Language - PowerPoint PPT Presentation

1 / 55
About This Presentation
Title:

Language

Description:

What is the difference between studying natural competences and side-effects? ... 'Big Bird is washing Cookie Monster' 'Cookie Monster is washing Big Bird' ... – PowerPoint PPT presentation

Number of Views:74
Avg rating:3.0/5.0
Slides: 56
Provided by: ledaco
Category:

less

Transcript and Presenter's Notes

Title: Language


1
Language
  • Does the child come factory equipped with any
    knowledge of the world?
  • What is the difference between studying natural
    competences and side-effects?
  • What does learning mean?
  • How many learning processes are there?
  • Is instinct the opposite of learning?
  • What is the design of the instinct that causes
    learning in a given domain?

2
Communication versus Language
  • Communication in other animals
  • Sound one meaning
  • Small vocabulary
  • No rules for combination
  • Human language
  • Sound meaning arbitrary relationship
  • Large vocabulary
  • Finite set of symbols ? infinite number of ideas

3
What is language? A grammar?Definitions
  • A language is a rule-governed system that allows
    one to produce an infinite number of sentences
    from a finite set of symbols
  • A grammar is a system that arranges or groups
    words together into meaningful phrases and
    sentences

4
Language as a rule-governed system
  • Hierarchical structure
  • Sounds versus phonemes
  • Morphemes content versus function
  • Phrases
  • Sentences

5
(No Transcript)
6
Phonemes
  • Sounds versus phonemes
  • Phoneme a sound distinction that is relevant to
    speech perception
  • Bus v. Bussss not phoneme diff
  • Bus v. fuss v. but phoneme diffs
  • 40 phonenes in English
  • Sound distinctions formed by different
    arrangements of the vocal tract
  • Pa ta (lips, palatte) ba pa (voice onset time)

7
(No Transcript)
8
Different languages use different sets of phonemes
  • Efharisto !Kung clicks
  • Every child starts out able to make all phonemes
  • Babies start to lose the ability to make phonemes
    not in their language
  • Start losing by 12 months By 20s, lose ability
  • We can perceive 6 phonemes / second
  • 250 words/ minute
  • Even though NO gaps or silences that mark off one
    speech unit from another
  • Unlike writing. Jeet?

9
Morphemes
  • Phonemes combine to form morphemes
  • Morpheme the smallest language unit that carries
    a bit of meaning
  • Strangers 3 morphemes
  • Strange, er (one who is), s (more than one)
  • Stranger one who is odd or alien strangers
    ones who are odd or alien
  • Content morphemes (strange)
  • Function morphemes (er, and)
  • Serve grammatical functions

10
Language as a rule-governed system
  • In English 80,000 morphemes
  • Combine to form several hundred thousand words
  • Words strung together to form phrases
  • Phrases strung together to form sentences

11
Language as a rule-governed system
  • Sentence
  • Mini-drama Doer, act, done-to
  • Doer Noun phrase
  • Act Verb
  • Done-to Object

12
(No Transcript)
13
Same meaning, different grammatical forms
  • Propositions (sentences in the language of
    thought)
  • Mental representation of meaning
  • Same proposition can be expressed by different
    grammatical forms
  • Different surface structure to the sentence
  • E.g., active, passive

14
(No Transcript)
15
(No Transcript)
16
Language as a rule-governed system
  • Grammatical rules or a system of syntax
  • Rules for stringing phonemes, morphemes, words
    together to make meaningful phrases and sentences

17
Grammar
  • Prescriptive grammar (school)
  • Dont end sentence with preposition, dont split
    infinitives
  • Psychogrammar (psycholinguists sense)
  • Refers to mental structures
  • A finite rule system or program or circuit design
    capable of generating and recognizing the
    sentences of a language
  • Possessed by every native speaker
  • Rules usually implicit

18
A Grammar Instinct?
  • The grammars of all human languages are slight
    variations on a single underlying system
    psychogrammar
  • All human societies have language
  • Lang, grammar are universal in humans
  • Not a cultural invention (unlike, e.g., computer
    programming, writing)
  • Every brain intact person can acquire language
  • Does not correlate with IQ, SES, etc
  • Intuitively simple sentences (Where did he go?
    The guy I met killed himself) require rule
    systems more complex than any existing computer
    language system can handle

19
Grammar instinct (contd)
  • No correlation between the complexity of a
    grammar and the degree of technological
    sophistication of a culture
  • E.g. English one word for we. Cherokee 4
    words (you I, another person I, several
    other people I, you, one or more other
    persons, I
  • Or SES, ethnicity He is working vs. He be
    working (a currently ongoing activity) and He
    working (has a job)
  • All human grammars are a subset of all possible
    grammars, and are slight variants on what appears
    to be a universal psychogrammar.

20
Language development
  • Some milestones

21
12 months Gibberish babbling One word utterances
7 months syllable babbling
22
2 years, 9 months to 3 years, 6 months all
hell breaks loose stage!
18 months 2 word strings
23
Language Acquisition
  • By virtue of what cognitive mechanisms do
    children acquire the words of their language and
    its grammar?
  • Are the learning mechanisms involved SPECIALIZED
    for language acquisition, or are they designed
    for more general purpose learning?
  • Are some features of language acquired via
    language-specific learning mechanisms, and others
    by learning mechanisms with a more general
    function?

