Introduction

Classroom acoustic conditions often pose challenges for student listening. Sources of intrusive noise include computers, air conditioning, adjacent classrooms, and particularly the students themselves1. Listening in noise can be further confounded by the negative effects of reverberation, distance from the teacher, and the increasing adoption of open-plan room configurations2,3. Combined, these factors impede the audibility of the teacher’s voice, by resulting in unfavorable signal-to-noise ratios (SNRs). Signal-to-noise ratios measure the difference in intensity of a target signal compared to competing background noise. Typical closed-plan classrooms in the United Kingdom (UK) have SNRs of −7 dB to +5 dB4, which are markedly different from the recommendation of +15 dB at minimum in educational settings5. Given that 45–60% of a mainstream primary school day is spent on actively listening6, it is critical that students receive clear auditory input from their peers and educators for effective engagement.

Inadequate classroom acoustics have profound consequences for many aspects of primary-aged learning, including academic progress7,8,9 and reading10. Chronic exposure to classroom noise, encompassing both internal and external sources, has been negatively associated with learners’ attention, concentration, and memory abilities11,12. The consequence of noise on cognition is further demonstrated by reduced performance on academic tasks in the presence of competing speech sounds. Shield and Dockrell9 found poorer outcomes on tests of English, Science, and Numeracy in classrooms with higher external and classroom noise levels, particularly in children aged 11 and older. This increased effect with age was attributed to compounding exposure over time. A review of studies investigating the association between noise and academics, indicated that memory, spelling, mathematics, and reading were impaired by other background talkers13. The authors proposed that these effects vary based on the task and noise type and that possible explanations could be cognitive interference or distractibility. More recently, Rance et al.14 demonstrated greater reading fluency progress when students were situated in more acoustically ideal enclosed classrooms as opposed to open settings. In addition to behavioral evidence, teacher ratings have demonstrated improved verbal language comprehension, expression, and achievement on recognition tasks in quieter learning conditions10.

All students are adversely affected by the classroom acoustic environment, but some groups are particularly vulnerable. In UK primary schools, children under age nine years are most impacted by noise levels15.This is consistent with the progressive development of speech recognition skills, which continue to improve until approximately age ten16. Before this, neurodevelopment in the auditory system and the ability to use language contextual cues to establish missed messages are immature. Similar academic risks have been demonstrated in children who are not taught in their primary speaking language17,18. Additional learners, who are more affected by the acoustic environment than others, can be grouped under the umbrella term of children with listening difficulties (LiD)19. These are children who have a poorer understanding of verbal information in noisy environments19,20,21. Dillon and Cameron19 summarized the possible causes of LiD as deficits in cognition, auditory processing (AP), language development, hearing (or a combination of these factors), and in many cases, the exact etiology is uncertain. There is substantial evidence supporting higher academic risk for LiD populations, including those with hearing loss22, auditory processing disorder23, autism spectrum disorder24,25, attention deficits25, learning disorders26, and language disorders27. With increasing awareness and accessibility to diagnostic services globally28, the prevalence of LiD in childhood has risen. For example, diagnoses of neurodevelopmental conditions in children aged 3–17 years in the United States increased by 9.5% from 2009 to 201729. In Australia, there has been a shift towards more integrated education, where 89% of primary and secondary students with disabilities now attend mainstream schools rather than specialist institutions30. Therefore, it is essential that mainstream classrooms are acoustically optimized to ensure inclusive learning for all students. Installation of soundfield amplification (SFA) systems is one approach to improving classroom listening2,31. These systems function by sending the teacher’s voice (recorded via a lanyard-worn microphone) to a series of loudspeakers arranged around the classroom. In doing so, they elevate important instructions above the ambient noise and evenly distribute auditory information throughout the room to allow a teacher to be heard without having to raise their voice. Typical amplification ranges from 8 dB to 10 dB, which in a standard classroom can mean the difference between negligible and complete access to the spoken message32. The benefit of SFA extends beyond student access to the educators themselves, who report less voice strain and therefore exhaustion from teaching, as well as increased job satisfaction33.

