6: Language Acquisition

1
6 Language Acquisition

• Outline
• How does language develop?
• Grammar
• Sequence of acquisition
• Issues
• Nativist theories
• Constructivist theories
• Learning outcomes

2
How does language develop?

• 2 categories of theories
• Nativist theories
• children are born with innate KNOWLEDGE that
  helps them make sense of the world
• Constructivist/emergent/empiricist theories
• language is LEARNED by building up knowledge from
  the environment
• Grammatical acquisition
• Learning how to fit words together into
  meaningful sentences

3
Syntax

• What grammatical categories do words fit into?
• Noun, verb, pronoun, adjective, determiner
• What grammatical relations exist between these
  categories (subjects, verbs, objects)
• e.g. John kissed Mary SUBJECT VERB OBJECT

4
Morphology

• Refers to inflections (bits of words)
• e.g. add -ed to make past tense
• add -s to make plural
• Sometimes refers to little or function words
  - e.g. on, a, the, are, up

5
Sequence of acquisition

• At end of 1 word stage children start to put
  words together into 2 word utterances
• telegraphic speech stage (NB often called
  Stage 1 speech (Brown, 1973)
• similar meaning to words used in 1 word stage
• negation, nonexistence, call attention
• soon branches out to other meanings (Brown, 1973)

6
Sequence of acquisition

• Children in telegraphic speech stage are said to
  leave out the little words and inflections
• e.g. Mummy shoe NOT Mummys shoe
• Two cat NOT two cats
• But
• individual variation - analytic vs. gestalt
• cross-cultural variation e.g. Spanish

7
Sequence of acquisition

• Addition of grammatical morphemes
• children add little words and inflections
• e.g. daddy shoe -gt daddys shoe
• two cat -gt two cats
• occurs over long period of time
• More complex constructions
• passives, compound sentences, negatives, questions

8
Issues in grammatical acquisition

• Modularity
• Domain specific?
• Domain general?
• Learning Innateness
• How much is innate?
• Learnability theory

9
Nativist solutions

• Children cannot learn language without innate
  principles/knowledge to guide them (Chomsky,
  1957 Gold, 1967 Pinker, 1979)
• Poverty of the stimulus (Berwick Weinberg,
  1984)
• Overregularisation errors (e.g. runned)
• No negative evidence
• Goal find the universal knowledge and specify
  how children use this to learn their language

10
Chomksy (1959,1965, 1981 etc).

• Language is a distinct piece of the biological
  make-up of our brains distinct from more
  general abilities to process information or
  behave intelligently (Pinker, 1994, p. 18)
• Language acquisition device (LAD) - a mechanism
  with access to the grammatical rules of all human
  languages (Universal grammar - UG)
• Some features of language are universal across
  the worlds languages - PRINCIPLES (e.g. noun
  category)
• Others vary PARAMETRIC VARIATION (e.g. English
  requires a subject - John kissed Sarah, Italian
  doesnt)

11
Chomskys theory

• Good points
• explains why language is learned relatively
  quickly
• explains how language is learnt despite poverty
  of the stimulus, no negative evidence etc
• Bad points
• very little evidence for adultlike grammatical
  knowledge in young children (Braine, 1976)
• young children make errors Chomsky would not
  predict (e.g. omit obligatory constituents such
  as determiners, possessives)
• cannot explain why children make grammatical
  errors (e.g. doggie go walkies) even after
  extensive language exposure

12
New Nativist theories

• Children are prevented from making use of their
  full knowledge
• Continuity theories
• (e.g. Valian, 1986, 1991)
• children have UG
• children have performance limits that restrict
  the number and type of words they can PRODUCE

13
Continuity theories

• Good points
• explains why childrens speech is different to
  adult speech
• explains how language is learnt despite poverty
  of the stimulus, no negative evidence etc
• Bad points
• cannot really explain the restricted nature of
  very early speech
• e.g. children should treat me and I equally -
  as both belong to nominative pronoun category
• BUT, children make many more errors with me
  than with I (Pine, Lieven Rowland, 1998)

14
Competence theories

• Some aspects of innate knowledge are not
  available to the child until later in development
• These aspects MATURE at a later stage
• e.g. Radfords small clause hypothesis (1990)
• Until 24 months, children have access only to
  lexical categories (noun, verb, preposition,
  adjective)
• At 24 months, functional categories come online
  (determiner, inflection, complementizer)

15
Radfords theory

• Good points
• explains why childrens speech is different to
  that of adults
• explains how language is learnt despite poverty
  of the stimulus, no negative evidence etc
• Bad points
• doesnt fit the data
• young children produce functional categories
  (e.g. a doggie), older children still make errors
  (Gathercole Williams, 1994 Rowland, 2000)

16
Constructivist ideas

• speech to children is not impoverished (Snow,
  1977)
• Children do not need innate knowledge to learn
  grammar
• Semantic accounts
• (e.g. Bates, 1979)
• children learn grammar by mapping semantic roles
  (agent, action, patient) onto grammatical
  categories (subject, verb, object)

17
Cognitive accounts

• Children learn grammar by mapping cognitive
  categories onto grammatical ones
• e.g. disappearance words such as gone would be
  learnt after the child learnt object permanence
  (Gopnik Meltzoff, 1987)

18
Semantic and cognitive accounts

• Good points
• explained acquisition of grammar without having
  to resort to innate knowledge
• Bad points
• does not fit the evidence.
• Children seem to learn grammar without going
  through semantic routes (i.e. dont start off
  with agent and patient, Maratsos, 1979).
• Cognitive accounts cannot explain how some
  grammatical relations (e.g. the passive) were
  learnt

19
Semantic-distributional accounts

• Positional and semantic commonalities in the
  language guide the learning of grammatical rules
• e.g. a dog, a cat, a bird, a car -gt a noun
• Theory is often implemented as a computer program
  a cognitive model

20
Semantic-distributional accounts

• Good points
• demonstrate that lots of information about
  language is present in the statistical
  distribution of language (e.g. Finch Chater,
  1992).
• Bad points
• Criticised by nativists
• Cannot handle long-distance dependencies?
• BUT Elman (1993) demonstrates that connectionist
  models can learn long-distance dependencies
• Unrealistic assumptions

21
Learning Outcomes

• So much to learn that you may get lost - so
  concentrate on the theories!
• Outline the sequence of acquisition of grammar
• Critically compare and evaluate the nativist and
  constructivist views of language acquisition

22
Reading

• Essential Reading (on Digital Resources)
• Thornton, S. (2002). Growing minds. Basingstoke
  Palgrave Macmillan. Ch 3. pp.42-73
• Further Reading
• See pdf handout

23
Questions to ask

• Why do nativists argue that we need innate
  grammatical knowledge?
• Why do constuctivists argue that we dont need
  innate knowledge?
• Can constructivist accounts answer all the
  criticisms from nativists?
• What is the evidence for/against these different
  types of theory?