient, LAeq,vocal and ST were investigated in repeated measures mixed models with the teacher as random intercept. Because of the limited number of subjects, only main effects were investigated. All factors and covariates were entered in the statistical model and eliminated stepwise conditioned of P [ 0.2. A two-tailed P value B0.05 was considered statistically significant. To investigate the relation between noise and voice measures and change in fatigue during a workday, the individual mean exposure to ambient noise (LAeq,ambient,day), the individual mean vocal load (LAeq,vocal,day) and the individual mean speaking time during lessons (STday) were calculated. Associations between daily exposure metrics and fatigue were expressed by Spearman’s rank correlations. Since our a priori hypothesis is a positive association between noise and voice measures and an increase in fatigue indices, we considered a one-tailed probability of the null hypothesis P B 0.05 as indication of statistical significance. To control for the possible confounding effect of classroom type and school subject, the rooms were divided into three categories: gymnasiums (used for the school subject: Sports), classrooms for special subjects (with lessons in Science, Home economics) and general classrooms (Danish, Math, Social subjects, English, German as well as kindergarten class). Statistical analyses of associations were conducted in a stepwise manner: first, including all types of rooms (and school subjects), second, excluding gymnasium, and finally, excluding both gymnasiums and classrooms for special subjects.

Results Classroom volumes and reverberation times General classrooms were all less than 250 m3 of volume, and RT ranged from 0.39 to 0.72 s (Table 1). Only a limited number of classrooms for special subjects and gymnasiums were included in this study. In general, classrooms for special subject were somewhat larger and the RT was also longer (Table 1). The three gymnasiums were of course much larger in volume and also had the longest RT.

Statistical analysis

Noise exposure

Ambient noise, vocal load and speaking time were approximately normal distributed and entered untransformed in the statistical models. The influence of teacher’s sex and age, school (dummy coded), school subject,

School subject had (repeated measures that noise exposure than during other

123

a significant effect on noise exposure ANOVA). Post hoc analysis showed during Sports was significantly higher school subjects (Table 2). With the

Int Arch Occup Environ Health Table 1 Volume and reverberation time in classrooms by classroom type

Mean and range

General classrooms (n = 32)

Special subjects classrooms (n = 5)

Gymnasiums (n = 3) 1,020 (969–1,055)

Volume (m3)

173 (145–243)

326 (229–525)

RT (s) (125–8 kHz)

0.54 (0.39–0.72)

0.67 (0.47–0.83)

1.13 (0.95–1.30)

RT (s) (125–500 Hz)

0.63 (0.40–0.87)

0.70 (0.50–0.86)

1.13 (0.95–1.29)

RT (s) (1–2 kHz)

0.48 (0.35–0.67)

0.69 (0.50–0.87)

1.29 (1.17–1.47)

RT (s) (4–8 kHz)

0.45 (0.36–0.62)

0.60 (0.42–0.75)

0.95 (0.75–1.15)

exception of Sports, the mean noise exposure during a lesson was 61.8–81.8 dB(A) (n = 76) with an overall mean for all school subjects of 71.6 dB(A) and 75 and 90 % percentiles 74.4 and 77.0 dB(A), respectively. During Sports, the mean noise exposure was 78 dB(A) and the range 73.9–83.0 dB(A), which was 6.6 dB(A) higher on average compared to other school subjects (SD = 1.2 dB(A), P \ 0.001, t test). There were no significant differences between the schools with respect to noise exposure. Moreover, the pupils’ age, the teachers’ sex or age or the room volume also had no significant effect on noise exposure.

The mean speaking time for different school subjects varied between 49 and 71 % (excluding Sports) and was 41 % for Sports. There was no significant influence of school subject on speaking time. There was considerable variation in speaking time within each school subject (Table 2). Effect of reverberation time The association between RT and noise exposure, vocal load and speaking time was investigated in mixed models. Overall, there was no significant effect of the mean RT (all frequencies) on noise, vocal load or speaking time.

