UC Berkeley Psychology Professor Mahesh Srinivasan began his journey in academia after receiving his bachelor’s degree in Symbolic Systems from Stanford and his Ph.D. in Developmental Psychology from Harvard. Throughout his career at Berkeley, Professor Srinivasan has taught and inspired many undergraduate and graduate students.
Director of the Language and Cognitive Development Laboratory, Professor Srinivasan employs empirical and computational approaches to investigate the emergence and interaction of linguistic, cognitive and social skills during human development. His research spans various cultural contexts to gain a comprehensive understanding of these processes. Most recently, Dr. Ruthe Foushee (Assistant Professor at the New School for Social Research and former UCB PhD student) and Professor Srinivasan studied 5 to 15-month-old infants from a Tseltal Mayan community in Mexico. The results of their research, published in the journal Proceedings of the National Academy of Sciences, challenge the prevalent idea that babies primarily develop language from the speech that caregivers direct to them (as opposed to the speech they overhear).
Professor Srinivasan spoke to Berkeley Social Sciences recently about his most recent research and what is next for him. His interview is edited for clarity.
Tell us more about your research?
A great deal of research has suggested that the language that caregivers direct to their young children is a critical driver of their children’s language learning and cognitive development. This conclusion has driven interventions globally to encourage parents to speak more to their children. Yet the overwhelming majority of this research has been conducted in Western, child-centered contexts in which the parental practice of speaking to young children is common. Moreover, some have suggested that children may develop language on similar timetables across environmental contexts, including those in which child-directed language is scarce. One possible explanation for the robustness of language development across variable environments is that children are able to learn not just from language directed to them, but also from language which has been directed to others, that they overhear.
Our study examined this hypothesis with infants in a Tseltal Mayan community in Chiapas, Mexico. Infants in this community are rarely spoken to, yet have the opportunity to overhear a great deal of other-directed language by virtue of being carried on their mothers’ backs. We conducted two studies with these infants to measure their early knowledge of words in the Tseltal language. We reasoned that if these infants rely exclusively on the small amounts of language directed to them to learn new words, they should not exhibit a level of word knowledge analogous to their U.S. peers, who are much more frequently spoken to. Yet in fact, our studies documented considerable word knowledge among these infants, and provided evidence that they may have derived this knowledge in part via overhearing.
What specific types of words did the Tseltal infants exhibit knowledge of?
We conducted two studies, utilizing a method that has been used to detect early word knowledge among U.S. infants, in which infants’ eye-gaze patterns are measured as they listen to speech. The first study found that Tseltal infants exhibited knowledge of common nouns in Tseltal (e.g., words for food like corn and squash, words for animals like chickens and cows, etc.), similar to previous demonstrations with U.S. infants. This finding alone is unexpected according to current theorizing, which predicts that Tseltal infants should exhibit minimal word knowledge, by virtue of receiving little child-directed speech. But our second study went a step further, to examine whether Tseltal infants understand words that they could only have learned from overhearing. This second study found that Tseltal infants understood honorific greeting terms (e.g., words for greeting an older woman, younger man, etc.) that, by definition, are never directed to them.
How do the findings of these studies contrast with current theories in language development?
Current research and theory holds that children primarily learn from the speech that caregivers direct to them because it is thought to have special properties that make it especially easy to learn from. Some have argued that the speech that children overhear, by contrast, holds little value for young language learners. Our findings, however, suggest that for some infants, learning from overhearing may be an important path toward developing language. From this perspective, one reason that children may be able to successfully acquire their native languages across variable contexts is because they are flexible in how they learn: They can learn not only from language directed to them, but are also able to learn from the language they overhear. It is well-known that the psychological sciences have suffered from reliance on convenience samples; our findings show how considering how language learning proceeds within understudied contexts of development can shift our understanding of the informational sources and mechanisms that children employ to learn language.
Could you please expand on the adapted gaze-tracking method used for detecting word knowledge? How does it work?
To measure infants’ knowledge of words we used an implicit, gaze-based measure that has been used to demonstrate knowledge in American infants as young as 6 months of age. On each trial, infants are presented with two images, like an image of a baby and an image of corn. The caregiver is then prompted to repeat phrases they hear (over headphones they are wearing) that refer to one of the two images (e.g., “Where is the baby?”). To measure infants’ recognition of words, we explore how much more infants fixate on a picture just after it has been named, compared to a matched trial when it has not been named. So for example, we measure how often infants look at the baby when it has been named compared to another trial where the same two pictures are shown, but the corn was named. This method effectively controls for asymmetries in any general preferences infants might have for looking at one image over another.
What are the implications of these findings for the advice often given to parents about speaking to their young children?
Our study suggests that — in a context where infants are getting a lot of exposure to meaningful linguistic interactions happening around them — infants can learn more from these interactions than has typically been appreciated in the literature. Although research has emphasized the importance of child-directed language for children’s early language development, our findings provide an example of how children may also be able to tune into and learn from the rich interactions around them, which is something that busy caregivers — who are often encouraged to speak to their infants — could keep in mind. Of course, there are reasons to think that the language that parents tailor to their young children could be quite easy for children to learn from. But our findings suggest that, even in contexts in which children do not receive much of this kind of language, they nevertheless develop early word knowledge, in part by attending to and learning from conversations around them.
Do you think that if this study was conducted under different cultural contexts the results would differ?
Yes, it is possible that the results would differ in other contexts. For example, it is possible that the experience of Tseltal infants constitutes a best-case scenario for learning from overhearing. This is because these infants are often carried on their mothers’ backs, and thus have almost a “front-row seat” to their mother’s interactions. They are consequently in a good position to hear what their mothers are saying, and to even have some visual access to the focus of their mother’s attention, which could scaffold guesses about the meaning of overheard speech. It is thus possible that infants in other contexts would learn less from overhearing.
It is also possible that Tseltal infants may be particularly adept at learning from overhearing compared to infants in other contexts. For example, it is possible that children who are frequently spoken to — as is the case for many U.S. children — may come to expect to be spoken to, and may pay comparably little attention to interactions around them. By this account, children may adapt their language learning strategies to the nature of their environment, with some children learning to learn primarily from child-directed language, and other children learning to learn primarily from language directed to others.
What is next for your research?
As noted above, an exciting question opened by our findings is whether, across different contexts, children are equally attentive to and able to learn from child-directed and other-directed language, or whether children might instead adapt their learning strategies in response to their environments. By the latter account, children raised in child-centered contexts may come to expect their attention to be managed by their caregivers and learn to ignore interactions around them; conversely, children raised in contexts in which child-directed language is rare and other-directed language is common may develop a strong ability to attend to and learn from interactions around them. We are excited to explore this adaptation hypothesis in future work. We are also interested in understanding the different kinds of other-directed language that children overhear (e.g., speech to another adult vs. another child), and how these different kinds of other-directed language support learning.