Enhanced speech recognition in both normal hearing and hearing-impaired students using SFA is well established34,35,36,37. In addition, several studies have presented anecdotal benefits from teachers, including improved student attention, greater listening and engagement, reduced discipline challenges, and more relaxed interactions with pupils6,38,39.

While the benefits of SFA for listening in the classroom are established, it remains unclear whether increasing access to the teacher’s voice results in measurable improvements for student learning. Increased rates of academic progress have been reported from SFA use in hearing-impaired students (Mainstream Amplification Resource Room Study), first nations populations across Canada and Australia33,40,41,42 and students who are culturally and linguistically diverse (CALD)33,41. However, the literature on the general student population has reported inconsistent findings. Within these SFA studies, most have used literacy or reading-based measures because they are thought to be strong predictors of educational progress43,44. Massie and Dillon41 investigated the benefit of SFA on academic progress in 12 cross-cultural classes of Australian year 2 students (n = 242) across one school year. Teacher ratings of reading, writing, and numeracy indicated significantly greater academic development during SF amplified periods (two terms). A more recent Canadian study explored academic outcomes in year 1 students (n = 486), with half of the classes using SFA and the others acting as controls for one school year. Using the standardized developmental reading assessment (DRA), they failed to show a significant difference in reading outcomes between the amplified and unamplified classrooms45. Another study on older UK students aged 8 to 11 years investigated differences in outcomes for five classes using SFA (n = 114) compared to three control classes (n = 72) over a six-month period46. They found no significant benefit from SFA for numeracy, reading, or spelling.

While the discrepancies in SFA study findings on academic outcomes are unclear, they may be partially attributed to differences in study design. Massie and Dillon41 employed a within-subject crossover design, where each classroom alternated between amplified and unamplified conditions each term. In contrast, the other investigations used between-subjects designs, comparing classes with amplification to age-matched control classes45,46. Within-subject designs may be more sensitive to detecting effects in educational research as they control for individual differences and classroom-specific factors. This approach increases statistical power, which may explain why an effect was seen exclusively in Massie and Dillon41. Given this study was limited to year 2 students, future research should employ a similar study design in older age groups to determine whether similar academic benefits apply to these students.

The current longitudinal study aimed to explore the effect of periods of SFA and non-amplification on the development of reading fluency in grade 3–4 children. Based on previous findings in the Australian context, it was hypothesized that reading fluency would progress faster when children were exposed to a clearer acoustic signal through SFA. A secondary aim was to determine whether student characteristics, including intelligence, attention, listening ability in noise, and memory, influenced the effect of SFA on literacy development. Given that children with difficulties in these areas are more impacted by poor classroom acoustics, it was expected that children with poorer listening and cognition would be afforded larger benefits by SFA.

Results

Reading fluency at baseline

Baseline reading ability was normally distributed across the cohort. Mean reading fluency rate at baseline for the cohort was 106.4 ± 39.3 words per minute, consistent with published normative data for this age group47.

A range of participant characteristics thought likely to impact reading fluency were correlated with baseline abilities. The findings are summarized in Table 1.

At baseline, reading fluency was significantly associated with participant IQ, attention, auditory working memory, and speech discrimination in noise. There was no association between reading fluency and age. IQ was also significantly associated with attention and speech discrimination in noise ability.

The significant variables from this baseline analysis (plus test site [school]) were combined in a general linear model to determine the independent predictors of reading fluency (see Table 2). Auditory working memory (p = 0.016), attention (p = 0.002), and school (p = 0.006) were significant. Speech discrimination in noise (p = 0.107) and IQ (p = 0.988) were not independent predictors of baseline reading fluency. Tukey post hoc comparisons showed that reading fluency findings for the rural school were significantly poorer than for the metropolitan schools.

Table 1 Correlations between baseline WARP and other baseline measures
Full size table
Table 2 General linear model of factors influencing baseline reading fluency
Full size table

Reading fluency development across the data collection period

A mixed effect analysis was undertaken on participants who completed both SFA-On and SFA-Off assessments (n = 84). Individual participants were treated as a random variable with fixed factors of classroom condition (baseline, SFA-On, SFA-Off), condition order (On first, Off first), and school. Auditory short-term memory and attention were also included as covariates.

