Language Acquisition - created from Mind Map

Página 1

Questions in Acquisition:

Lightfoot (1982) writes about the central problem: how do children master their native languages?

Lightfoot argues that the made problem is that there is a 'defiency of the stimulus' meaning that children cannot possibly learn language entirely from the input they get as they do not experience everything they show they can both understand and produce.

Main three problems include:

- The language they hear is not always grammatical standard language.

- They cannot hear everything that can be produced because language is infinite.

- Children find a way of understanding what is and what isn't their native language, often without being told.

HOW DO CHILDREN ACQUIRE LANGUAGE:

- It's universal as long as there's exposure.

- It's rapid; it takes 3-4 years to go from first words to fluency.

- Children make relatively few errors during acquisition considering the number of logical mistakes they could make and the complexity of linguistic systems being acquired. They acquired language easily compared to adults learning a L2 language.

- Acquisition is fairly uniform despite variable input. Not all cultures have child-direct speech/speak to babies and infants. Children also hear a lot of mistakes.

COMMON MISTAKES MADE:

- Overgeneralisation

- Sound Substitution

MISTAKES NOT MADE:

Binding Principle--> "Names/Noun Phrase's can't co-refer as an object prior in the sentence"

Children never make this mistake. Comprehension tests show that they understand that 'He was in the box when the smurf ate the handburger' is ungrammatical if the smurf and 'he' are the same person.

Wanna Contraction --> Children won't use 'wanna' to mean 'they want to do something'

THIS IS THE RABBIT I WANNA VANISH (Must mean that they want the rabbit to vanish, not that they want to vanish the rabbit).

STAGES OF SPEECH

Babbling stage -> Holophrastic Stage -> Telegraphic Stage

TELEGRAPHIC SPEECH: Words chosen contribute to lexical meaning. Grammatical function words are usually missing.

NULL SUBJECT STAGE:

- Salient property of telegraphic speech.

**ALL children drop subjects - NOT all children need to learn to use them again like those in English - EG. Italian and Japanese only use subjects to emphasize**

Subject isn't necessary when it is directly referenced, often by context.

"THE MOTHERESE HYPOTHESIS"

Caretaker speech is structured in such a way that it facilitates acquisition, however it cannot be used to say WHY children learn language as not all children experience motherese, in fact not all culture even speak to children/infants who cannot yet speak themselves.

Properties of Child Directed Speech (C.D.S):

* Clear Articulation

* Higher Pitch

* Exaggerated Intonation

* Shorter Sentences

* Fewer Errors

* Repetitive

BUT:

- Not syntactically simpler

- No correlation between simplicity and grammatical development ALTHOUGH there are some aspects of caregiver speech style that affect acquisition.

SUCH AS:

- Caregivers asking more Yes/No questions see's quicker auxillary growth in the child.

- Use of diexis by the caregiver enhances growth rate for noun inflections WHEREAS high levels of commands used by the caregiver reduces inflections.

Infants do prefer CDS over adult direct speech; probably because the exaggerated intonation holds their attention.

Prosodic qualities may provide input on stress patterns and other syntactic properties: BOOTSTRAPPING.

UNIVERSAL GRAMMAR

**An Innate Template For Language Learning**

We are born with a set of tools which we use with input to come up with a set of rules for language, this has to be reassessed with new input.

Children filter everything through their current grammar, which is why negative evidence/correction doesn't work.

What would be UG?

- Cross-Linguistic

- Should share deep grammatical commonalities across languages.

- Should be linguistic

*Example: Are we born with the ability to discriminate phonological pairs? Is this linguistic since Chinchilla's can do it? It can't be. The ability to hear, and process cognitively, the different phonemes is innate because other animals, like chinchilla's can do it, so it can't be linguistic.

However: The ability to categorise phoneme with predictable allophonic variants is UG, however initially the distinctions are not based on this understanding of different categories but on the input and this bit ISN'T UG.

Stress/Rhythm Class - NOT UG. Not all languages have stress and those that do have different stress patterns. Therefore this must be learnt via input rather than being UG.