Vocal load and speaking time Effect of ambient noise on vocal load The teachers’ vocal load varied from 78 to 83 dB(A) (excluding Sports), and it was around 90 dB(A) during Sports (see Table 2). The mean difference between Sports and other school subjects was 8.6 dB(A) (t test, SD = 1.5 dB(A), P \ 0.001). The ISO standard 9921 contains criteria to characterize the loudness of the human voice. The criteria are based on sound pressure levels measured at 1 m distance. We extrapolated the criteria to 0.2 m, which is the approximate distance between the microphone and the teacher’s mouth in this study. Since we had to ignore directional effects, the extrapolated criteria at 0.2 m at the shoulder position are probably slightly overestimated, leading to a underestimation of the number of measurements that exceeds the criteria. The results are presented in Table 3. In 61 % of the lessons, the measured vocal loads were equivalent to talking with a raised or a loud voice. In 39 % of the lessons, the measured vocal loads were equivalent with normal or lower than normal voice. With regard to Sports, all measurements correspond to raised, loud or very loud voice. The pupils’ age, teachers’ sex or the volume of the room had no significant influence on voice measurements. The linear association between vocal load and the teacher’s age was nearly significant: LAeq,vocal decreased by 0.13 dB(A) per year of age (mixed model, SD = 0.07 dB(A)/year, F = 3.3, P = 0.079).

Not surprisingly, the vocal load showed a significant linear association with noise exposure that remained significant after adjusting for gender and age of the teachers and limiting the analysis to teaching in general classrooms. The effect of noise on vocal load was estimated to 0.65 dB(A) per dB(A) noise (SD = 0.12 dB(A)/dB(A), F = 5.4, P \ 0.001). The teachers’ vocal load was 10.1 dB(A) (SD = 3.3 dB(A), N = 75) higher than the noise exposure in general and special subjects classrooms and 11.7 dB(A) (SD = 3.8 dB(A), N = 12) higher during Sports lessons in gymnasiums. Daily noise exposure and daily vocal load The average daily noise exposure for all teachers was 73.0 dB(A) (range 65.5–81.2 dB(A), n = 34) (Fig. 1a). When excluding teachers that had Sports during the day, the average was 72.2 dB(A) (range 65.5–78.7 dB(A), n = 28) and 75 and 90 % percentiles 75.1 and 76.0 dB(A), respectively. The distribution characteristics for the daily vocal load were mean 80.0 dB(A) (range 71.2–93.9 dB(A), n = 34) (Fig. 1b) and mean 79.5 dB(A) (range 71.2–86.0 dB(A), n = 28) and 75 and 90 % percentiles 82.5 and 84.6 dB(A) when excluding teachers that have had Sports during the day.

123

40.7 (16.4)

17.9–75.8 (12) 40.1–53.5 (4)

49.0 (6.0) 71.1 (14.9)

53.0–83.7 (4) 34.3–93.2 (13)

62.5 (15.0) 50.6 (17.6)

24.4–74.2 (7)

90.0 (5.3)

81.8–98.9 (12) 76.4–79.6 (4)

78.3 (1.4) 82.5 (3.0)

80.1–86.6 (4) 75.4–90.1 (13)

82.1 (4.5) 83.1 (4.4)

73.9–86.6 (7)

78.3 (2.8)

73.9–83.0 (12) 65.2–74.0 (4)

70.9 (4.0) 71.7 (3.5)

67.7–74.9 (4) 66.6–81.8 (13)

72.6 (4.1) 74.8 (4.6)

66.5–81.7 (7)

Math

17.1–94.0 (32)

57.0 (18.6) 50.8 (14.1)

25.4–72.8 (16) Min–Max (n)

Mean (SD) Speaking time ST (%)

65.0–91.2 (32)

81.4 (5.6) 80.5 (3.8)

73.4–87.5 (16) Min–Max (n)

Mean (SD) Vocal load LAeq,voice [dB(A)]

71.0 (4.3)

62.8–79.7 (32) 64.4–76.9 (16)

71.0 (3.5) Mean (SD)

Ambient noise exposure LAeq,ambient [dB(A)]

