Speech Recognition Abilities in Normal-Hearing Children 4 to 12 Years of Age in Stationary and Interrupted Noise

Abstract

Objectives: The main purpose of this study was to examine developmental effects for speech recognition in noise abilities for normal-hearing children in several listening conditions, relevant for daily life. Our aim was to study the auditory component in these listening abilities by using a test that was designed to minimize the dependency on nonauditory factors, the digits-in-noise (DIN) test. Secondary aims were to examine the feasibility of the DIN test for children, and to establish age-dependent normative data for diotic and dichotic listening conditions in both stationary and interrupted noise.

Design: In experiment 1, a newly designed pediatric DIN (pDIN) test was compared with the standard DIN test. Major differences with the DIN test are that the pDIN test uses 79% correct instead of 50% correct as a target point, single digits (except 0) instead of triplets, and animations in the test procedure. In this experiment, 43 normal-hearing subjects between 4 and 12 years of age and 10 adult subjects participated. The authors measured the monaural speech reception threshold for both DIN test and pDIN test using headphones. Experiment 2 used the standard DIN test to measure speech reception thresholds in noise in 112 normal-hearing children between 4 and 12 years of age and 33 adults. The DIN test was applied using headphones in stationary and interrupted noise, and in diotic and dichotic conditions, to study also binaural unmasking and the benefit of listening in the gaps.

Results: Most children could reliably do both pDIN test and DIN test, and measurement errors for the pDIN test were comparable between children and adults. There was no significant difference between the score for the pDIN test and that of the DIN test. Speech recognition scores increase with age for all conditions tested, and performance is adult-like by 10 to 12 years of age in stationary noise but not interrupted noise. The youngest, 4-year-old children have speech reception thresholds 3 to 7 dB less favorable than adults, depending on test conditions. The authors found significant age effects on binaural unmasking and fluctuating masker benefit, even after correction for the lower baseline speech reception threshold of adults in stationary noise.

Conclusions: Speech recognition in noise abilities develop well into adolescence, and young children need a more favorable signal-to-noise ratio than adults for all listening conditions. Speech recognition abilities in children in stationary and interrupted noise can accurately and reliably be tested using the DIN test. A pediatric version of the test was shown to be unnecessary. Normative data were established for the DIN test in stationary and fluctuating maskers, and in diotic and dichotic conditions. The DIN test can thus be used to test speech recognition abilities for normal-hearing children from the age of 4 years and older.

Fig. 1.

Age Dependency of DIN SRT and pDIN SRT for monaural presentation. (A) pDIN SRT vs. age. (B) DIN SRT vs. age. The thick line represents an exponential fit to the data. DIN indicates digits-in-noise; pDIN, pediatric digits-in-noise; SNR, signal to noise ratio; SRT, speech reception threshold.

Fig. 2.

Equivalence of DIN SRT and pDIN SRT. A, DIN SRT vs. pDIN SRT. Pearson r = 0.74 for a single test and Pearson r = 0.80 for the average of test and retest. The thick line represents the equal-SRT line. B, The difference in pDIN SRT–DIN SRT as a function of age. The slope and offset of the linear fit (represented by the thick line) are not significantly different from zero. DIN indicates digits-in-noise; pDIN, pediatric digits-in-noise; SNR, signal to noise ratio; SRT, speech reception threshold.

Fig. 3.

DIN SRT as a function of age for different conditions: (A) N0S0 stationary noise, (B) N0S0 fluctuating noise, (C) N0Sπ stationary noise, (D) N0Sπ fluctuating noise. The thick lines are exponential fits to the data. The results from the first test for each condition are shown. The adult data are summarized in a box-whisker plot [median group SRT (horizontal line), 25th and 75th percentile SRT (box ends), 10th and 90th percentile SRT (whiskers), and outliers (open circles)]. DIN indicates digits-in-noise; N0S0, diotic; N0Sπ, dichotic; SNR, signal to noise ratio; SRT, speech reception threshold.

Fig. 4.

Adult speech recognition functions and baseline-predicted masking release. A, Diotic and dichotic speech recognition functions for adult subjects for the four listening conditions tested. Data points show the mean percentage correct at each SNR. The lines show a logistic function fit to the individual data points. B, Masking release as a function of the baseline SNR for FMB, BU, and the combined effect. Note that there is a reduced release of masking at high SNR relative to low SNR, as illustrated with gray arrows for the FMB case. BU indicates binaural unmasking; DIN, digits-in-noise; FMB, fluctuating masker benefit; FMB&BU, the combined effect of fluctuating masker benefit and binaural unmasking; N0S0_fluct., diotic presentation using fluctuating noise; N0S0_stat., diotic presentation using stationary noise; N0Sπ_fluct., dichotic presentation using fluctuating noise; N0Sπ_stat., dichotic presentation using stationary noise; SNR, signal to noise ratio; SRT, speech reception threshold.

Fig. 5.

Masking release (mean and standard error of the mean) for different age groups. The measured masking release (black line) is compared with the baseline-corrected masking release for adults (gray line), that is, the predicted masking release, an adult would experience relative to a baseline SNR equal to the SRT-DIN of the children in each age group. Results are shown for (A) FMB, (B) BU, and (C) FMB&BU. Only for fluctuating noise, part of the reduced masking release observed with children can be explained by the lower SNR in the N0S0_stat. condition for children. For each condition tested (except BU in 10- to 12-year-olds), adults have more masking release than children, even when accounting for baseline performance. The error bars are smaller than the associated symbols in some cases. BU indicates binaural unmasking; DIN, digits-in-noise; FMB, fluctuating masker benefit; FMB&BU, the combined effect of fluctuating masker benefit and binaural unmasking; N0S0_stat., diotic presentation using stationary noise; SNR, signal to noise ratio; SRT, speech reception threshold.