The knowledge of language can be divided in various ways. One way is to separate out the various components of the knowledge, as this course is divided, into knowledge of sounds, words and sentences. We can then explore what children learn about language in each of those areas, and we will do that below.
We can also relate the acquisition of language to other areas of child development and learning, and compare how language is acquired to how other skills are acquired.
The general finding is that children do NOT acquire language through imitation, reinforcement, analogy or motherese.
| Theory | Empirical Problem |
| Imitation | Children create novel forms, e.g. "holded", which they have never heard. |
| Reinforcement | Parents correct for truth, not grammar. |
| Analogy | Which analogies work? Which don't? Too vague. |
| Motherese | Other cultures don't have motherese. |
Rather, they construct grammars of particular languages by making choices from a set of options available to them. This set of options is known as Universal Grammar. That is, they choose the fundamental elements and the rules of combination for their language. We can characterize what they learn about sounds, words and sentences by looking at what they know abou the elements and rules of combination in each area.
It is difficult to test children when they are first born. We can measure their interest in speech sounds by measuring their sucking rate. By doing this we have discovered that they come pre-equipped to head phonetic contrasts even for languages not spoken around them. We now know that tuning in to their ambient language begins at about 10 to 12 months, at which time they begin to fail to distinguish contrasts not found in their ambient language. That is, they seem to be discovering phonemes at this point.
Even deaf children babble. They seem to be testing out their vocal apparatus, and perhaps tuning the articulation to their own audition. Babbling initially follows statistics closely, with the most common cross-linguistic sounds and patterns babbled the most But later on they babble less common sounds, even sounds not present in their ambient language environment.
At this stage utterances are one word long, though they often contain complex messages. Children have been shown to understand differences in word order at this stage even though they can't produce sentences with different word orders.
Most children go through a stage where sentences seem to be limited to two words. There can be a large variety of two-word patterns, however, and the sentences continue to encode much more complex meanings.
After the two-word stage there is an explosion in the child's capacity to form sentences, and developmental patterns are more difficult to describe. At this point it is better to simply test particular aspects and constructions from adult grammars to determine which aspects the children have acquired.
Animals communicate. But they don't seem to have communication systems like human language. Their communication systems lack recursion, and thus cannot create an infinite number of expressions from a set of elements and rules of combination.
Being human. The language faculty seems to be unique to humans. So language acquisition is easy for us: we're pre-programmed to do it.
Children often over-generalize a rule, applying it to cases where it is not used in adult grammars, e.g. "hold-ed" instead of "held".
This leads to a "U-shaped curve" if we plot percent correct against time. Initially, they memorize all cases, they they learn the rule and over-use it (getting words like "holded" "wrong"), then they re-learn the exceptions to the rules.
As discussed above, the first year shows vast changes in the sound system. Children begin with the ability to distinguish between many different speech sounds, but by about 10 months begin to tune in to the ambient language and exclude non-contrastive distinctions.
They tend to start with combinations of sounds that are more common in the world's languages. Thus, they start with small, CV syllables. This can many words homophonous, e.g. "do" and "drew" both as [du]. Another common process is "stopping", pronouncing stops for fricatives, e.g. [top] for "soap". Unstressed vowels are also often deleted, as in [næna] for "banana". In general, they perception is better than their production, and they will often object to adult mimicry of their speech.
Plurals are one morphological process that is learned relatively early. The "wug" test demonstrated children's use of word formation rules.
In terms of categories of words, nouns are generally learned before verbs, and content words (nouns, verbs, adjectives, ...) are generally learned before function words (auxilliaries, prepositions, ...).
They often over- or under-generalize the meanings of words, for example using "dog" to refer to all animals, or using "dog" for only their own dog.
Even though children may leave out many words, the words that are present generally follow the correct word order from abbreviated phrase-structure rules.
They tend to use intonational patterns to ask questions rather than moving the words around.