Classroom conditions were significantly associated with reading fluency development (F = 20.87, p < 0.001), whereas school (F = 2.12, p = 0.072) and condition order (F = 0.00, p = 0.968) were not. Attention ability was a significant covariate (F = 16.26, p < 0.001), but auditory working memory (F = 3.47, p = 0.067) was not.

Tukey post hoc comparisons for the classroom condition factor showed that the mean reading scores for SFA-On (M = 116.2, p < 0.001) and SFA-Off (M = 116.0, p < 0.001) were significantly higher than the baseline measurement (M = 107.6), indicating general improvement in reading fluency over the course of the study. There was, however, no significant difference in the mean reading fluency scores between the SFA-On and SFA-Off conditions (p = 0.985).

Analysis of within-child changes across conditions revealed no significant difference in reading fluency development through the SFA-On vs SFA-Off study phases (t = 0.12, p = 0.908). Mean change in WARP score for SFA-Off was 3.9 ± 11.2 words per minute, and SFA-On was 4.1 ± 10.6 words per minute.

Factors affecting reading development

Mean reading fluency improvement in the SFA-On and SFA-Off conditions was calculated for each participant who completed the protocol (n = 84), and used to derive a “Soundfield Effect” using Eq. (1).

$$Soundfield\,Effect\,Score\,=\,\Delta WARP\,SFA-On-\Delta WARP\,SFA-Off$$
(1)

Forty-one participants (48.8%) had a positive score, indicating a higher rate of reading fluency when assessed during the SFA-On condition.

A general linear model analysis with Soundfield Effect score for each participant as the dependent variable was undertaken. Order of testing, age, baseline reading score, auditory working memory, speech discrimination in noise, and attention were not significant in the analysis (p < 0.05).

IQ (F = 4.5, p = 0.036) was a mildly significant predictor of the Soundfield Effect score. That is, children with a lower IQ had a greater degree of reading fluency improvement in the SFA-On condition (Fig. 1).

Fig. 1: Correlation between non-verbal intelligence and soundfield effect score.
figure 1

The relationship between participant non-verbal intelligence, as measured using the TONI-4, and the effect of SFA on reading fluency development, as measured using the Soundfield Effect score. The blue dots display individual data points, and the orange line depicts the best-fit linear regression. 95% CI for Pearson Correlation. CI 95% confidence interval, r Pearson correlation coefficient, SFA soundfield amplification, CI confidence interval, TONI-4 test of non-verbal intelligence fourth edition.

Full size image

Discussion

SFA is a well-established classroom intervention that increases student access to instruction while also reducing teacher voice strain33. However, previous findings regarding the effect on academic results are mixed. The present study investigated the effect of SFA on reading fluency in pupils aged 8–10 years, with progress expected to advance more rapidly during periods of SFA use. The findings indicated no benefit under sound-amplified conditions. A secondary aim was to determine whether participant characteristics impacted the degree of reading progress attained. While most participant characteristics showed no relationship, participant non-verbal IQ was associated with SFA usage.

An individual’s academic progress results from a multifaceted interaction between innate characteristics and environmental factors48. The links between academic development, cognition, and AP are well established14,44. Children with AP deficits (such as speech recognition in noise), show poorer scholarly outcomes compared to their peers44,49. Furthermore, cognitive skills, particularly attention and working memory, significantly impact AP test performance and functional difficulties44,50. The findings of the current study were consistent with previous literature, showing baseline reading fluency was related to speech discrimination in noise, attention, auditory working memory, and non-verbal IQ. Non-verbal IQ was also associated with speech discrimination and attention, emphasizing the interrelationship between these skills in addition to their cumulative impact on academic skills such as reading fluency. School and location are also strong predictors of academic progress51. Baseline reading fluency was poorer for the rural school compared to metropolitan locations. This finding aligns with established literature that urban students generally outperform their regional counterparts due to broader socioeconomic and educational resource disparities51,52.