Kindergarten class

Danish and social subjects

English and German

123

Effects of daily noise exposure and vocal load on change in fatigue symptoms The change in voice symptoms over a workday was significantly correlated with the teacher’s average noise exposure during teaching (Spearman’s rho = 0.4, P = 0.011) (Table 4). The strength of the association increased when excluding teachers that taught in gymnasiums or in special subjects classrooms during the day (see Table 4). The same tendency was observed for the association between the change in voice symptoms and the teacher’s average vocal load, where Spearman’s rho increased to rho = 0.29 and borderline significance when limiting the analysis to general classrooms (Table 4). The change in feeling of fatigue in the head correlated with the mean noise exposure with borderline significance (rho = 0.30, P = 0.08) when limiting the analysis to teachers in general classrooms. Neither the stress or energy scores nor the feeling that the workday had been strenuous correlated significantly with noise exposure or vocal load scores (Table 4). The duration of exposure may influence the association between noise and symptoms. However, we found little changes in the associations when excluding teachers with less than four lessons from the analyses. Effects of daily noise exposure and vocal load on change in cognitive test scores

Min–Max (n)

Sports Science Home economics

Classrooms for special subjects (n = 5) General classrooms (n = 32)

Table 2 Measurements (mean, SD, range, n) of school teachers’ ambient noise exposure, vocal load and speaking time by school subject

Gymnasiums (n = 3)

Int Arch Occup Environ Health

The percentage of correct responses in the TBT was negatively and significantly correlated with the teacher’s vocal load averaged over a working day (Table 4). The change in TBT performance also showed a moderate correlation with the teacher’s average noise exposure, in particular when limiting the analysis to general classrooms (Spearman’s rho = -0.35, P = 0.11). The scores in the SART did not correlate significantly with noise exposure or vocal load (Table 4).

Table 3 Distribution of speaking levels according to ISO 9921 in all school subjects except Sports, and in Sports lessons All school subjects, except Sports

Sports

n

Percentage

n

Percentage

1

1.3

0

–

Relaxed [68–73 dB(A)]

3

3.8

0

–

Normal [74–79 dB(A)]

26

33

0

–

Raised [80–85 dB(A)]

33

42

2

17

Loud [86–91 dB(A)]

15

19

6

50

0

–

4

33

Lowest [\68 dB(A)]

Very loud [C92 dB(A)]

Int Arch Occup Environ Health Fig. 1 Distribution of the teacher’s average daily noise exposure (left panel) and vocal load (right panel) during teaching (n = 34). The noise exposure was measured at shoulder level and the contribution from the teacher’s own voice was excluded

25

30

20 20 15

10 10 5

0

0 65

70

75

80

70

85

75

80

85

90

Table 4 Spearman’s rank correlations of average noise exposure during teaching and teacher’s average vocal load with changes in noise-related symptoms and cognitive performance during a workday Noise effect

Expected direction of effect

Mean noise exposure

Mean vocal load

Working area

Working area

All rooms

General and special subjects classrooms

General class rooms only

0.40* (33)

0.47** (27)

0.53** (23)

All rooms

General and special subjects classrooms

General class rooms only

Self-reported measures Change in voice symptoms Change in energy

Increase

0.02 (33)

-0.17 (28)

-0.18 (23)

0.13 (33)

Change in stress

Increase

-0.01 (33)

0.05 (28)

0.00 (23)

-0.10 (33)

0.00 (28)

-0.05 (23)

Change in fatigue (head)

Increase

0.11 (34)

0.21 (28)

-0.08 (34)

0.02 (28)

0.15 (23)

Strenuous workday

Increase

-0.01 (34)

0.08 (28)

-0.05 (23)

0.01 (34)

0.08 (28)

0.05 (23)

Change in inhibition errors in SART

Increase

0.01 (25)

0.08 (20)

-0.23 (15)

0.18 (25)

0.19 (20)

0.13 (15)

Change in percentage correct in two-back test

Decrease

-0.23 (23)

-0.30 (19)

-0.35 (14)

-0.55** (23)

-0.54** (19)

-0.47* (14)

Decrease

0.30(*) (23)

0.23 (33)

0.24 (27) -0.05 (28

0.29(*) (23) -0.13 (23)

Cognitive measures

The expected direction of the effect is the change assuming noise as associated with increasing mental exertion and development of fatigue. The number of measurements is shown in parentheses. One-tailed significance test of the hypothesis that higher sound levels is associated with increased fatigue: ** P \ 0.01; * P \ 0.05; (*) P \ 0.1

As above, excluding teachers with less than four lessons did not substantially change these findings.

