6 Online databases of child language
The availability of large collections of written and spoken material stored online has been of enormous benefit to the study of child language acquisition. Instead of relying on how adults think children speak, we have been able to use corpora to study in great detail how they actually speak, from languages across the world. It has provided a solid empirical base against which to test our theories of language acquisition as well as provide rich data for computational models to train on and simulate.
The Child Language Data Exchange System (CHILDES) is a repository of child and adult speech containing millions of words and thousands of hours’ worth of speech from typically developing and clinical populations. Often recorded at home, these samples represent naturalistic language use as children play, eat and perform other everyday activities. Wordbank is a database of children’s vocabulary development that includes data from 92,771 children, 42 languages and 105,290 administrations of the Communicative Development Inventory (CDI) – a collection of parent-report questionnaires about language ability. Both databases are open access, are being added to regularly, and it is well worth browsing the data for yourself, trying the interactive analyses to see what you can find out, which you will do in Activity 2.
Activity 2 A web of words and meaning
When children are learning a language, they are building a network of words and relationships between words. As they learn more about language their language networks grow. Researchers have simulated these networks online for us to explore. You’ll now have a look at one using Workbank. Access the link in a new tab or browser.
As you can see there are lots of words connected by links. The ‘Age of acquisition’ slider button is the average months-of-age that children acquire those words.
As you explore this app, investigate two questions:
- What happens to the parts of speech (e.g., nouns, verbs, adjectives, function words) as you change the ‘age of acquisition slider’ from 6 to 36 months of age?
- What do you notice about words that cluster together in the network?
Discussion
- The complexity of the network grows over time, as children learn to use more grammatical categories like noun, verb and adjective in increasingly varied ways. This generates many paths through the network making it resilient and efficient system and it also allows children to use language in increasingly flexible and expressive ways.
- Ideas that are more related to one another cluster together in the network. For example, ‘hair’ is closely related in the network to ‘nose’, ‘mouth’, and ‘eye’, and ‘grandma’ is closely related to ‘mommy’, ‘daddy’, and ‘grandpa’ in another cluster. Scientists believe the mind and brain stores language like a giant thesaurus, where words that close in meaning ‘activate’ others nearby.