Young children’s speech: from a simple phonological rule to the murky phonetic reality

Abstract

Word productions of children between 1;0 and 3;0 often deviate phonologically from adult targets. This paper revisits a textbook case of child phonological development — the neutralization of the voicing contrast in onset stops — and shows how it becomes both messier and more interesting once acoustic measurements are allowed into the picture. Strikingly, the direction of this neutralization appears to differ between Dutch- and English-acquiring children: Dutch children favor voiceless realizations, while English children voiced ones. Acoustic measurements explain why Dutch- and English-acquiring children seem to neutralize the contrast in opposite directions, while longitudinal measurements (all measurements using Praat of course) from a Dutch-acquiring child uncover a U-shaped VOT trajectory and, tucked inside it, a covert contrast that impressionistic transcription alone would have missed entirely. A follow-up production experiment shows that toddlers can actively enhance this contrast in a minimal-pair context, hinting at real phonological knowledge beneath the phonetic noise. The moral: categories exist from early on in developing speakers, but are realized by gradient, acoustically subtle means, a conclusion that is right up Paul Boersma’s alley.