Discussion This is the first study, to our knowledge, where school teachers’ noise exposure has been measured according to the principles normally employed to assess occupational noise exposure. The legislation in EU makes it mandatory to offer hearing protection to employees exposed to noise

above an equivalent 8 h level of 80 dB(A). Below this limit, the risk of noise-induced hearing damage is considered negligible (ISO 1999). With the possible exception of Sports lessons, our results indicate that noise exposure during school lessons normally are below this limit. However, with regard to Sports, the highest observed noise exposure was 83 dB(A), and therefore, it cannot be excluded that the equivalent 8-h noise exposure of some Sports teachers could exceed 80 dB(A). Due to the limited number of observations, it is not possible to judge how frequent this exposure limit may be exceeded. With regard

123

Int Arch Occup Environ Health

to the risk of hearing impairment in Sports teachers from a long-term excessive noise exposure, it should be remembered that often, and especially in the summer time, Sports activities take place outdoors which will reduce the yearly noise burden of the Sports teachers. Our measurements of noise exposure are generally in good agreement with the levels measured in an English study (Shield and Dockrell 2004) that showed an average equivalent sound level of 72 dB(A) in occupied classrooms. Similarly, a Swedish study have found equivalent sound levels ranging from 59 to 87 dB(A), but the majority fell in the range 60–70 dB(A) (Wa˚linder et al. 2007). The fact that we did not see any effect of RT on ambient noise levels is in agreement with Sato and Bradley (Sato and Bradley 2004) who similarly failed to find such an association in their study of acoustical conditions and noise levels in 41 classrooms. This led them to suggest that differences in classroom RT in the 0.4–0.6 s range presumably have only negligible influence on activity noise levels in classroom. Indeed, a small effect of RT on noise levels can easily be masked by the large variability in noise levels between classes and lessons. With regard to vocal load, we found that the vocal load of the teachers were on average 10 dB(A) higher than the background noise levels during classroom teaching and 11.7 dB(A) higher during Sports lessons in gymnasiums. These values are in reasonable agreement with the teacher voice signal-to-noise ratio of 9.3 dB(A) observed in school teachers (Sato and Bradley 2004) and 9 dB(A) with large individual differences in preschool teachers (So¨dersten et al. 2002; Lindstrom et al. 2011). Deviations may be due to differences in the methodologies, for example, the distance between the microphone and mouth of the teacher. Using ISO 9921 categories for speech loudness, we estimated that teachers spoke with an equivalent voice level corresponding to a raised voice 61 % of the lessons (excluding Sports). This is slightly more than 43 % of the time that day-care centers have been observed to speak with a raised voice and considerably more than the 14 % of the time for hospital nurses (Sala et al. 2002). On average, teachers raised their voice level by 0.65 dB(A) per dB(A) increase in noise, which is the socalled Lombard slope. Other studies have found different Lombard slopes in teachers from 0.22 dB/dB measured in the laboratory (Whitlock and Dodd 2006) to 0.82 dB/dB measured during actual teaching (Sato and Bradley 2004). Large interindividual variations have been noted in daycare center teachers (from 0 to ca. 0.9 dB/dB) (Lindstrom et al. 2011). However, voice problems are prevalent among teachers, and the teaching profession is overrepresented among patients diagnosed with voice disorders (Fritzell 1996; Russell et al. 1998; Smith et al. 1998). Voice problems are also recognized as an important factor of

