Recording procedure
The neural responses to a click stimulus and to the consonant-vowel /da/ were recorded while the neonates were naturally sleeping in their cradle. A vertical montage was used with the active electrode located above nasion at the 10% of the distance from nasion to inion59 (Fpz according to the International 10-20 EEG system), the ground electrode placed between Fpz and nasion, and the reference electrodes placed at the mastoids. According to previous studies39,49,51-53, only the right mastoid (ipsilateral reference) was used. Stimulus presentation, EEG recording and data repository were carried out by the portable SmartEP equipment, including the cABR and Advanced Hearing Research modules (Intelligent Hearing Systems, Miami, Fl, USA). Firstly, to confirm auditory pathway integrity, two blocks of 2000 ABRs to a click stimulus (100 µs square; 60 dB SPL) were recorded. Then, the FFR was collected following the standards published in our previous studies39,53. Four blocks of 1000 artifact-free responses were recorded to the consonant-vowel /da/. The speech stimulus was created using a Klatt-based synthesizer60 and modified in Praat61. The stimulus duration was 170 ms (10 ms onset period, 47 ms for the consonant transition and 113 ms for the vowel region) with a constant fundamental frequency (F0) of 113 Hz. During the consonant transition, the first and second formants (F1, F2) varied from 553 to 688 Hz and 1438 to 1214 Hz, respectively, remaining stable in the vowel region. The speech stimulus was presented at 60 dB SPL every 100.27 ms in alternating polarities though ER3C shielded earphones of 300 Ω (Etymotic Research Inc., Elk Grove Village, IL, USA).
The continuous EEG sampled at 13333 Hz was online bandpass filtered from 30 to 1500 Hz and epoched in sweeps of 270.27 ms (-40.95 ms to 229.32 ms). Any sweep containing activity > ±30 µV was rejected during the recording. Less than 3% of sweeps were rejected in click or /da/ stimuli blocks and impedances were kept <8 kΩ throughout the entire recording. Responses were averaged for each stimulus and neonate and the subsequent evoked potentials were offline bandpass filtered from 80 to 1500 Hz.