Language Acquisition

Language Acquisition

When we talk about language acquisition, technically, we are referring to the acquisition of the mental system that allows people to speak and understand a particular language.

Why we look at the above as the above:

  1. Mature language users/speakers have acquired a system of productive rules (they can generate endless combinations of utterances)
  2. Children apply rules productively and generally to words they have never heard (and do so without guidance - often coming up with incorrect but understandable words)
    • e.g. bringed brought broughted brang - they say all

Investigation

To investigate these we use a mix of the following:

  • Study naturalistic observation
  • Experimentation

Naturalistic Observation

Naturalistic Observation is longitudinal (spaced out over a period of time), and takes context into account. The observation sessions are regular, and we precisely note age and acquisition order.

We look at:

  • diary studies (counting the words they say)
  • recordings
  • videoing
  • notes

Experimentation

Experimentation is cross-sectional (compares linguistic knowledge across different children and different points in their development (ages)).

Specially designed tasks are used to elicit linguistic activity relevant to the phenomenon (e.g. play sesame st or show them toys). The child's performance is then used to formulate and test hypotheses about the state of their grammatical system at the time.

We test children's:

  • Comprehension
    • We can use picture selection format - a format in which children perform actions that demonstrate their understanding
    • Steven Crane does tests involving something/everything/nothing/anything/ etc
  • Production
    • A child may be asked to describe what's happening in a picture/cartoon
    • Production lacks behind comprehension - you understand more than you can produce.
    • These tests tell us what kids can do, but not what they can't do
  • Imitation
    • Production and comprehension rules are often used in conjunction with these
    • Children won't imitate correctly if they can't understand it - it's dependent on the current state of their grammatical knowledge.
    • e.g. "can" and question swapping - "what can you eat", "what I can eat" etc

Language Acquisition Paradigms

There is evidence for each of these being the most effect paradigm and system of study, but there are many debates about them.

  • behaviourist
  • nativist
  • cognitivist
  • social interactionist

Phonological Development

  • babies are exposed to a number of different sounds in their environment (even before birth)
  • they can distinguish speech from non-speech from birth
  • they respond differently to human voices vs other sounds3, and in a few weeks can distinguish to their mother's voice

High Amplitude Sucking

To test phonological development, a child is given a dummy where every time it sucks it a noise/sound is generated. Upon first hearing a sound they suck frequently, (attributed to interest), and when the sucking decreases the sound is changed to a new one. If the sucking resumes frequency we can presume their ability to differentiate between the two.

To determine whether children are built with an ability to tell the difference, or just learn quickly, infants are exposed to two fairly similar sounds that are not a part of their native language (e.g. glottal stops for Western children, or [th] for Asian) and tested to see if they can determine the difference.

Infants under about 6 months are able to discriminate between *all* sounds, whereas anyone (adults etc) over that age can only distinguish between sounds in their own language.

(Interestingly it seems that the language experience/decline in ability affects mainly foreign sounds that are phonetically similar to sounds in our native tongue; 14 month children through to adults can differentiate between clicks in Zulu).

Speech Sounds vs Meaningful Words

Despite their ability to distinguish between sounds like b and p, children under 18 months can't tell the difference between the two toys pok and bok. From this we learn that although they can recognise the contrast, they have not yet learnt that this is linguistically significant.

Babbling

Children begin to develop the articulatory movement needed to produce distinctions for the phonemic contrasts of their own language via babbling. Babbling starts around 6 months, and their are significant similarities regardless of the language exposure

Screen%20Shot%202012-09-08%20at%2012.50.44%20PM.png

Babbling increases in frequency until about 12 months, and then has a transitionary period out whilst the child learns to talk - it stops after 50 words.

Deaf Children

Very early on, deaf children babble the same as hearing children - however (probably due to not being able to hear themselves), this dies off. Deaf children of deaf parents (i.e. those exposed to signing natively) babble manually.

Acquisition Order

There are general trends in the way speech sounds are acquired in production and perception:

  • Vowels are acquired before consonants - easier to say, don't obstruct flow of air, etc
  • Stops are acquired before other consonants - t, d, g, k, p, b
  • Place of articulation matters - we acquire labials then velars then alveolars then palato-alveolars then interdentals
  • Contrasts are picked up earlier when they are in word-initial positions

Screen%20Shot%202012-09-08%20at%2012.56.04%20PM.png

Screen%20Shot%202012-09-08%20at%2012.56.07%20PM.png

Language Production

Children perceive what adults do, but can't necessarily produce it (e.g. a child who says "this is my fis" will disagree when an adult says "your fis?" but agree with "your fish?").

All children reduce syllables and substitute sounds in order to make production easier.

Syllable Simplification

Children delete the final consonant of a word (it's much rarer to delete initial), or remove segments, in order to reduce syllables down.

e.g. dog - do.