123

˚ hlander et al. 2011; de absenteeism among teachers (A Medeiros et al. 2012). Thus, our findings indicate that classroom noise could contribute to increase the vocal load of the teachers. The speaking time varied greatly between teachers and lessons, but was generally in the order of 40–60 % of the time. It did not depend on the school subject, although speaking time during Sports tended to be in the lower end. Sala et al. (2002) observed an average speaking time of 40 % (range 19–63 %) in preschool teachers, which was significantly longer compared to the value they found for nurses (mean 28 %, range 6–50 %). Somewhat lower values in teachers, namely 21 and 22 %, was found in the studies of Titze et al. (2007) and Masuda et al. (1993). As noted by Sala et al. (2002), the use of different methods makes a direct comparison of values difficult. Differences may also be due to different pedagogical traditions between countries. Our final aim was to investigate whether the noise exposure and vocal load was associated with development of symptoms of fatigue, stress, low energy and voice symptoms during the workday. The results indicated that changes in voice symptoms correlated positively with the teachers’ average noise exposure during the workday and with average vocal load during the workday, although the latter correlation was only borderline statistically significant. A borderline significance was also found for the correlation between mean noise exposure and an increase in the feeling of ‘‘fatigue in the head,’’ while no associations were found with changes in energy or stress scores. At present, we have no explanation why the worsening of voice symptoms correlates better with noise exposure than with vocal load. One possibility is that voice symptoms exert at negative feedback on vocal load, that is, voice level is moderated downward when the voice starts feeling coarse. Some support for this notion comes from the study by Pelegrin-Garcia et al. (2010), who observed that a group of teachers with voice problems talked 1.4 dB lower than teachers without voice problems. Further, teachers with voice problems decreased their vocal load when the room reverberation time increased while teachers without voice problems did the opposite. Finally, the reduction in percentage of correct responses in the cognitive test TBT correlated significantly with the teachers’ average vocal load during the day. There was also a correlation with noise exposure in the expected direction, but this effect was just above the level of statistical significance. The reason that vocal load correlated stronger with cognitive fatigue measures could be that an increased vocal load during the day is better indicator for the mental workload than the background noise level itself. For example, when interviewed about noise, teachers have told us that they not always feel challenged

Int Arch Occup Environ Health

by a high level of classroom noise because high noise levels indicate active participation by the pupils in the teaching. It is only when it obstructs the teaching that noise becomes a problem. Speculatively, a raised vocal load may reflect situations where noise has become an obstruction to teaching. Before turning to the conclusions, some limitations of this study needs to be addressed. First, general classrooms were in focus of this study. Hence, other schools subjects taking place in other types of rooms were undersampled. Some school subjects, such as music and art, were not measured. Therefore, extrapolations to other teaching situations based on the results in this study should be done with some caution, in particular with regard to school subjects that are represented by few teachers and measurements (Sports, and also Home economics and Science). Secondly, the measurements made during school lessons may not reflect the total occupational noise exposure, for example, playground duty could contribute significantly with noise from playing children. The teachers that participated in this study may also have had different possibilities for restitution during the workday, for example, the number of staff meetings and free periods during the day could differ between the teachers, which would result in a weakening of the associations between exposure and symptoms. Another limitation is the low number of participants in the cognitive tests which makes it more difficult to control for confounding effects and reduced the possibilities to investigate the influence of the duration of the noise exposure on fatigue symptoms. We controlled for the effect of the working area by doing the analyses for different subgroups. Confounding from the teachers’ age, the pupils’ age and the school subject are also unlikely because our initial analyses showed nonsignificant effects on noise exposure and vocal load from these factors. But we cannot exclude the possibility of residual confounding from these factors, or the influence of unknown confounding factors. In conclusion, the model suggested in the introduction that links acoustical characteristics of school classrooms with noise levels and fatigue symptoms in teachers was partly supported by the findings in this study. The teachers’ noise exposure in the classroom is probably too low to be of concern with regard to noise-induced hearing loss. Nevertheless, the majority of the teachers were speaking with a raised voice during most of the lessons, and we found a strong positive association between classroom noise and vocal load. Finally, we found some indications that these objective measures of classroom noise and increased vocal load during teaching were associated with an increase in voice symptoms and development of signs of cognitive fatigue in teachers.

