Previously, it had been thought that language was a conditioned response, based on imitation/reward. But this ideas is now discredited:

• Can't be purely imitative, because children say things that are wrong and that they wouldn't have heard someone else say, e.g. "mummy get it my ladder". [Fromkin]
• "reinforcement seldom occurs, and when it does, it is usually incorrect pronunciation or incorrect reporting of facts that is corrected" [Fromkin]

Single-word utterances

During this phase, which occurs between 10-20 months, the vocabulary builds up slowly, one word at a time, and includes different types of words (things, actions, etc.) but has no grammatical constructs. At this stage, words may be holophrasitic - that is, the utterance conveys more meaning than just the word itself - e.g. "up" means "I want you to pick me up".

While it's not possible to classify the words as "nouns", "verbs" etc., because they're not combined in any grammatical sense, there does seem to be a bias towards "object-referring" words. The words do not include any grammatical glue such as "the", "to", "is", etc..

Two-to-three word utterances

Towards the end of this stage, children start trying to form more complex proto-sentences. "Vertical constructions" occur when there is a relationship between two separate utterances that, if combined, form a more sophisticated sentence. There may also be a kind of rehearsal, with the child saying "me want" then "want that", before "me want that". It appears that the child is forming the idea of grouping separate words into chunks, which can then be used in a hierarchical grammatical construction.

Evidence for this is found in the way that children apply negation, e.g. an utterance such as "no sit there" requires understanding of "sit there" as a distinct unit of meaning.

Up to this point, the features of grammar that are seen are:

• universal rather than language-specific
• context-dependent - e.g. may be accompanied by gesture
• content-laden
• object-oriented

Trask says that appears that children are trying to formulate rules and try various different constructs until they hit on the right one. E.g.

• no I want juice
• I no want juice
• I don't want juice

There is very strong evidence that language ability is innate:

• deaf infants "babble" with their hands, and progress through same stages as hearing children as they learn the language. "If the deaf child's parents are not completely fluent in sign language, but use it somewhat inconsistently and inaccurately, the child still learns the language perfectly" [Trask]
• children of speakers of pidgin create a new language based on pidgin that has all the grammatical rules of a proper language (a creole) - more sophisticated than the one spoken by their parents
• Nicaraguan Sign Language invented by deaf children in an orphanage who weren't exposed to any other language - by the second generation, NSL has large vocabulary and sophisticated grammar

There does seem that there may be a cut off age (critical period) after which language acquisition ability is turned off. (Genie, older NSL children were not able to learn language so well)

Grammatical Systems

Book readings:

• Crystal, Section 41
• Trask: "Language: The Basics" ch. 8
• Fromkin: "An Introduction to Language" ch. 8