The impact of child age and sentence-type on the acoustics of infant-directed speech

• In Event: Poster Session 04
  In Poster Session: PS 04 Topic - Language, Communication

Abstract

Infant-directed speech is important for early language acquisition. Cross linguistically, infant-directed speech is characterized by exaggerated prosodic contours, including higher mean pitch and expanded pitch range (Fernald & Kuhl, 1987; Fernald et al., 1989). Research has demonstrated differences in the characteristics of infant-directed speech (IDS) based on age (Stern et al., 1983) and sentence-type (Geffen & Mintz, 2017) but has not examined the two factors together. Stern et al. (1983) examined longitudinal changes in maternal speech directed to their infants from the neonatal period until 24 months. Acoustic features included F0 contours (e.g. final pitch change, F0 range) and duration of vocalizations and pauses. As infants got older, pitch contouring (including Max F0 and F0 range) began to narrow and vocalizations became longer. Prosodic differences also exist between different sentence types, with yes/no questions differing from statements and wh- questions in their pitch contour (e.g. mean F0, Max F0) and intensity in the final syllable (Geffen & Mintz, 2017). However, research has not examined how the global changes in infant directed speech may be reflected in specific sentence types. The current study evaluates whether the acoustics of IDS differ as a function of child's age and sentence-type. We focus on 9 and 12-month-olds, which allows for a comparison of prelinguistic infants with those just starting to produce one-word utterances.
The study combines two corpora of native-English adult speakers. Both Corpus1(9mo) (described in Geffen & Mintz, 2017), from the Brent corpus of the CHILDES database (Brent & Siskind, 2001) and Corpus2 (12mo) (described in Thompson, 2019) from naturalistic home-recordings, included statements, yes/no and wh-questions, and were acoustically coded in Praat. Three 2-way mixed ANOVAs with Age (9 and 12 months; between-subjects factor) and Sentence-Type (S, WH, YN; within-subjects factor) were carried out on OverallF0range, FinalVowelDuration, and FinalVowelF0range. OverallF0range showed a main effect of Age,F(1,278)= 923.83, p < .001, and Age X Sentence-Type interaction, F(2,556)=4.77, p = .009. Post-hoc analyses showed that statements exhibited larger F0 range than wh-questions for young infants, t(96)=2.21, p=.029. FinalVowelDuration also had a main effect of Age, F(1,287)=100.38, p<.001, with younger infants hearing lengthened final vowels.
Results demonstrated a developmental shift in acoustic characteristics of IDS, with more exaggerated prosody to younger infants, supporting Stern et al. (1983) and suggests that IDS to older children no longer privileges prosody as strongly. The interaction between developmental shift and sentence type was only present as a global acoustic feature (i.e., OverallF0range) rather than localized ones (e.g. FinalVowelDuration). This variation by sentence type may help infants highlight important prosodic information given the link between prosody and syntax in distinct sentence types (e.g. yes/no questions are characterized by AUX-inversion and utterance-final rise). The younger infants may in turn bootstrap from these acoustic features as they integrate semantic, syntactic, and prosodic information during online language processing. Future studies should investigate if and how these acoustic features can differentially impact infants’ language learning.

Group Authors

Susan Geffen, Tianlin Wang

Authors

• Susan Geffen, Occidental College
  Presenting Author

• Tianlin Wang, State University of New York at Albany (SUNY Albany)
  Non-Presenting Author