The purpose of the present study was to introduce a computational simulation of timing in child-adult interaction. The simulation uses temporal information from real adult-child interactions as default temporal behavior of two simulated agents. Dependencies between the agents’ behavior are added, and how the simulated interactions compare to real interaction data as a result is investigated. In the present study, the real data consisted of transcriptions of a mother interacting with her 12-month-old child, and the data simulated was vocalizations. The first experiment shows that although the two agents generate vocalizations according to the temporal characteristics of the interlocutors in the real data, simulated interaction with no contingencies between the two agents’ behavior differs from real interaction data. In the second experiment, a contingency was introduced to the simulation: the likelihood that the adult agent initiated a vocalization if the child agent was already vocalizing. Overall, the simulated data is more similar to the real interaction data when the adult agent is less likely to start speaking while the child agent vocalizes. The results are in line with previous studies on turn-taking in parent-child interaction at comparable ages. This illustrates that computational simulations are useful tools when investigating parent-child interactions.
Cite as: Marklund, E., Pagmar, D., Gerholm, T., Gustavsson, L. (2017) Computational Simulations of Temporal Vocalization Behavior in Adult-Child Interaction. Proc. Interspeech 2017, 2208-2212, doi: 10.21437/Interspeech.2017-1289
@inproceedings{marklund17b_interspeech,
author={Ellen Marklund and David Pagmar and Tove Gerholm and Lisa Gustavsson},
title={{Computational Simulations of Temporal Vocalization Behavior in Adult-Child Interaction}},
year=2017,
booktitle={Proc. Interspeech 2017},
pages={2208--2212},
doi={10.21437/Interspeech.2017-1289}
}