The current study did not find that SFA significantly affected the rate of reading fluency development in 8–10-year-old students. Previous studies looking at academic outcomes from similar interventions have had mixed findings. In students of similar age, Dockrell and Shield46 showed no overall change in academic achievement, attention, or language processing from SFA. Similarly, Millett and Purcell45 found no amplification effect on reading outcomes. In contrast, Massie and Dillon33 demonstrated significantly increased numeracy and literacy skill development in year two children when SFA was used. This discrepancy in outcomes may be attributed to the age difference across study populations. SFA studies incorporating broader age ranges have consistently shown greater auditory and learning benefits for the younger age groups6,53,54. Speech perception matures gradually, and ongoing auditory and language development may explain why these benefits are limited to younger students17,18. Significant differences in AP abilities persist even between 7-year-olds and 8–12-year-olds55. Moreover, younger children are more susceptible to the adverse effects of noise and reverberation, making them more likely to benefit from auditory enhancements15. While SFA can provide greater access to instruction for all students, these findings suggest the translational impact on reading fluency applies particularly to younger year levels.

The lack of significant benefit from SFA may have instead been attributed to classroom acoustic factors. While the occupied classroom noise levels of approximately 65 dBA represented ideal candidates for SFA implementation, the system’s impact on overall acoustic conditions was limited. The mean 7 dB SNR improvement with SFA in study classrooms (from approximately −7 dB to 0 dB with SFA) was consistent with other studies documenting typical classroom SNR improvements from −4 dB to +4 dB4. However, this improvement fails to meet the recommendation of the +15 dB SNR at minimum for optimal learning32. Despite not reaching ideal conditions, other studies have suggested that even modest improvements in SNR can enhance speech intelligibility and reduce listening effort for both typical students and those with LiD4,56. Furthermore, previous research indicates enhanced speech recognition in both normal hearing and hearing-impaired students with SFA34,35,36,37. Therefore, it is likely that SFA is positively impacting access to the teacher’s voice without translating to measurable improvements in learning outcomes for most students.

Intelligence was a mild predictor of the reading fluency advantage obtained during the SFA-On study periods. This was even though all subjects had IQ scores within normal limits (>85). To our knowledge, this is the first investigation to show the association between soundfield benefit and intellectual abilities. The mechanism behind this may be explained by Cognitive Resource Allocation theory, which proposes a limited pool of resources that are flexibly allocated depending on task demands57. With increasing auditory difficulties, cognitive processing becomes compromised due to the greater resource requirement for the perception of an acoustic signal57. Individuals with higher intelligence can allocate more resources to demanding tasks58, which may explain the observed relationship between non-verbal IQ and reading benefits derived from SFA.

The other cognitive and listening skills assessed, including attention, working memory, and speech recognition, were not related to the reading fluency change from SFA. This finding is unexpected, considering the well-established correlations between intelligence and other skills associated with listening44. Previous studies have shown that other at-risk student groups gain more from SFA compared to their peers. For example, while showing no overall change in academic progress with SFA, Dockrell and Shield46 found improved performance in those students previously identified as requiring individualized learning support. Combined, these findings indicate individuals may obtain an academic advantage from SFA despite no effect on the general student body.

There were some methodological limitations associated with this study. While the recruitment rate for each class was not documented, based on the average Victorian class size of 2259, our observed range of one to 12 participants per classroom indicates a potentially low study uptake rate. Moreover, while the rural school showed the lowest Index of Socio-Educational Advantage (ICSEA) score of 956, all schools were relatively close to the median of 100060. This risks the sample not representing the broader student community and degrees of advantage observed across the education system. However, the fact that the baseline reading levels of our participants were normally distributed and consistent with published normative values suggests that we tested a representative sample. Another potential limitation of the study is that only 84 of the 152 students were available to complete the longitudinal protocol. This may have lessened the reliability of the findings. The attrition rate was, however, within our expected range, and pre-study Power calculations (based on Rance et al. [61]) indicated that only 70 participants were required to achieve a Power value of 0.8. Furthermore, comparison of the baseline findings suggested that there were no significant differences (p > 0.05) in demographic characteristics or study results between those children who were or were not available for each of the data collection points. As such, the absence of the latter is unlikely to have biased the longitudinal findings. An additional constraint was the lack of regular monitoring of device use during SFA-On periods. While the participating schools/teachers were typically enthusiastic about the use of SFA, the lack of a data-logging facility in the systems meant that we could not be certain about the number of hours used per day during the SFA-On phases. Future studies should consider teacher questionnaires or random visits to ensure compliance and appropriate use. A final limitation was restricting our measure of academic progress to the outcome of reading fluency. While it is well established that reading is a strong predictor of academic development44, a more detailed evaluation of student learning (particularly involving numeracy and spelling) may have revealed benefits specific to the provision of SFA41.

