Syntax and Morphology

Morphemes

Language is made up of unit blocks (morphemes) that are combined into words. They include the word-content morphemes, and the inflectional morphemes that provide grammaticality.

English is quite morpheme-poor, but morphological acquisition plays a big part in languages with richer morphology.

Development of Affixes

Initially production is devoid of any morphological structure, but gradually inflectional and derivational morphemes start to occur.

The development of affixes seems to be cross-linguistically consistent (in the stages, not the specifics).

Stage 1

Children start memorising forms on case by case (hence they end up with ran and went).

Stage 2

They go on to discover that rules exist around 2;6. At this point they overgeneralise, creating words such as wented and runned.

Stage 3

They master exceptions to general rules.

Brown's Orderly Sequence

Roger Brown developed a study for determining the order of acquisition of bound morphemes and functional categories. It's been duplicated thousands of times.

Screen%20Shot%202012-10-01%20at%2010.01.29%20AM.png

It appears to be unrelated to the frequency with which children hear the different forms;

Screen%20Shot%202012-10-01%20at%2010.02.41%20AM.png

Explaining Factors

Several factors cross-linguistically explain this. The table below lists a number of properties, and those items that are learnt earlier feature more of them.

Screen%20Shot%202012-10-01%20at%2010.14.15%20AM.png

Tests to Study Development

Jean Berko Gleason in the 1950s developed the Wug test; he provided children with nonsense words and asked them to form plurals and past tense versions.

Screen%20Shot%202012-10-01%20at%2010.18.57%20AM.png

The easiest forms to acquire show productivity and lack of exceptions, but almost all forms are acquired by age five.

Syntactic Development

Choosing the Holophrase

Children are more likely to say the most semantically informative word for their holophrase. E.g. "juice" not "want".

They express semantic relations in their utterances; expressing the most important concept.

Screen%20Shot%202012-10-01%20at%2010.25.08%20AM.png

Telegraphic Syntax

We don't have enough evidence to know if children have actually acquired syntactic categories. They seem to use appropriate word order, but the utterances are so short that we can't claim for them having or not having them.

We can, once again, examine the semantic relations they're expressing. We're not sure whether they develop semantic relations or syntactic categories first.

Screen%20Shot%202012-10-01%20at%2010.28.01%20AM.png

Future Word Order

Children seem to show very few word order errors, even in languages with more variable word order (using the different patterns with roughly the same relative frequency as adults).

Phrase Structure Rules

After about 2 years language development is put at about one word per 2 hours (Steven Pinker).

It appears that acquisition of XP categories happens in steps; lexical items followed by functional ones.

Children later develop inversion and then wh-questions.

Recasts

We know that children do not learn in a behaviourist manner.

Whilst they don't learn from corrections, they do get given alternative utterances through recasts (e.g. "Dad, did you see the dog runned away?" replied to with "Yeah, she ran really fast, didn’t she?").

However mothers of two year olds only recast ~26% of the time, and they don't always fix them (sometimes just repeat them).

Saxton

Mathew Saxton, in 1997, found a scenario in which recasts aren't of no use; by presenting 4/5 y/o children with progressive forms of irregular nonsense verbs and asking them to provide the past tense he could measure the effect of two strategies.

E.g. "pelling" -> "pold"

  1. Recast the correct past tense to them
  2. Tell them the correct past tense initially.

With strategy 1 there was a 30% success rate after one recast, whereas with strategy 2 no child used the correct form.

After five weeks of biweekly sessions, children in strategy 1 had almost 100% success, whereas strategy 2 had about 40%.

Sensitivity to Recasts

Recasts appear to be more useful at the point when children have learnt the form but do not always remember it. For instance Roger Brown's participant Eve was sensitive to recasts when she was already correct 50% of the time.

Impact of Recasts

Children who hear recasts are simply not faster learners or better speakers, hence it cannot be necessary, or even significant.

We can say that it may be helpful to hear recasts, but that is about it.

This is similar to motherese.

Outperforming their Competence

Children use all kinds of shortcuts to perform; for instance they pick up on a word in a sentence and apply an action to it, regardless of the rest of the words. (e.g. getting a diaper everytime someone says the word).

To study these we have to eliminate contexts where world knowledge is sufficient. E.g. they also use word order strategies, which is great except for passive phrases.

Comprehension v Production

They also sometimes comprehend more than they produce; e.g. holophrastic stage children performing well in preferential looking tests for unusual telegraphic combinations.

Children prefer to hear, and respond better, to grammatical sentences. They prefer "throw me the ball" rather than "throw ball", even if they can only produce the latter.

Do Children Have Grammar?

The whole argument here is centered on whether children know verbs are an abstract category.

Arguments for Child Abstract Grammars

  • Children have word order rules, even in telegraphic combinations.
  • Children's language has hierarchal structure and abstract principles
    • Some words go together more than others
    • Sentences aren't just one word after another, they're in specific patterns and orders.

Arguments for Child Limited Syntactic Understanding

  • A theory is that children use semantic, not syntactic categories.
  • Children may use lexically based (case-by-case) rules
  • Usage-based grammar (claims that linguistic knowledge derives from experience; put chunks of language they know together and then notice patterns)

Genetic Langauge Inheritance

If we have language genetically, we must have genetic language disorders. This is, actually, the case.

Twins

Identical twins are more linguistically similar than fraternal twins or siblings; this cannot be environmental, given same household.

Adopted Children

Adopted children with biological relatives who suffer from language disorders are 3 times as likely to suffer from one as those without such relatives.

Animals

Children acquire language but kittens don't. It could be biological (but teaching chimps to sign doesn't work 100%), or it could be that we have an innate knowledge.

Principles and Paramaters (Universal Grammar)

Because general learning mechanisms appear too weak to accomplish language acquisition, we assume we must have something inbuilt to build off. This is also linked to Poverty of the Stimulus, where children never hear the full story, and they also hear meny errors in speech.

The theory is that we have principles and paramaters in our head, and when we discover our target language we set those switches to the correct location.

Acquisition Device not just for Language

Perhaps these parameters and this amazing way of acquiring knowledge isn't just for language; perhaps it helps us learn reasoning, logic, generalisation and statistical learning.

Communication involves many different considerations, which gives credence to this. Children have to learn to understand intent and patterns in order to learn to communicate, and these are general cognitive abilities.