The Social Origins of Vocal Sequence Learning: a Comparative Approach

• In Event: The Diversity of Social Contingency and its Contributions to Social and Communicative Development

Abstract

Across many species that communicate acoustically, distinct vocal elements or syllables are often strung together, one after another, forming vocal sequences that are foundational to vocal communication. A developmental challenge for both songbirds and humans is to learn to produce vocal sequences quickly and efficiently, in terms of producing sequences with more syllables per second. Here, we provide comparative evidence that immature human and avian learners use social guidance to facilitate more efficient vocal communication. In both taxa, we observed: 1) a positive relationship between adult feedback to immature vocalizations and the development of efficient vocal sequences, 2) a positive relationship between vocal sequence efficiency and communicative maturity.
30 caregiver-infant dyads participated when infants were 5 (M=5.32) and 10 months of age (M=9.74). Dyads engaged in 15-minute unstructured free-play sessions. Caregivers were asked to play as they normally would at home. Audio-visual recordings of infants’ vocalizations and caregivers’ speech during free play were obtained. Infants’ vocalizations were segmented into vocal sequences if they were bounded by silent gaps longer than 1 second in duration. For each infant vocalization, we determined whether it elicited a caregiver response. Caregiver responses were defined as any verbal behavior which occurred within 2 seconds after infants’ vocalizations. Caregivers completed the Communicative Development Inventory (CDI): Words and Gestures when infants were 18 months old to assess infants’ word production (Fenson et al., 1994).
Nine juvenile zebra finch were raised by their parents until 35 days post-hatch (dph) and thus had access to their father’s (i.e., tutor’s) song during their sensory song memorization phase. Juveniles then received video playback responses to their song (1 hour/day) from 35-60 dph. During the playback sessions, each time the bird sang an experimenter triggered video of an adult female appearing to “fluff-up”. Fluff-ups are elicited by complex, attractive song (Vyas, et al., 2009). Songs were recorded and assessed at 45 and 90 dph for syllable rate and acoustic similarity to tutor (Tchernichovski, et al., 2000).
In both human infants and zebra finch, vocal sequences compress over development (Figure 1). The effect was a function of contingent social feedback. The degree to which human caregivers responded to infants' vocal sequences at 5 months predicted the extent of vocal sequence compression over development (Figure 2A). The amount of compression at 10 months predicted productive vocabulary at 18 months (Figure 2B). In zebra finch, the extent to which their immature song received contingent fluff-ups predicted their syllable compression at 90 dph (Figure 2B). Their compression rate predicted overall similarity to their tutor's song (Figure 2D).
Socially guided vocal learning is a crucial learning mechanism of vocal development, with strong parallels across humans and songbirds (Carouso-Peck & Goldstein, 2019). Our findings point to a parallel learning mechanism – the perception of social contingency on immature vocalizations - underlying vocal development across two distantly related species. Thus, our study provides evidence of a shared social mechanism for vocal development in both songbirds and humans.

Authors

• Steven L Elmlinger, Cornell University
  Presenting Author

• Samantha Carouso-Peck, Cornell University

• Annalise H. Wilk, Cornell University

• Michael H. Goldstein, Cornell University