In summary, findings from the present study support the use of SFA in mainstream classrooms. Amplification advanced reading fluency development for those students with lower intellectual abilities, while having no consequences on the academics of other students. While the repercussions of poor acoustics vary between individual children, there will always be a group of students whose academic development is at risk in less-than-optimal listening conditions, therefore, SFA should be considered important support in classrooms that do not meet recommended acoustic standards. Future research on larger cohorts may consider additional participant characteristics such as neurodevelopmental diagnoses, and in what acoustic circumstances these technologies may be optimally used.

Methods

Ethics approval

This study was approved by the University of Melbourne Ethics Department (2023-25998-37500-4), the Research in Victorian Schools and Early Interventions Office, Melbourne, Australia (2023_004702), and the Melbourne Archdiocese Catholic Schools Research Committee (1257). Prior to participation, families were informed that involvement was voluntary, and written consent was obtained from each child’s caregiver.

Participating schools

Twenty-three grade 3 and 4 classes from six mainstream primary schools in Victoria, Australia, participated in the study. One school was located rurally while the remaining five were based in Metropolitan Melbourne (Table 3). The ICSEA score for each school can be found in Table 3. The ICSEA considers student and school-level factors to determine advantage relative to other schools. It ranges from 500 to 1300, with a median score of 1000 (SD: 100)60. All sites are enrolled at the beginning of the same academic year to ensure teaching staff consistency across the study period. Classes were considered eligible if they were situated in an enclosed classroom and had no existing SFA in use.

Classroom noise levels

Acoustic measurements were completed in each classroom in the SFA-On and SFA-Off conditions. Students and teachers completed their usual activities during these periods. In each condition, 30-min samples were obtained using an SVAN971 sound level meter, documenting noise level (in dBC) in 1-min intervals. The mean occupied noise levels for each test environment are summarized in Table 4.

Table 3 School demographic information
Full size table

SFA characteristics

The SFA system utilized was the Front Row Juno. One system was installed in each classroom, and educators were instructed not to move the equipment during the study period. The gain provided by the Juno system was determined by recordings of the teacher's voice levels in each unoccupied classroom. Teachers were instructed to read a prescribed text (WARP reading list 4). Average voice level in dBA was measured over a 30 second period in 6 seating positions around the classroom with and without SFA. The gain provided by the system was the difference between the mean unamplified and amplified level. Mean voice levels with and without SFA across the 23 classes are summarized in Table 4.

Table 4 Classroom details and acoustic measures
Full size table

Study design

This study employed a within-subject crossover design, as has been used in previous Australian SFA trials41. Each class participated in the study across three academic school terms and was randomized into one of two condition sequences: (1) On first: SFA-On for term 2, SFA-Off for term 3, and SFA-On for term 4, or (2) Off first: SFA-Off for term 2, SFA-On for term 3, and SFA-Off for term 4. Twelve classrooms (80 subjects) started with SFA-On first, while the remaining eleven classrooms (72 subjects) had the opposite order (see Table 4). A member of the research team visited each site at the start of each term to check equipment functionality and ensure correct settings (On or Off) depending on the determined condition schedule.

