A Cross-Cultural Examination of Child Directed Speech Across Development

CASILLAS, M.; BUNCE, J.; KIRBY, J.; SODERSTROM, M.; ROSEMBERG, C.; STEIN, A.; BERGELSON, E.

Baltimore, Maryland

Congreso; 2019 SRCD Biennial Meeting; 2019

Society for Research in Child Development

Several characteristics of child-directed speech (CDS) have long been proposed to support language development (e.g. Golinkoff et al., 2015). More frequent exposure to CDS is a strong predictor of children?s later linguistic competencies (e.g., Rowe, 2008; Schneidman & Goldin-Meadow, 2012). However, the amount of CDS children hear varies across cultural and demographic factors, including socioeconomic status (Hoff, 2006).Recent work exploring the quantity of CDS and adult-directed speech (ADS) heard by children acquiring North American English (NAE) found that 3?20-month-olds heard an increased proportion of CDS with age (Bergelson et al., 2018). Intriguingly, this increase was due to the quantity of CDS remaining relatively constant across age, while quantity of ADS decreased. One interpretation proposed by the authors was that adults more actively divide their time between child-rearing activities and adult-related activities for older children, and may be more likely to converse with other adults away from the (increasingly independent) child. Alternatively, the reduction in ADS may have arisen from technical issues related to their use of LENA for talker classification and segmentation.In the current analysis, we explore factors related to the decrease in ADS, replicating these findings with NAE using manual segmentation and talker classification, and expanding analyses to include other language communities.Our data come from 58 children ages 2?36 months from the ACLEW dataset (Bergelson et al., 2017; 30 children acquiring NAE, 10 UK English, 8 Argentinian Spanish, and 10 Tseltal (Mayan)). The dataset includes daylong audio recordings capturing the natural language environments. For each recording, we analyzed all of the speech the children encountered during 10 randomly selected 2-minute segments (Tseltal: nine 5-minute segments) in which we annotated the speaker gender, age, and addressee (e.g., child, adult) for each utterance. We calculated the minutes per hour of CDS, ADS, and all speech.First stage analyses indicate that the overall language input children heard across samples, child age, and culture was quite variable (Fig 1). We found similar patterns of CDS and ADS quantity to those reported in Bergelson and colleagues? (2018) study: overall rates of CDS were fairly similar across groups and remained consistent across development. Changes across development in the proportion of CDS appear to be driven by changes in the overall rate of ADS. Most groups reflected the Bergelson et al finding of a decrease in ADS over development. Ongoing annotation will permit a finer-grained analyses of whether sub-group differences (e.g. Argentine Spanish, cf Figure 1) are robust and meaningful. Further analyses will examine the influence of other factors, such as speaker gender, number of speakers, and maternal education.These preliminary findings indicate that across several languages and samples, the quantity of CDS remains relatively stable over the first 36 months. They further undergird results based on LENA algorithm output, replicating the overall pattern with more precise manual annotations of raw audio recordings. Summarily, changes in the proportion of CDS young children hear over developmental time appear to be driven by changes in the quantity of ADS in their input.