Examining the Temporal Features of Infants’ Cry and Speech-like Vocalizations and Associations with Adult Input in an Early Intervention Classroom

• In Event: Poster Session 10
  In Poster Session: PS 10 Topic - Language, Communication

Abstract

Infants’ early environments are replete with statistical regularities and temporally structured information that shapes everyday learning. One metric to assess temporal patterning over longer timescales is the burstiness coefficient, which quantifies the spacing of behavioral events over time and provides an index of event regularity (Goh & Barabási, 2008). Day-long recordings of infant-caregiver interactions in the home revealed that adult speech vocalizations exhibited the highest degree of burstiness followed by infant cry and speech-like vocalizations, respectively (Abney et al., 2017). Still unknown, however, is the extent to which the burstiness of adult vocalizations in contexts outside of the home, including in early education settings, influences the burstiness of infants’ own vocalizations. Further, we sought to understand whether the burstiness of infants’ vocalizations is modality-dependent or a stable property of the child. Understanding whether burstiness is a stable property of the child would allow us to make future predictions about the Here, we longitudinally assessed the burstiness of infants’ cry and speech-like vocalizations in an early intervention classroom and examined the relationship between the burstiness of infants’ vocalizations and the burstiness of their input from teachers.

We conducted 29 day-long observation of 16 infants in an early intervention classroom. Audio from individual infants was recorded using Language ENvironment Analysis (LENA) recorders worn by each child in a vest. We employed LENA’s pattern recognition software to distinguish between infants’ cry and speech-like vocalizations and distinguish infants’ vocalizations from adults. To quantify the temporal structure of classroom vocalizations, we calculated separate burstiness coefficients (Figure 1) for infants’ cry and speech-like vocalizations as well as adult speech vocalizations. Burstiness measures the distribution of time intervals between consecutive vocalization onsets on a scale of -1 (periodic) to 1 (bursty), with 0 indicating randomly spaced vocalizations. A bursty distribution is characterized by clusters of vocalizations followed by long periods with few to no vocalizations.

Both infants’ cry (B=.75) and speech-like vocalizations (B=.76) were highly bursty, meaning that infants’ vocalizations often clustered together in time rather than occurred periodically over the day (Figure 2). Cry and speech-like vocalizations were highly correlated with each other, (r=.789, p<.001), and did not exhibit change in their degree of burstiness over time (cry: χ2(1)=.74, p=.39; speech-like: χ2(1)=.06, p=.80). Next, we examined whether the burstiness of infants’ vocalizations was associated with teacher input. Multilevel models indicated that the burstiness of infants’ speech-like vocalizations was predicted by the burstiness of adult speech, (χ2(1)=4.64 , p=.031), whereas cry burstiness was not (χ2(1)=2.80 , p=.09).

Infants’ vocalizations in the classroom are highly clustered in time, suggesting that burstiness may be a distinct property of infants’ early communication. We observed differential associations between the burstiness of infants’ cry and speech-like vocalizations and the burstiness of their input from teachers, suggesting that the temporal patterning of teacher vocalizations may rely on the communicative value of infants’ vocalizations. Together, these demonstrate how objective measurements of classroom vocal activity can reveal subtle patterns pertinent to infant’s developing vocal interactions.

Authors

• Stephanie Custode, University of Miami
  Non-Presenting Author

• Samantha Mitsven, University of Miami
  Presenting Author

• Regina Fasano, University of Miami
  Non-Presenting Author

• Lynn K Perry, University of Miami
  Non-Presenting Author

• Guangyu Zeng, University of Miami
  Non-Presenting Author

• Daniel Messinger, University of Miami
  Non-Presenting Author