Participants

While SFA was used by all children in each class, only those whose parents/caregivers provided written consent were behaviorally evaluated in the study. One hundred and fifty-two students (66 male) aged between 8.2 and 10.7 years (mean = 9.4, SD = 0.6 years) were enrolled. A breakdown of the participant age range across schools is shown in Table 3. All enrolled children had normal sound detection thresholds (screened at 20 dBHL) for pure tones at octave frequencies between 250 Hz and 8 kHz.

Data collection occurred in the final week of each school term. The final week of each term was selected to ensure the longest possible exposure to the intervention while maintaining comparable time frames across all assessment periods. Each term lasted 10 ± 1 weeks. Baseline measures were completed on each child at the end of the term they were enrolled in the study. One hundred and twenty-eight were enrolled at the beginning of term 2, and the remaining 24 participated from the beginning of term 3, meaning the latter subjects completed two terms in the study instead of three. Reading fluency was re-assessed at the end of subsequent terms, and changes in reading fluency across each term were determined. Where development across two terms with the same condition was measured (e.g., SFA-On phases for classes following the on/off/on schedule), an average of the change values for the two terms was used.

Participants were individually removed from classes for behavioral data collection by an experienced researcher. These assessments were completed in a quiet room at each school with minimal background noise (<40 dBA). Eighty-four participants completed the longitudinal protocol, allowing within-child comparison of development across both SFA-On and SFA-Off study phases. The remaining 68 participants were missing data for at least one condition due to school absences on the allocated testing days.

Materials

Reading fluency was selected as the primary academic outcome in line with research indicating that it is a strong predictor of educational outcomes in primary school-aged children43,44. The Wheldall Assessment of Reading Passages (WARP) measured reading fluency47. Participants were given a 200-word passage to read aloud, and the percentage of correctly read words in 1 min was calculated. A previous normative study indicates the following expectations for age ranges included in the sample: 109 ± 40 words/min for 8-year-olds; 118 ± 39 words/min for 9-year-olds, and 135 ± 39 words/min for 10-year-olds47.

The Test of Non-Verbal Intelligence (TONI-4) was used to assess baseline cognitive ability61. Participants had to select the missing item from a closed-set series of pictures to complete a visual matrix, with subsequent matrices increasing in complexity. Scores are compared to age-specific normative data to provide a normalized intelligence quotient (IQ).

The Digit Span digit reverse subtest of the Clinical Evaluation of Language Fundamentals 4 (CELF-4) evaluated auditory working memory ability62. Participants listened to a series of digits that progressively increased in length and had to correctly repeat each sequence in reverse order. Age-corrected scaled scores were calculated by the assessor, where a mean scaled score is 10 with a SD of 3.

The Listening in Spatialized Noise—Sentences Test (LiSN-S) DV90 condition was used to determine binaural speech discrimination in noise ability63. Under headphones, participants had to repeat sentences presented by a target speaker from the front amongst different competing background speech presented from the side. The test software calculated a speech reception threshold (SRT)—the signal-to-noise ratio at 50% of the target words were correctly discriminated. Each child’s SRT was then compared to normative data to calculate an age-specific z-score.

The integrated visual and auditory continuous performance test (IVA-CPT) measured auditory and visual attention skills64. This involved an 8-min computer-based task in which children had to indicate if they saw or heard a ‘number one’ stimulus while ignoring any ‘number two’ stimulus. The test software calculated a scaled attention quotient based on age-specific normative data. This score is derived from vigilance (errors of omission), focus (consistency of processing speed), and speed (reaction time) towards both the auditory and visual stimuli during the test. An average scaled attention quotient is 100 with a SD of 15.

Statistical analyses

Data was analyzed by the first author using MINITAB 19 software. Assumptions for parametric analyses were met. To determine independent predictors of reading fluency, correlations were run between baseline reading scores, participant age, and cognitive abilities. The significant factors from this analysis were then included as independent variables in a general linear model with baseline reading fluency as the dependent variable. A general linear model was also conducted on the SFA Effect Score (difference in reading development scores for the SFA-On and SFA-Off conditions), including test order, age, and baseline cognitive abilities as independent variables. In all multivariate analyses, Tukey post-hoc tests were conducted to assess significant differences.