The words from children to children: a study of the linguistic environment of children from diverse sociocultural groups in Argentina

ALAM FLORENCIA; ROSEMBERG CELIA RENATA; GARBER, LEANDRO; AMDEM PABLO

Congreso; MPAL; 2023

Max Plank

A series of studies from the last few decades has shown that during the first years of life the characteristics of the at-home linguistic environment have a significant impact on young children’s  language acquisition (Fernald, Marchman & Weisleder, 2013). Despite the fact that different studies have indicated that children from different social groups and cultures tend to spend several hours a day with their siblings (Oshima-Takane, Goodz & Derevensky, 1996), most studies analyzing the linguistic environment of children under 2 years of age have focused on the mother's speech production in dyadic interactions (Hoff, 2003; Rowe, 2012). In particular, research conducted with indigenous and rural populations has shown that older children perform caregiving and socialization tasks for young children (Maynard, 2002) and that young children's interactions are in fact not dyadic but involve multiple participants of various ages (Ochs & Schieffelin 1984; De León 1998). Only a few recent works  have addressed the extent to which the language produced  by children of different ages contribute to shaping the at-home linguistic environment of young children. Results showed that children from rural and indigenous communities hear more speech from other children than their peers of urban residential families (Loukatou et al., 2022; Scheinedman & Goldin-Meadow, 2012). However these studies were carried out between extreme groups (urban-middle income vs indigenous rural communities) from different countries. This work aims to analyze the linguistic environment in the households of a socioeconomically and socioculturally diverse sample of Argentinian children (up to 4 years old), considering the proportion of speech they heard from other children (until 12 years old), and its quantitative and some qualitative characteristics (the quantity and diversity of words and the amount of questions addressed to the child and overheard). The corpus comprises audio recordings of spontaneous speech from 46 children living in residential urban areas, 16 children from urban marginalized areas, 21 children from indigenous marginalized communities in semi urban areas. Three  recordings of 4 hours of each child's linguistic environment were carried out over the course of one year (984 total hours of recordings) starting at 14 months. The 2 middle hours of the recordings of 43 children (9 from indigenous marginalized communities in semi urban areas, 12 from urban marginalized areas, and 22 from urban residential households)  were transcribed in CHAT, and analysis of quantity and diversity of words and amount of questions addressed to the child and overheard were carried out with CLAN program (MacWhinney, 2000). Transcriptions were only made for 28% of the corpus as they demand a great amount of human and economic resources, so in order to analyze the proportion of speech children hear from other children in the 984 hours of recordings we employ a pre-trained diarization model that, given an audio recording, classifies each frame into the type of speaker (adult male, adult female and child) (Lavechin et al., 2020). Beta regressions were used to test differences between the groups. Results show that children from impoverished indigenous communities in semi rural areas hear more amount of speech (automatic diarization analysis) quantity and diversity of words, and more amount of questions (manual analysis over the transcriptions), both addressed to them and overheard from other children than their peers from the other socio-cultural groups. Preliminary findings on longitudinal variations show a similar pattern. These findings contribute to the description of the linguistic characteristics to which children of different socio-cultural groups are exposed.