Reduction of Consonant Clusters

Children remove sounds in order to reduce a structure to CV (consonant vowel - the most commonly found pattern universally).

Screen%20Shot%202012-09-08%20at%2012.59.12%20PM.png

Substitution

Children replace sounds with easier alternatives.
E.g. Zebra - Debra

Assimilation

Children also merge sounds with those around them, adjusting them to flow together nicely.

Mental Lexicon

Everyone has an innate mental lexicon - a network structuring and linking words together with other similar sounding (phonetics) or meaning (semantics) words.

We do forget words as we get older (not using network pathways causes the signals to weaken, etc).

Defining a Word

The problem of defining what a word is gets tricky when you think of whines and expressions of frustration and also non-referential signals of greeting and social routines.

  1. A word is a symbol
  2. There is an arbitrary relationship between form and meaning
  3. A word refers to something

Phonological words

Some words are easily grouped into all being the same thing.
e.g.: thingamabob, whatchamacallit, watzit

Protowords vs First Words

Protowords are a child's way of referencing something without any relation to the language. E.g. 'noonoo' for blanket.

First words are an approximation of a word in the target language, e.g. 'bankie' for blanket.

Prelexical Use

Referentiality

It's hard to tell if early words are situation specific or contextually flexible - in situations
are early words purely context bound - are words prelexical or are children prelexical.

Categories of First Words

Nelson took 18 kids and categorised their first 50 words:

  • specific nominals (my teddy) - most at first
  • general nominals (food)
  • modifiers (nice)
  • action (jump) - most as they get older
  • pronouns (his)
  • personal social words (hello)
  • grammatical function words (in)

Hypotheses Explaining the Categories

Natural Partitions Hypothesis

kids vocabulary arises out of the structure of the world they experience
- nouns label things, verbs vary from language to language - noun meaning is more stable, verbs vary from language to language

Meaning Hypothesis

other hypothesis: children acquire words they can link to a meaning

Language Learning is not a Word Learning Problem Hypothesis

another theory:
cognitive/linguistic understanding has nothing to do with how some words are acquired earlier than others. language learning is not a word learning problem. Children use nouns because their knowledge of language allows them to guess at the meanings of those words.

Cultural Differences

children acquiring korean mandarin and japanese show less noun bias than english learning kids. lolwut why - cultures are not more verb than noun based or vice versa. some people believe it, some do not. language doesn't reflect culture.

Why the 50 Explosion

Children learn their first 50 words slowly until around 15-24 months of age, and then there's a rapid explosion in word acquisition. There are many factors thought to come into account, but they mainly focus on the fact that learning helps us learn - as we know more it's easy to pick things up faster.

  1. Maturation of internal word learning constraints
  2. Threshhold at 50 words helps in further acquisition
  3. Realisation that words refer to object categories, not just objects.
  4. Maturation of children’s phonological abilities.
  5. Ongoing cognitive development

Word Comprehension

Children can comprehend words far before they can produce them. Usually they respond to their own name first, and then to shorter phrases (e.g. give me a kiss), and then to individual words.

Children often say things without knowing the meaning, just understanding the context (e.g. "love you!" or "late.").

Acquisition Rate/Content

Some children start with referential words, other with context-bound, and children pick up words at different times (others pick up completely different vocabularies).

We don't know how much of the difference between word types and vocabularies is due to the the input they receive, and how much is due to different routes in their early development

Rate Factors

It could be dependent on the following (we don't know):

  • The mount of talk children hear?
  • Birth order? (More one-on-one attention)
  • Socio-economic status (They believe this highly in the US. Debra thinks it's rubbish).
  • Nature of speech babies hear - simplicity vs richness of vocab
  • Informativeness in CDS (Child Directed Speech) - expansion and explanation.
    • Expansion: (ooh chair, sit on chair, can you sit on the chair, make the dolly sit on the chair, etc chair words) Some people think it's good. Argument - there are cultures without CDS, they acquire language the same by 6ish. (Some cultures believe kids aren't human beings).
  • Speech in response to their own behaviour

Or on the following factors within the child:

  • Age when joint attention skills start (you get there's, or declarative/imperative pointing from them)
  • Phonological memory - more advanced vocabulary skills
  • Predict further vocabulary development (phonological memory can predict this)
  • Sex differences - girls have a slight early advantage

Process of Word Learning

There are different types of learners - Analytic vs holistic learners (some get details then build a big picture, some get big picture then find details). We don't know there are which, but some go top down and others bottom up.

There's also the problem of word segmentation: How do you find the word within the stream of speech? (e.g. some kids think that elemeno is a letter of the alphabet).

