Statistical analyses

Data were checked for normality using the Shapiro–Wilk test. Waveform morphology of responses to different narrators was compared using Pearson correlations of the responses between 0 and 15 ms for the ABR waveforms or 0 and 40 ms for both ABR and MLR waveforms. The Wilcoxon signed-rank test was used to determine whether narrator differences (waveform correlations) were significantly different than the correlations of the same EEG split into even and odd epochs with equal numbers of epochs from each narrator. For experiment 1, responses containing an equal number of epochs from the first and second half of the recording had a median (interquartile range) correlation coefficient of 0.96 (0.95–0.98) for the 0–15 ms lags, indicating good replication.

The intraclass correlation coefficient type 3 (absolute agreement) was used to verify good agreement in peak latencies chosen by an experienced audiologist and neuroscientist (MJP) at two different time points, 3 months apart. The ICC3 for all peaks were ≥0.90, indicating excellent reliability for the chosen peak latencies.

Independent t-tests with μ = 0 were conducted on the peak latency differences of ABR/MLR waves for unaltered and broadband peaky speech. For multiband peaky speech, the change in peak latency, τ, with frequency band was modeled using a power law regression (Harte et al., 2009; Neely et al., 1988; Rasetshwane et al., 2013; Strelcyk et al., 2009) according to the formula

where

and where f is the band center frequency normalized to 1 kHz (i.e., divided by 1000), τsynaptic is the synaptic delay (assumed to be 0.8 ms; Eggermont, 1979; Strelcyk et al., 2009), and τI−V is the I–V inter-wave delay from the subjects' responses to broadband peaky speech. The power law model was completed for each filter cutoff of 30 and 150 Hz in the log–log domain, according to the formula

and using linear mixed effects regression to estimate the terms d and b (calculated as b = 10^intercept). For subjects who did not have an identifiable wave I in the broadband peaky response, the group mean I–V delay was used – this occurred for 1 of 22 subjects in experiment 1 and 2 of 11 subjects for responses to the female narrator in experiment 2. Because only multiband peaky speech was presented in experiment 3, the mean I–V intervals from experiment 2 were used for each subject for experiment 3. Random effects of subject and each frequency band term were included to account for individual variability that is not generalizable to the fixed effects. The mixed effects regressions were performed using the lme4 (RRID:SCR_015654) and lmerTest (RRID:SCR_015656) packages in R (RRID:SCR_001905) and RStudio (RRID:SCR_000432) (Bates et al., 2015; Kuznetsova et al., 2017; R Development Core Team, 2020). A power analysis was completed using the simR package (RRID:SCR_019287; Green and MacLeod, 2016), which uses a likelihood ratio test on 1000 Monte Carlo permutations of the response variables based on the fitted model.