BLOCKING - Pinker says is part of UG. All children go through it in a fairly uniform way... would need to prove it was cross-linguistic. However, it works alongside INPUT as input determines overregularisation. The more frequent a word in the input the less it is overgeneralised.

Speech Perception:

Problems in Speech Perception:

- We are all biologically different with different size heads and vocal tracts which makes the sounds we all make different.

- EINMAS calls this a 'limited linguistic environment'.

An example of this is COARTICULATION: Individual sounds cross over and overlap the boundaries of phonemes in speech. For example vowels can become nasalised. SO HOW DO CHILDREN DECODE THE INDIVIDUAL PHONEMES?

PRE-BIRTH:

* Neo-Natal babies can hear rhythm and intonation and debatably vowel-formant information.

DAYS OLD:

* Recognise mother's voice at 3 days old.

* Show preference for parental language at 4 days old.

Between 1-6 MONTHS:

*EINMAS found: 'Underpinnings of later speech production' using high amplitude sucking procedure.

* Phonemic discrimination between minimal pairs crossing V.O.T boundaries, with the exception of /b/ and /p/.

* Suggestion that this ability is innate and part of UG -> Discriminatory abilities found cross-linguistically even in languages that don't natively make the distinctions.

-> Children are born with the ability to discriminate, then adapt it to their own native environment from the input they receive. (So English babies only contrast Voiceless Aspirated & Voiced Stops).

HOWEVER this may not be UG as it's not necessarily a linguistic ability; Chinchilla's can do it.

FEATURE DETECTOR HYPOTHESIS:

It's a biological ability, later modified by the input of the native language.

5-6 MONTHS:

- Using head turn preference procedure, children can show ability to consitently discriminate between language categories in their native language. Reflects experience with particular language, and is NOT reliant on phonological acquisition or lexical comprehention.

7-12 MONTHS:

- Infants loose the ability to discriminate between non-native phonemes.

- Babbling starts to take on the characteristics of the native language.

EXAMPLE: Walter and Tees (1984) --> At 6-8 months American Infants could correctly distinguish Hindi & Thompson Salish consonants 85% of the time, by 1 year this drops to 25%.

1 YEAR:

* First Words Appear

* Complete loss of non-native consonant production/comprehension abilities.

KUHL ET AL: (2001) Found that Japanese infants found /r/ and /l/ distinctions much harder.

PRODUCTION ABILITIES:

1-4 MONTHS:

* Sucking, Burping and Crying.

* AT 2-4 MONTHS --> Cooing/Laughter produced socially.

- Have a shorter vocal tract, more mammal like than human.

7-12 MONTHS:

- 7 Months: Reduplicated, Canonical Babbling (MaMaMa)

- 10 months: Varigated Babbling. (MaBaBo)

*It is easy to tell whether the baby is babbling in a foreign or native language.

10-12 MONTHS:

- First Words

- Symbolic Use of Language (A sequence of sounds takes on a fixed meaning, although this isn't necessarily the adult word. Bo for bottle etc).

Morphological Development

BROWN (1973)

* Proposes the 14 grammatical morphemes and stages of their development.

--> Browns stages related to the MLU of a child.

Stage One - 1.5-2.0

Stage Two - 2.0-2.5

Stage Three - 2.5-3.0

Stage Four - 3.0-3.5

Stage Five - 3.5-4.0

Characteristics of stage one:

- Mainly using nouns, verbs and adjectives.

- Predicative constructions.

- Children filter out the elements their grammar doesn't recognise.

Stage two:

- Children start to acquire the 14 grammatical morphemes.

* Order of Acquisition*

1) Present Progressive '-ing'

2) Plural '-s'

3) Possessive '-s'

4) Past Tense '-ed'

5) 3rd Person Singular '-s'

THE WUG TEST

--> Issue: the test was carried out on first grade children aged 6/7 years who are expected to know almost all grammatical rules.

*Berko-Gleeson used novel words and picture cards to elict various inflections.

