Who are you talking to like that? Exploring adults' ability to discriminate child- and adult-directed speech across languages

AbstractChild-directed speech (CDS) shows similar characteristics across many languages, but is known to vary across cultural and demographic groups (Lieven, 1994). Is CDS consistently discriminable from adult-directed speech (ADS) despite these differences? Perhaps: adults listening to scripted female CDS can discriminate ADS-vs-CDS in a language they don’t speak (Bryant et al., 2012). We build on this finding by asking North American English speakers to classify utterances from the natural language input of 10 Tseltal Mayan children as ADS or CDS (n = 1836 utterances). Binomial mixed-effects regressions of accuracy show that listeners are more accurate on utterances from females (mFemale = .81, mMale = .67) and adults (mAdult = .82, mChild = .72), with a larger gender effect for child speakers (m: ΔGirl-Boy = 0.31, ΔWoman-Man = 0.09). This suggests that (a) ADS-CDS discrimination of natural speech in an unrelated, non-familiar language is reliable (mAll = 0.78) and also (b) modulated by speaker type.

Return to previous page