Testing the relationship between IDS preference and vocabulary development: An MB1 Follow-up study

• In Event: Updates from ManyBabies: Collaborative efforts to promote best practices and improve robustness in developmental science

Abstract

Decades of research support the idea that infant-directed speech (IDS) supports language development. Under laboratory conditions, IDS can facilitate word segmentation (Thiessen, Hill & Saffran, 2005), word recognition (Singh, Nestor, Parikh, & Yull, 2009), and word learning (Graf Estes & Hurley, 2013). Use of caregiver IDS (but not adult-directed speech, ADS) in children’s everyday experience predicts enhanced language development (e.g., Ramirez-Esparza, Garcia-Sierra and Kuhl, 2014; Shneidman, Arroyo, Levine & Goldin-Meadow, 2013; Weisleder & Fernald, 2013). One commonly suggested mechanism for the impact of IDS on language development is via attentional mechanisms - infants will attend more to speech with IDS properties. In support of this idea, infants show preferential listening to speech in the IDS register compared with speech in ADS (The ManyBabies Consortium, 2020).
However, to date this preference for IDS has not been robustly and directly tied to language outcome measures like vocabulary knowledge. One study (N = 36) found that individual preferences for IDS over ADS between 6 and 12 months of age predicted expressive language outcomes at 18 months, at least in typically developing infants (siblings of children with autism did not show this association; Droucker, Curtin, & Vouloumanos, 2013). The ManyBabies dataset provides a unique opportunity to build on this finding with a larger and more diverse sample of typically developing infants.
The original data contributors of ManyBabies 1 were invited to contribute to a follow-up study in which caregivers of the original participants (tested between age 3 and 15 months) reported on their child’s productive vocabulary at 18 and/or 24 months using a Communicative Development Inventory. Nine laboratories contributed data from North American families (N = 228), 4 laboratories from UK families (N = 88), and 8 laboratories from non-English-speaking families (N = 189).
In pre-registered analyses, we examined the relationship between infants’ IDS preference (a standardized difference score between their mean looking time to IDS vs. ADS ranging from -1.6 to 1.6) and vocabulary size (standardized CDI score in the North American sample, raw vocabulary size otherwise, see Fig. 1). We tested the possible impact on this relationship of age at preference measurement and at vocabulary size measurement, controlling for effects of gender or method of IDS-preference testing.
We conducted separate mixed-level analyses for the UK and North American samples. In both the UK and North American samples, we found that infants’ IDS preference did not significantly correlate with their later vocabulary size [ps > 0.3], even after controlling for infants’ gender and their age at preference testing and at vocabulary measurement. The relation between preference and vocabulary size did not significantly interact with infants’ age (both at time of preference and time of vocabulary measurement) [ps > 0.5]. Ongoing analyses are testing the statistical support for these findings via a Bayesian approach, and possible cross-linguistic effects. However, our findings do not support the theorized relationship between IDS preference as measured in the laboratory and later vocabulary via parental report, possibly because individual-level data may not be sufficiently reliable (Schreiner et al., in prep.).

Group Authors

The CDI Follow-up Group

Authors

• Melanie Soderstrom, University of Manitoba
  Presenting Author

• Arthur Capelier-Mourguy, Lancaster University
  Non-Presenting Author

• Sin Mei Angeline Tsui, Stanford University
  Non-Presenting Author