*Children did well with invarient forms.

*Struggled with plural '-s' in words ending in silibant sounds and '-ed' in words ended in stops.

MLU:

Proposed as a method for working out a childs grammatical development.

Useless cross-linguistically for of variations in morphology. A verb/utterance method better in cross-linguistic cases.

PINKER:

On regular and irregular morphology and overgeneralisation.

Usually happens during stages 2-4 of Browns stages. Marcus et al, found that children usually use correct and incorrect stage at same time.

PINKER proposes BLOCKING as a method for infants learning irregular morphology.

*Children can't learn via negative evidence as corrections to a child's language have shown no obvious improvements and often adults do not correct a child's speech.

*Children often use the correct irregular term first, then start to overgeneralise when they learn a grammatical rule.

*Suggests this is because of the PERFORMANCE EFFECT --> children have had less exposure to language and their ability to retrieve information cognitively is less reliable. Therefore after grammatical rule is learnt sometimes children simply apply the rule because they cannot retrieve the irregular form.

--> This is supported by the fact that children make more overgeneralisations with less frequent words... and adult often make them with rare forms like dreamed/dreamt.

Early Syntax

Phonological Development

Phonological Problem:

- How do children segment continuous speech? We don't give them words in isolation so how do they come to understand word boundaries?

**C.D.S only exaggerates pauses that are already there in the natural flow of speech**

BOOTSTRAPPING

Information avalible in one domain that gives the child information about another.

E.G --> Prosodic Bootstrapping: Using prosodic features as cues to word boundaries, cues to language discrimination, and cues to clausal or phrase boundaries and word order.

WORD BOUNDARIES:

Listener = 3:1 chance of a strong syllable being the start of a new lexical word.

METRICAL SEGMENTATION STRATEGY:

Cutler & Norris 1988

Listeners identify new words with the occurence of a strong syllable in English.

Young children are highly sensitive to stress-baed rhythm in language.

6-8 MONTHS:

Babbling conforms to the rhythmic properties of language.

9 MONTHS:

Juscyzk shown using the head turn preference test that infants prefer trochaic stress over final stress.

STATISTICAL CLUES:

How do infants know words are trochaic?

STATISTICAL LEARNING HYPOTHESIS:

--> Infants use transitional probability to work out initial boundaries are use this to figure out trochaic stress pattern.

Thiessan & Saffron 2003

FOUND --> 7 Months: Use statistical domain general method of figuring out word bounaries.

---> 9 months use the language specific method of stress pattern.

FIRST WORD PHONOLOGY:

1-2 YEARS: Classification of sounds into phonological categories.

INGRAM 1976:

Common processes incl.

**SUBSTITUTION (Example: Fricatives to Stops)

**ASSIMILATION (Example: Paper becomes [be:b] ie. Voicing before vowels.

SYLLABLE STRUCTURE REDUCTION:

**Children won't produce clusters in early speech**

Coda Omission

Reduplication

Weak Syllable Omission

Cluster Reduction

Children prefer to use least marked options (CV syllable structure) because they are cognitively easier:

- Obligatory Onset

- No Coda

- No Complex Coda's or Onsets

MARKEDNESS THEORY (Jackobsen)

Children also acquire less marked structures before marked ones.

SONORITY HIREARCHY:

Most sonorant towards nucleus, least further away.

**Children remove elements of clusters in a way that keeps the largest contrast**

Vowels = Most SONORANT

Nasals/Liquids

Fricatives (Excepting /s/ and /z/)

Stops = LEAST SONORANT

Children resist [st] because it violates the sonority hirearchy. SMITH 1973 (Straw becomes sraw, traw, but NEVER staw)

COVERT CONTRASTS:

Sometimes children may not be able to produce a contrast that they can comprehend. ROY & MASILSON (1995) found that spectographs can actually show that the children ARE making a contrast, it is just adults with fully formed phonemic categories can't perceive it.

Known as COMPREHENSION-PRODUCTION MISMATCH.