Scientists continue to discover more about language acquisition from an early age. A new study examines how young children use different information sources to learn new words.These sources could include everything, from whether they've ever seen an object before (which indicates whether it has a name they have heard before) to what they may be talking about when a new word appears.Researchers created a cognitive model to help understand how these sources interact. They also proposed a social inference approach, where children use all of the information available to infer the identity and purpose of an object.Michael Henry Tessler, a developmental psychologist at the Massachusetts Institute of Technology (MIT), says that you can think of the model as a small computer program. "We input children’s sensitivity to different types of information. This is measured in separate experiments. The program then simulates what would happen if these information sources were combined in a rational manner.""The model predicts what will happen in hypothetical new situations where all information sources are available."Research in philosophy, developmental psychology, and Linguistics informed the theoretical system that researchers created. To assess the sensitivity of children aged 2-5 years to different information sources, data were also collected from tests. These data were then used to build the model.After generating predictions, the researchers ran real-world experiments on 220 children to test their ability to infer meanings of words like duck, apple and pawn when they were presented with the objects on a tablet screen.Children were provided with a variety of cues about the relationships between words, objects. This included a voiceover by a presenter and a mix of labels they wouldn't or could have known. The researchers were able to test three sources of knowledge: their previous knowledge, cues provided by the presenter, as well as context within a conversation.Children see a portion of the word app test. (Frank and al, Nature Human Behaviour 2021The results of the last experiments showed that the model approach was very similar to the results. This suggests that children use these information sources in predictable and quantifiable ways as they build their vocabulary.Tessler says that computational modeling has the virtue of allowing you to articulate alternative hypotheses and alternative models with different internal wiring. This allows you to determine if other theories make better predictions.These results suggest that there are other hypotheses that can be dismissed: for instance, that some information sources are not used or that the way that sources are processed changes with age.The research provides a mathematical perspective on how language learning occurs in children. However, it is still early days for this approach. More studies will be required with larger numbers of children to further develop the idea.It is amazing to see how we can go from knowing only a handful words to knowing many thousand in just a few years. This knowledge can be used to inform everything, from therapy to teaching.Manuel Bohn, a developmental psychologist from Germany's Max Planck Institute for Evolutionary Anthropology, says that children learn words in real life in complex social situations in which multiple types of information are available."They must use their vocabulary to interact with the speaker." Word learning requires the integration of multiple information sources.Nature Human Behaviour published the research.
