Quantifying Infants’ Movements to Maternal Singing - a Comparative Methodological Approach

Authors

  • Zoe Viola Bergmann University of Vienna
  • Trinh Nguyen University of Vienna

Abstract

Rhythmic movements in infants are motivated by music; but also, they are found to offer practical training for performing coordinated and precisely timed movements, which are also necessary for speech [1]. New research reveals that increased rhythmic movements and high neural tracking of live maternal infant-directed (ID) play songs relate to infants’ linguistic development [2]. Nguyen et al. [2] measured neural and behavioral responses in 7-month-old infants to their mothers’ singing of play songs and lullabies as well as the vocabulary size at 20 months. To study infants’ behavior, commonly their movements are manually labeled as infants often dislike wearing markers, and motion tracking which can be applied successfully on adults is tricky to be transferred to babies due to their different proportions. Likewise, manual labeling is time-consuming and prone to subjectivity. New advances in animal research to quantify the behavior of small organisms bear great potential to allow studying the movements of infants. DeepLabCut (DLC), a tool for markerless pose estimation building upon a deep learning algorithm, aims to overcome limitations imposed on experiments, that use markers that ought to be expensive and distracting [3]. This research project aims at replicating and refining Nguyen et al. [2] study results which relied on two independent human coders for their behavioral data analysis. Less than a handful of studies have been published that apply DLC to study infants’ behavior and no study to date presents the results of two independent behavioral analyses, comparing the methods of manual human coding and DLC on the same dataset. Such comparative approaches and repeated analyses are effortful and often inaccessible to researchers who work under the diverse constraints of their resources. This research project bears the potential to inform future decisions of researchers in the field when it comes to designing their experiments and opting to find the most suitable method of analysis. Analyzing the dataset again, using a novel method, gives rise to potentially concluding new best scientific practices in the analysis of behavioral data of infants.

References 

[1] J. M. Iverson, "Developing language in a developing body: The relationship between motor development and language development," Journal of Child Language, vol. 37, no. 2, pp. 229-261, 2010.

[2] T. Nguyen, S. Reisner, A. Lueger, S. V. Wass, S. Hoehl, and G. Markova, "Sing to me, baby: Infants show neural tracking and rhythmic movements to live and dynamic maternal singing," Pre-Print, bioRxiv 2023.02.28.530310, Feb. 28, 2023.

[3] A. Mathis, P. Mamidanna, K. M. Cury, T. Abe, V. N. Murthy, M. W. Mathis, and M. Bethge, “Deeplabcut: Markerless pose estimation of user-defined body parts with deep learning,” Nature Neuroscience, vol. 21, no. 9, pp. 1281–1289, 2018.

Published

2023-06-05