2018 Sun Session A 0930
Sunday, November 4, 2018 | Session A, East Balcony | 9:30am
Language Input in a foragerfarmer population: Estimations from daylong recordings
C. Scaff, J. Stieglitz, A. Cristia
If children growing up in some preindustrial societies experience little adult childdirected speech from adults, then such high quality directed input might not be necessary for typical language acquisition (e.g., Ochs & Schieffelin, 2001; Shneidman, 2010). There is little quantitative data assessing whether the ifpremise is true (a summary in Cristia, Dupoux, Gurven & Stieglitz, 2017), and much of it can be critiqued on the grounds of ecological validity and/or sample sizes. We report the first ecologicallyvalid, largescale analysis of speech experiences by infants and children growing up in a foragerhorticulturalist population.
The Tsimane of lowland Bolivia are composed of extended family clusters, who share food and labor. On average, women have their first child by 19 years of age, and a total fertility rate of about 9 births (Kaplan, Hooper, Stieglitz & Gurven, 2015). Infants are kept close to their mothers, and regularly carried in a sling so that mothers can perform subsistence activities; toddlers are often cared for by older siblings or other kin. Such a setup could lead to small amounts of maternal speech directed to young children.
We recruited 15 families in one village, and fitted children under age 6 years (n=25) with daylong audiorecorders (LENA® or similar). For each recorded day, we had one minute per hour (average 13 coded minutes, range 416) for which a trained phonetician annotated each minute using Praat (Boersma, 2001). Vocalizations were classified into: the focal child, the “main female voice” (MFV, usually the mother), other female adults, other male adults, and other children. We automatically classified as potentially childdirected any vocalization that either occurred next to a vocalization by the focal child, or which did not have a neighboring vocalization by some other speaker (i.e., monologues). Finally, we summed all the speech produced by anyone but the key child across coded minutes, and normalized to estimate average number of speech minutes per hour.
We found that young Tsimane children experienced little childdirected speech, with much higher levels of overheard input, and nonsignificant age effects [see Figure 1]. Most input was attributed to MFV or peers, rather than other female or male adults [see Figure 2]. There was a decrease of overall input from the MFV [ß=8.34 (3.85), p=0.04], with no other significant changes as a function of the target child’s age.
Thus, the overall pattern fits well with Ochs & Schieffelin description of a rich linguistic environment early on, composed primarily of thirdparty conversations, which raises many questions. Does this mean that adult childdirected input is less important than thought, or is it the case that Tsimane children acquire language more slowly? We look forward to addressing these questions (and gathering data to check the validity of the methods above) in a field trip planned for July 2018. If the patterns above hold but Tsimane children are not obviously delayed, this would suggest that even small quantities of childdirected input from adults suffice, with important implications both for theorybuilding and applied science.
References
Ochs, E., & Schieffelin, B. (2001). Language acquisition and socialization: Three developmental stories and their implications. Linguistic anthropology: A reader, 2001, 263301.
Shneidman, L. A. (2010). Language input and acquisition in a Mayan village (Unpublished doctoral dissertation). The University of Chicago.
Cristia, A., Dupoux, E., Gurven, M., & Stieglitz, J. (2017). Child Directed Speech Is Infrequent in a Forager Farmer Population: A Time Allocation Study. Child development.
Kaplan, H., Hooper, P. L., Stieglitz, J., & Gurven, M. (2015). The causal relationship between fertility and infant mortality. In P. Kreager, B. Winney, S. Ulijaszek, & C. Capelli (Eds.), Population in the human sciences: Concepts, models, evidence (p. 361). Oxford University Press.
Boersma, Paul (2001). Praat, a system for doing phonetics by computer. Glot International 5:9/10, 341345