Acknowledgments This study was funded by the National Working Environment Fund (Project No. 16-2008-03). We are grateful to Bru¨el & Kjær Sound & Vibration A/S for loan of equipment and provision of expert technical assistance. Conflict of interest of interest.

The authors declare that they have no conflict

References ˚ hlander VL, Rydell R, Lo¨fqvist A (2011) Speaker’s comfort in A teaching environments: voice problems in Swedish teaching staff. J Voice 25(4):430–440 Bailey A, Channon S, Beaumont JG (2007) The relationship between subjective fatigue and cognitive fatigue in advanced multiple sclerosis. Multiple Scler 13(1):73–80 Banbury SP, Macken WJ, Tremblay S, Jones DM (2001) Auditory distraction and short-term memory: phenomena and practical implications. Hum Fact 43(1):12–29 Clausen T, Christensen KB, Lund T, Kristiansen J (2009) Selfreported noise exposure as a risk factor for long-term sickness absence. Noise Health 11(43):93–97 Clausen T, Kristiansen J, Hansen JV, Pejtersen JH, Burr H (2013) Exposure to disturbing noise and risk of long-term sickness absence among office workers. A prospective analysis of register-based outcomes. Int Arch Occup Environ Health 86(7):729–734 de Medeiros AM, Assuncao AA, Barreto SM (2012) Absenteeism due to voice disorders in female teachers: a public health problem. Int Arch Occup Environ Health 85:853–864 Fritzell B (1996) Voice disorders and occupations. Logoped Phoniatr Vocol 21:7–12 Hodgson M, Nosal EM (2002) Effect of noise and occupancy on optimal reverberation times for speech intelligibility in classrooms. J Acoust Soc Am 111(2):931–939 Hughes RW, Jones DM (2003) Indispensable benefits and unavoidable costs of unattended sound for cognitive functioning. Noise Health 6(21):63–76 International Organization for Standardization (1990) Acoustics— determination of occupational noise exposure and estimation of noise-induced hearing impairment (ISO 1999). International Organization for Standardization (ISO), Geneva Kjellberg A, Iwanowski S (1989) Stress/Energi-formula¨ret: Utveckling av en metod fo¨r skattning av sinnessta¨mning i arbetet. Stockholm, National Institute of Occupational Health, 26, 1–21 Kjellberg A, Wadman C (2002) Subjektiv stress och dess samband med psykosociala fo¨rha˚llanden och besva¨r-En pro¨vning av Stress-Energi-modellen. In: Marklund S (ed) Arbete och Ha¨lsa. Stockholm, Arbetslivsinstitutet, 12, 1–32 Kjellberg A, Ljung R, Hallman D (2008) Recall of words heard in noise. Appl Cogn Psychol 22(8):1088–1098 Kristiansen J (2010) Is noise exposure in non-industrial work environments associated with increased sickness absence? Noise Vib Worldw 41(5):9–16 ˚ M, Shibuya H, Kristiansen J, Mathiesen L, Nielsen PK, Hansen A Petersen HM et al (2008) Stress reactions to cognitively demanding tasks and open-plan office noise. Int Arch Occup Environ Health 82(5):631–641 Kristiansen J, Lund SP, Nielsen PM, Persson R, Shibuya H (2011) Determinants of noise annoyance in teachers from schools with different reverberation times. J Environ Psychol 31:383–392 Kristiansen J, Persson R, Lund SP, Shibuya H, Nielsen PM (2013) Effects of classroom acoustics and self-reported noise exposure on teachers’ well-being. Environ Behav 45(2):283–300