Improving the process

Doing the following helps children pick up words:

  • Recurring sound sequences
  • Stress or rhythm
  • Child directed speech

Word Referent Mapping/Gavagai

The Gavagai story describes the problem of specificity of language - if a foreign language speaker points at a rabbit and says "gavagai!" does he mean that gavagai means rabbit? How about:

  • Running rabbit?
  • Rabbit on a hill?
  • Brown rabbit?
  • Brown rabbit on a hill?
  • Rabbit leg?
  • Look at the rabbit!?
  • Get me that rabbit!?

etc etc - how do we know when to use a word.

Some people think this mapping problem is a logic/philosophical problem, whereas some people think it's a social problem of intent of meaning that the other person is trying to convey - they believe pragmatics convey everything.

Over/Underextension

Children struggle with this - they overextend (they generalise the word 'dog' to all four legged animals), or underextend (specifise it to just their pet).

Overextensions in comprehension are rare and not predictable from overextensions in production.

It’s possible then, that overextensions reflect lexical access processes, not word meaning.

Potential Ways of Mapping

Initial Fast Mapping

The Initial Fast Mapping Hypothesis says that: Every time a new word is encountered a new hypothesis is generated for it as to its meaning.

By the time kids are word learners, they may have non-linguistic understandings of objects and events in their world which they bring to the word learning task - for instance they assume things like:

  • Words must refer to whole objects
  • Different words are for referring to different things (saying cup and glass would contradict this).

Pragmatic Bases

When we say pragmatic principles we mean the child's understanding of how language is used.
Pragmatic bases of word learning:

  • Principle of conventionality (meaning of a word is determined by convention)
  • Principle of contrast (different words have different meanings)

No true synonymy in language - big sister v large sister etc

Salience Theory

The theory is that how salient something is determines when it will be acquired; children pay attention to the newness of an item in a physical environment.

This makes sense given how their input is based on conversations about the here and now, but environmental support is not accepted to be the whole story in determining word/meaning assigning and acquisition.

Syntax as a Clue to Semantics

We can often tell the parts of speech of something based on its context.

Sibbing

Roger Brown was the first serious child psycholinguist. He had a study called 'sibbing', where he would show children an image of a pair of hands kneading a mass of material in a container and tell them it was one of the variants on 'sib' - 'sibbing', 'a sib' or 'some sib' - before asking them to describe it.

If the children heard 'sibbing' they thought that was kneading, whereas 'a sib' meant the container, and 'some sib' the material.

Gorping

Naigle's experiment on gorping tells us about syntactic bootstrapping (the theory that when verbs are presented within their syntactic context and with an image or scene in front of them, can provide information about their meaning).

Two year olds were shown a picture that involved rabbit repeated pushing duck down into squatting position with its left hand, while both rabbit and duck were making circles in the air with their right hands.

The first group were told that “the rabbit is gorping the duck” and in the other “the rabbit and the duck are gorping”, and were later shown two videos on different tv screens, each with one of the actions (one with pushing down into squatting, the other with making circles), and asked “Where’s gorping now? What’s gorping?”.

Each group identified gorping as a different action.

Word Learning based on Child vs Environment

There are three basic capacities that underlie children's word learning

  • Understanding of mental states (i.e. my Uncle is angry, my Auntie is talking on the phone not to me, social/cognitive understandings)
  • Understanding of the kind of things that get labelled (i.e. cognitive biases)
  • Understanding of syntax clues to word meaning

Taxonomic Assumptions

It turns out that children between 2-3 years of age assume that words that label things of the same kind are words that are of the same kind.

Some kids when asked to put things that go together, together, very young ones make the 'mistake' of putting drivers with cars, rather than people vs cars
but do they sort red triangle, blue triangle, red circle, blue circle as shapes or colours. often shapes it turns out.

Conceptual Assumptions

So, given children do know that some things go together and others don't - what are these conceptual assumptions and how do they make them:

  • Do they classify according to objects or substance?
  • What else is a determining factor? (e.g. shape over colour)

Initially rather than anything else they tend to classify by what they are used for - so hats would go together rather than purple things. Later on they do notice that colour can be used as a category and so they'll ask which way they should classify things (all people tend to classify objects by intent of the maker/asker)

Conceptually Available Distinctions

Words mark some, but not all, conceptually available distinctions; this leaves children to figure out how their language divides the world into word-sized packages.

For instance some languages have the same word for fingers as for toes, whereas English has two different ones.

More differences in language-specific lexicalisation:

  • English verbs = manner of motion
  • Spanish verbs = path of motion
  • Korean = motion up or down, different verbs for caused or spontaneous motion (e.g. fell down and broke leg vs pushed and broke leg)

Lexicalised Concpets

Children use input to help them to figure out that a particular concept is lexicalised (that there is a word for it), even though they don’t yet know the word itself.

e.g. They know that there is a name for a certain colour, even though they don’t know what it is. They also know that a word’s meaning can often be worked out on the basis of its place in a sentence.