How do children actually learn words? While the facts of children’s language development are fairly well-known, it is still far from clear just how they learn them.
First the size of the problem. Adults know enormous numbers of words. The size of people’s vocabularies has proved hard to estimate. The psychologist Paul Bloom estimates that 17-year-olds know 60,000 actual words. Looking backwards at what they have achieved, this means that, ignoring the first year of life, they have learnt 3,750 words a year or about ten a day. Three-year-olds are managing only about 10 words a week; older children must be absorbing new words at a considerably higher rate.
The graph below shows the average number of words learnt by a child each day of their life.
As adults we are still taking in new words all the time, not only through exposure to words we have never happened to encounter before, say the terminology of a new subject one gets interested in whether football Catenaccio (a defensive system) or architecture Palladian (a particular style), but also through the new words coming into the language such as i-phone and coronavirus.
The traditional common-sense belief is that a child learns a word by associating the word with a visible object. So the mother points to a dog and says Look! Doggie! The child associates the word with the object itself. Next time they see it, the child can say doggie, to their parents’ delight. In some sense this explanation must be true. Words in our minds are linked to visible objects and events in the world: I can point to a bird flying and say flying. But this doesn’t work with things that cannot be seen; I know that – how do you point to an action of knowing? How do you see the object known as air?
Philosophers have pointed out a logical problem with associating words with things by pointing to them and naming them: how do you know which aspect of the world is the thing the word is associated with? That is to say, how do you know exactly what it is that someone is pointing to? Most of the things in the world have many features. A tennis ball has a colour (yellow), a material (rubber), a shape (a sphere), a purpose (it’s for tennis), an age (new or worn-out), and other characteristics (bouncy, breaks windows, etc). If someone who spoke another language pointed to a tennis ball and said Shradditch, how would you know which of these attributes they actually meant? It could be the name for the ball, it could be the colour, it could be anything else at all the speaker wanted to say.
The child has therefore to work out which aspect of the world goes with the word. The world is not conveniently divided up into objects with single features. Naming means being able to sort out which aspect of a thing is being named.
Not even the fondest of parents actually spends much time labelling things – Look Jimmy a train! Rather they do more complicated things like describing the train – What a nice train! or warning – Watch out! The train’s coming! or complaining – The train’s late again. We don’t often say That’s a train because it would be such a boringly obvious thing to say.
A story among linguists concerns a researcher who was studying a previously unknown language spoken by a remote tribe. To find out the words of the language, he got people from the tribe to sit down on the other side of a table one at a time. He started by pointing at the table and saying Table. The first tribesman nodded vigorously and said something which the researcher dutifully wrote down. When he did the same with the next tribesman, the answer was a different word. And the third and fourth tribesman produced still different words. ‘Wonderful!’, thought the linguist, ‘a language with four different words for "table’’!’ Only much later did he discover that the first tribesman had said wood, the next you’re pointing, the third table, the fourth I don’t know, all showing their interpretations of what was going the situation, only one of which corresponded to his version. In other words it was far from obvious which word went with pointing at a table, except in the eye of the researcher.
The same with children. When the mother says Milk, she might be referring to the bottle the milk is in or to the act of giving the child a drink. If someone points to a red ball in a picture and says red, what can the child get out of it? Red might be a word for the ball itself, or a word for a round shape, or a word for a picture, and doubtless there are other interpretations. The child must have a way of working out which of the possible meanings the adult is using.
The philosopher Willard Quine put it in terms of hearing someone who speaks another language say gavagai as a rabbit scurries by. Gavagai might be a word for ‘rabbit’ in general, or it might be the pet-name of an individual rabbit. Or it could mean ‘one for the pot’. Or it could describe any aspect of the rabbit (furry, white, big etc). Or any part of the rabbit – nose, legs etc. The problem with learning how words go with things we see is sorting out which aspects of what we see are important, which are irrelevant.
Modern language teachers face the same problem of conveying what new words mean to the students. A fashionable teaching method a few years ago was audiovisualism. This meant showing students pictures and giving them language to associate with it: a picture of a car and the French word l’auto; a picture of a smiling face and the French Bonjour. Often this failed to work because the students did not interpret the pictures in the way the teacher wanted. They saw the car as a coupé or an old jalopy rather than just a car. I once observed a primary school class teacher teaching the word car to L2 children in London. Following the lesson the children had to count traffic on the street, so I pointed to a car and said to a boy ‘What’s that?’. He answered ‘That’s a Ford. My father drives a Jag’. He hardly needed to be taught the word car. A smiling face in a picture could mean almost anything; in many cultures smiling is a sign of embarrassment rather than friendliness. The audiovisual method works well enough for words whose meanings can be demonstrated visually but not for abstract words, structure words, and many others. It relies on everybody solving Quine’s problem in the same way by having the same interpretation of the picture.
So, when children meet new words, how do they know which of the umpteen different possibilities for its meaning is right? Children might play it safe by waiting for the next time they hear gavagai; if it’s said when there are lots of rabbits about in different colours, that eliminates the colour meaning. If the next time it is said the rabbit in question is obviously no longer edible, that eliminates ‘one for the pot’. And so on till the child gets the right meaning. But this puts an enormous burden on children’s memory and language learning processes. They’d have to store, say, ten memories of the word along with the details of each situation before they got its meaning. And this would be happening for 3,750 new words a year, i.e. 37,500 memories.
The psychologist Lawrence W. Barsalou claims that this is indeed more or less how we learn vocabulary: a word is remembered in combination with the situation in which we learnt it rather than its meaning getting detached; its meaning is distributed around different parts of our mind rather than being stored as a single item. When they think of the word sky, people’s eyes often actually move upwards. Think of the word run; the areas that are activated in your brain are not just those for vocabulary but also for movement. In our minds words go along with situations. Activating a word brings up everything else in our mind that goes with it. Hence athletes with an injury can gain from practicing mentally what they cannot do physically.
The psychologist Michael Tomasello showed that to learn a word you did not even need to see the object it refers to. An adult looks for a gazzer in four boxes. Three of them reveal ordinary toys for which the child knows the words. The fourth is locked and the adult pretends to be frustrated by not being able to find the gazzer, shaking it to hear something rattle inside. Later the child is shown the three toys together with another new toy, for which he or she does not have a name, and asked to choose the gazzer. By picking it out correctly, children show that they have learnt a word for something they had never seen by working it out from the adult’s behaviour. It is not just associating words with things that is important; it is knowing what naming an object means.
Children's Early Words Children's Early Word Combinations Words in the UK National Curriculum BabyTalk words