The mystery of language: how children learn to speak their mother tongue

‘I want.’

‘This go on here.’

‘I’m eating my bread Mummy.’

‘I’m going to Cookie Monster’s place, and I’m going to havea peanut-butter sandwich.’

You’ve just read sentences produced by children at 1 year and 8 months, 2 years and 3 months, 3 years and 4 years respectively. The journey from single syllables to full sentences is remarkably rapid, and has puzzled scientists for decades.

Read more about children's development:

• Children can understand counting “years earlier than previously believed”
• Music helps to build the brains of premature babies

I would advise you to enjoy the simplicity of that baby talk while you can, because it’s in attempts to get to grips with this puzzle that the study of language becomes, well, difficult. This is where the really big questions are tackled.

If you’ve ever tried to become fluent in a second language you’ll know how laborious it is, how you seem to move forward at a snail’s pace if you’re lucky, and only then because of great effort.

Now, can you remember getting to grips with your first language? Do you recall struggling to figure out where the preposition should go, or memorising irregular past tenses and when to say ‘the’ instead of ‘a’?

No. Because it came ‘naturally’ to you. You were raised on it, not ‘taught’ – you didn’t have to have lessons. In fact, you could speak before you went to school, merely as a result of being in the company of the people who raised you.

There’s now evidence that the whole process starts even earlier than that – in the womb. According to German researchers, the melody of newborn babies’ cries is shaped by the sounds of their native language, which they must have heard through their mother’s bellies.

The ‘development’ or ‘acquisition’ of speech in children – so called because it can seem more like an organic rather than a willed process – is all the more remarkable given just how complicated language is. Think of the knowledge stuffed into that sentence: ‘I’m going to Cookie Monster’splace, and I’m going to have a peanut-butter sandwich.’

Listen to theScience Focus Podcast to find out more about the science of language:

First of all, you have to know the meanings of the verbs and nouns – ‘to go’, ‘Cookie Monster’, ‘place’, ‘have’, ‘peanut-butter sandwich’. Very roughly, children can produce around 10 words by the age of 13 months, 50 by the age of 17 months and 300 by the age of two.

By the time they’re 6, they are able to say around 3,000 words, and understand a third as much again. As soon as children can put syllables together, between the ages of 9 and 18 months, words become an important currency, a means of getting what they want: ‘milk’ or ‘teddy’; and of expressing their preferences: ‘no’, ‘more’, ‘again’.

New words are absorbed at the rate of around 1 to 3 a week up to the age of 18 months, when the pace starts to pick up. How are these words understood? Sometimes, the method developed by St Augustine will suffice: the parent presents an object to the child while naming it.

More often it will be through observing how they are used – the Wittgensteinian way (after all, with words like ‘no’ there is nothing to point at). Learning words is a feat in itself, but really that’s the easy part.

What about grammar? If words were all you needed, it would do just as well to say ‘Sandwich going I’m to Cookie peanut Monster’s butter and place I’m to going have a’. This doesn’t happen. Remember, language – the ‘code’ part at least – consists of meanings (semantics) and the rules forcombining them (syntax). The latter is essential for avoiding gobbledegook.

Another facet of grammar is knowing how parts of words are put together – morphology, which has its own sets of rules; for example, if you want to change the tense of the verb ‘walk’, you have to add different elements to it, in a specific way: ‘walking’, ‘walked’.

Read more about linguistics:

• What artificial languages can tell us about ourselves
• Telling stories: why your elderly relatives are the best storytellers

In the Cookie Monster sentence, the child displays extremely sophisticated knowledge of syntax and morphology.

She knows how to conjugate the verb ‘to be’ and how to abbreviate it (‘I’m’). She knows the conditions governing the use of present continuous tense in English. She uses a special marker (’s) to indicate possession, places the subject, verb and object in the right order and unites two clausesusing a conjunction.

Most adults would be unable to set out the grammar of even a simple sentence like this. But somehow, children know all about the relationships between the different words. They must do, because they’re able to choreograph them with apparent ease.

This choreography is all the more impressive when it comes to questions. ‘I’m going to Cookie Monster’s place’ is a declarative sentence. The equivalent question, the interrogative form, would be ‘Where am I going to?’ Simple, right?

Except that, to get there, you have to replace the destinationwith a specialised counterpart of ‘here’ or ‘there’, which is only used when the location is not known. Then you move it all the way to the front. You swap the auxiliary and the subject around, expanding the former in the process, and leave the main verb and the preposition in place.

The question of how human beings – children or adults – perform these kinds of complex operations without having been taught them, or even having any conscious knowledge of what they’re doing, has proved difficult to answer.

It is made trickier still if – as many believe – no child is given enough ‘input’ from the outside world to generate the ‘output’ they eventually become capable of.

Were language acquisition solely a question of learning by rote, it would in principle be impossible: one of the key distinguishing features of any given human language is that the number of expressions it contains is infinite. You cannot learn all the ways of putting words together by memorising them if they go on for ever.

An alternative is to try to learn rules: to identify patterns in the way adults speak, and apply them to new situations. This would explain why children often overgeneralise the rules they learn.

Take the English past tense, which is formed with ‘-ed’. A child will hear plenty of examples of it, will begin to understand that it refers to things that have already happened, and will start to use it.

The problem is, there are lots of irregularities, which do have to be learned by rote. These include past forms like ‘went’, ‘hit’ and ‘caught’. And because they’re applying rules, but haven’t memorised all the exceptions yet, the child will produce words like ‘goed’, ‘hitted’ and ‘catched’.

That’s not the end of the world. Irregular past tenses are fairly common, and as time passes, children begin to incorporate them. Sometimes, though, the process of correction isn’t as straightforward.

There are certain constructions that don’t seem to get ‘disconfirmed’ like ‘hitted’ and ‘catched’do – but which children still somehow understand are mistakes.

Take the following pairs of sentences:

Dad told a story to Sue

Dad told Sue a story

Jim showed the model to Bob

Jim showed Bob the model

Mum baked a cake for Jack

Mum baked Jack a cake

A child hearing these sentences would be able to identify a clear pattern. In a sentence where the object is somehow the beneficiary of an action, there are two ways you can arrange things: one where the recipient comes at the end, after a preposition, and one where the recipient comes straight after the verb, with the ‘gift’ coming afterwards.

That means she can then go away and use the rule for herself. She might say ‘Dad gave me a sandwich’ instead of ‘Dad gave a sandwichto me’. Or ‘Mum showed me a drawing’ instead of ‘Mum showed a drawing to me’.

But she might also reason that, because you can say ‘Dad said something nice to me’, then ‘Dad said me something nice’ is also fine. Or that since ‘Mum buttoned up the coat for Jack’ is OK, then ‘Mum buttoned up Jack the coat’ is as well.

Unfortunately, to most English ears these sentences sound very wrong. But does the child get evidence of this in the course of her everyday life? Are their assumptions about these kinds of sentences ever disconfirmed?

Hearing an irregular past tense is one thing, but if you simply never heard ‘Mum buttoned up Jack the coat’, at what point would you understand it was wrong? And yet, somehow, children do – mistakes like these are eventually filtered out.