123

Int Arch Occup Environ Health Lindstrom F, Waye KP, So¨dersten M, McAllister A, Ternstro¨m S (2011) Observations of the relationship between noise exposure and preschool teacher voice usage in day-care center environments. J Voice 25(2):166–172 Ljung R, So¨rqvist P, Kjellberg A, Green A (2009) Poor listening conditions impair memory for intelligible lectures: implications for acoustic classroom standards. Build Acoust 16(34):257–265 Masuda T, Ikeda Y, Manako H, Komiyama S (1993) Analysis of vocal abuse: fluctuations in phonation time and intensity in 4 groups of speakers. Acta Otolaryngol 113:547–552 Miyake A, Friedman NP, Emerson MJ, Witzki AH, Howerter A (2000) The unity and diversity of executive functions and their contributions to complex ‘‘frontal lobe’’ tasks: a latent variable analysis. Cogn Psychol 41(49):100 Nijs L, Saher K, den Ouden D (2008) Effects of room absorption on human vocal output in multitalker situations. J Acoust Soc Am 123(2):803–813 ˚ hlander V, Rydell R, Brunskog J, Pelegrin-Garcia D, Lyberg-A Lo¨fqvist A (2010) Influence of classroom acoustics on the voice levels of the teacher with and without voice problems: a field study. Proc Meet Acoust 11:1–9 ˚ M, Ohlsson K, Ørbæk P (2003) The Persson R, Garde AH, Hansen A influence of production systems on self-reported arousal, sleepiness, physical exertion and fatigue—consequences of increasing mechanization. Stress Health 19:163–171 ˚ M, Ørbæk P, Karlson B ¨ sterberg K, Garde AH, Hansen A Persson R, O (2010) Seasonal variation in self-reported arousal and subjective health complaints. Psychol Health Med 15(4):434–444 Picard M, Bradley JS (2001) Revisiting speech interference in classrooms. Audiology 40(5):221–244 Posner J, Russell JA, Peterson BS (2005) The circumplex model of affect: an integrative approach to affective neuroscience, cognitive development, and psychopathology. Dev Psychopathol 17(3):715–734 Russell JA (1980) A circumplex model of affect. J Pers Soc Psychol 39(6):1161–1178 Russell A, Oates J, Greenwood KM (1998) Prevalence of voice problems in teachers. J Voice 12(4):467–479

123

Sala E, Airo E, Olkinuora P, Simberg S, Stro¨m U, Laine A et al (2002) Vocal loading among day care center teachers. Logoped Phoniatr Vocol 27:21–28 Sandrock S, Schu¨tte M, Griefahn B (2009) Impairing effects of noise in high and low noise sensitive persons working on different mental tasks. Int Arch Occup Environ Health 82:779–785 Sato H, Bradley JS (2004) Evaluation of acoustical conditions for speech communication in active elementary school classrooms. Proceedings of the 18th international congress on acoustics, II:1187–1190 Shield BM, Dockrell JE (2003) The effects of noise on children at school: a review. Build Acoust 10(2):97–106 Shield B, Dockrell JE (2004) External and internal noise surveys of London primary schools. J Acoust Soc Am 115(2):730–738 Smith E, Lemke J, Taylor M, Kirchner HL, Hoffman H (1998) Frequency of voice problems among teachers and other occupations. J Voice 12(4):480–488 So¨dersten M, Granqvist S, Hammarberg B, Szabo A (2002) Vocal behavior and vocal loading factors for preschool teachers at work studied with binaural DAT recordings. J Voice 16(3): 356–371 Titze IR, Hunter EJ, Svec JG (2007) Voicing and silence periods in daily and weekly vocalizations of teachers. J Acoust Soc Am 121(1):469–478 van der Linden D, Frese M, Meijman TF (2003) Mental fatigue and the control of cognitive processes: effects on perseveration and planning. Acta Psychol 113:45–65 van der Linden D, Keijsers GPJ, Eling P, van Schaijk R (2005) Work stress and attentional difficulties: an initial study on burnout and cognitive failures. Work Stress 19(1):23–36 Wa˚linder R, Gunnarsson K, Runeson R, Smedje G (2007) Physiological and psychological stress reactions in relation to classroom noise. Scand J Work Environ Health 33:260–266 Whitlock J, Dodd G (2006) Classroom acoustics—controlling the cafe effect… is the Lombard effect the key? In: Proceedings of ACOUSTICS, Christchurch, New Zealand, 20–22 Nov 2006, pp 423–426