Mothers adapt their voice during children\u2019s adolescent development

Simon Leipold,Daniel A Abrams,Vinod Menon

doi:10.1038/s41598-022-04863-2

Simon Leipold, Daniel A Abrams + Show 1 more

Open Access

https://doi.org/10.1038/s41598-022-04863-2

Copy DOI

Abstract

Mothers alter their speech in a stereotypical manner when addressing infants using high pitch, a wide pitch range, and distinct timbral features. Mothers reduce their vocal pitch after early childhood; however, it is not known whether mother’s voice changes through adolescence as children become increasingly independent from their parents. Here we investigate the vocal acoustics of 50 mothers of older children (ages 7–16) to determine: (1) whether pitch changes associated with child-directed speech decrease with age; (2) whether other acoustical features associated with child-directed speech change with age; and, (3) the relative contribution of acoustical features in predicting child’s age. Results reveal that mothers of older children used lower pitched voices than mothers of younger children, and mother’s voice pitch height predicted their child’s age. Crucially, these effects were present after controlling for mother’s age, accounting for aging-related pitch reductions. Brightness, a timbral feature correlated with pitch height, also showed an inverse relation with child’s age but did not improve prediction of child’s age beyond that accounted for by pitch height. Other acoustic features did not predict child age. Findings suggest that mother’s voice adapts to match their child’s developmental progression into adolescence and this adaptation is independent of mother’s age.

Highlights

We investigated whether maternal vocal features were correlated with, and predictive of, the child’s age, with three primary aims for the analysis: For the first aim, we focused on pitch height, the most salient acoustical feature of mother’s v oice[25]
Results indicated a negative correlation of medium effect size between child’s age and pitch height of their mother’s voice (r = − 0.38, 95% confidence interval (CI) [− 0.60, − 0.11], t(48) = − 2.85, p = 0.007). This effect was statistically significant even after family-wise error (FWE) correction of the p value across the five investigated acoustical features. 4-fold cross-validation analysis confirmed the robustness of this association, r(observed, predicted) = 0.34, p = 0.006
Partial correlation analyses controlling for the effect of mother’s age corroborated the association between children’s age and pitch height of mother’s voice (rpartial = − 0.33, 95% CI [− 0.56, − 0.06], t(48) = − 2.41, p = 0.02)

Summary

Materials and methods

Brightness was inferred based on the spectral centroid (center of mass) of each sample’s audio signal in the frequency domain using the get_center_of_gravity command These analyses yielded one value for each of the acoustical features per sample, which were entered into statistical analysis. We repeated the assessment of the associations between acoustical features and age without averaging across samples using linear mixed effects models which included a random intercept for each participant. We ran a number of control analyses to determine the robustness of our findings with regard to the language mothers primarily used with their children (English or Spanish), potentially differential effects for mothers of girls vs boys, potential effects of pitch height drop-off within the nonsense words, and potential effects of the chosen pitch extraction parameters. We report the lower and the upper boundary of the 95% confidence interval (CI) for the parameter of interest (i.e. r or βstd)

Results

Discussion

Methods