Why toddlers suddenly learn to talk.

Kids become chatterboxes within months of barely being able to speak a few words. How come?

Children do not need any specialised learning to suddenly improve their vocabularies. Instead, their behaviour can be described by a simple mathematical rule of thumb.

Parents become familiar with the so-called “word spurt”, the slightly disconcerting stage of a child’s life when they go from hardly talking to suddenly uttering hundreds of new words, sometimes after hearing them only once. (This can be disconcerting for parents whose children are suddenly uttering profanities like a angry truck-driver.)

At 18 months the average child can say 50 words, but by age two, they have learned up to 350 words; half a year later their vocabulary has doubled to 600.

Scientists have proposed various theories to explain children’s language land-grab. Perhaps learning a few basic words helps a child learn others. The theory of “naming insight”, for instance, suggests that at around 18 months children suddenly realise that each object has a specific name. (It’s not until around two that a child learns that when they cover their eyes and ‘hide’, that you can still see them.)

Another theory, called “fast mapping”, suggests that children quickly understand that groups of objects are related, and therefore they learn unfamiliar words describing objects within familiar groups more quickly.

Characteristic curve

Of course, there may be a much simpler explanation. The acceleration in a child’s learning may inevitably happen due to the way most languages are structured.

All languages contain a characteristic distribution and pattern of words. Where most are of medium difficultly to learn, there are a few that are either very easy, or very difficult. Children always learn a number of words in parallel. These parameters have been factored into computational models which simulate how long it takes to learn 10,000 new words.

Guess what? The simulation the model produces the same characteristic acceleration in learning. Essentially learning one new word makes learning another new word even easier. This allows a child to move through words of medium difficulty more quickly since their learned in parallel. Acceleration is an unavoidable by-product of variation in difficulty. (It’s a network effect.)

Of course computer modeling isn’t the real world and may not accurately get to the heart of while kids learn language so amazingly fast, but adults don’t learn as quickly and don’t show a similar acceleration in their language learning.