24
Privileged hypotheses in word learning
  • Infinite of hypotheses a problem!
  • Quine On another planet, cannot assume beings
    there carve up the world as you do. Can you
    learn the meaning of words? Point to bear, says
    gavagi
  • Bear, bear in hat, fur, bear stuff, Bear, gift,
    classroom display, bear or light bulb, bear or
    chair...
  • Even worse whenever pointing to bear, you are
    also pointing to an example of
  • undivided bear parts, time slice of bear,
    bearhood...

25
Privileged hypotheses in word learning
  • Child has same problem! Gavagi?!
  • Without privileged hypotheses about what kinds of
    things words are likely to refer to, child would
    not be able to learn language at all!
  • Other names for Gavagi problem
  • Poverty of the stimulus (Chomsky)
  • Stim. does not uniquely determine a response
  • Combinatorial explosion (Math)
  • Frame problem (Artificial intelligence)
  • This problem applies to all learning, not just
    word learning

26
Some privileged hypotheses in word learning
  • Whole object assumption
  • Taxonomic assumption
  • Mutual exclusivity assumption
  • 2-6 years of age learn new word every 2 hours

27
Privileged hypotheses in word learning
  • Whole object assumption
  • Bear, not bear parts or substance.
  • Uses object concept system (cohesive, bounded
    entity that moves as a unit)
  • Taxonomic assumption
  • Basic level object (not superordinate or
    subordinate category)
  • Cup (not dinnerware (superordinate) or this
    particular cup (subordinate)

28
Privileged hypotheses in word learning
  • Mutual exclusivity assumption
  • No synonyms
  • If child already has a word for the whole object,
    assumes new word refers to something else e.g.,
    to a part or something it is made of
  • Point to pewter cup See this, its pewter!
  • What else is pewter? pewter tongs, china cup
  • If child already knows cup, she picks pewter
    tongs
  • If child does not yet have a word for cup, she
    picks the china cup (applies whole object
    assumption)

29
Use theory of mind/ joint attention in word
learning
  • Show 2 unfamiliar toys, let play
  • Put them back in bucket, then take one out for
    baby to play with
  • While baby looking at one, YOU look in bucket and
    say, Oh, here is a Toma!
  • Ask baby to show/find the toma
  • Uses cues from joint attention to tell when you
    are or are not naming something
  • Baby does NOT think toma is what SHE was looking
    at
  • 16-17 months at chance
  • 18-19 months chooses bucket toy

30
Grammar acquisition Privileged hypotheses?
  • Grammar is learned
  • without any explicit instruction
  • even in cultures where adults do not speak to
    children who cannot yet talk
  • Even though an infinite number of different
    grammars could have produced any set of sentences
    the child might hear.
  • Cat v. Tac Plural as mirror reversed?!
  • Gavagai again
  • So to learn grammar of native language at all,
    must have privileged hypotheses re possible
    rules i.e., psychogrammar

31
Language Acquisition Device?? (LAD)
  • LAD programs specialized for learning a language
  • Switched on early in life, switched off puberty
  • Amodal input can be acoustic or visual
  • Deaf speakers of American Sign Language
  • Same time course as hearing children
  • ASL grammar a variant of universal grammar

32
Learnability analysis One source of evidence
  • Think like an engineer
  • Is the proposed learning mechanism capableeven
    in principleof acquiring language
  • Proposal Children learn grammatical rules
    through feedback their parents correct their
    grammar when they make a mistake
  • But parents rarely correct grammar
  • Proposal Children learn language through operant
    conditioning. (General purpose learning)
  • Noam Chomsy finite state grammar (the kind of
    grammar that an associationist mechanism would
    have) cannot produce many grammatical sentences
    and would produce many ungrammatical ones

33
Learnability analysis One source of evidence
  • Proposal Children learn language through operant
    conditioning. (General purpose learning)
  • Noam Chomsy finite state grammar (the kind of
    grammar that an associationist mechanism would
    have) cannot produce many grammatical sentences
    and would produce many ungrammatical ones
  • I want neither coffee nor tea.
  • felt love Bill
    milk love
  • I am food fame cup
    food
  • want coffee coffee mocha
    you
  • went neither you nor
    tea

34
Empirical evidence Nim Chimpsky
  • Do we learn grammar through operant conditioning?
  • If true other animals that can be operantly
    conditioned should learn grammar as easily
  • Nim Chimpsky
  • No grammar
  • Limited vocabulary
  • V. smart though

35
Empirical evidence Deprivation studies
  • What happens if a child is not exposed to a human
    language before puberty?
  • Deaf children of hearing parents
  • First exposed to ASL (only lang) at different
    ages, tested as adults (30 years of learning)
  • Exposed lt 7 ? fine.
  • gt 7 many mistakes on function morphemes
    (grammar)
  • Other domains (math, biology) longer learning ?
    better performance
  • Dont see same effect for grammar

36
Is language just the expression of general
ability to manipulate symbols? (i.e., no LAD?)
  • Double dissociation between language ability to
    manipulate symbols in other domains
  • Stroke damage to Brocas area
  • Knock out ability to speak while leaving other
    intellectual abilities intact
  • Specific language impairment (SLI)
  • Hereditary disorder knock out ability to (e.g.)
    make plurals, past tense (a zag. 2 ___? He
    wugs. Yesterday he ___).
  • Can have other functions intact (computer
    programming)
  • Williams syndrome. Language fine, but mental
    handicap

37
What cues DO children use to acquire their native
grammar?
  • Privileged hypotheses ensure they consider only
    some possibilities, not all possibilities
  • Of those possibilities, how do they figure out
    which are true of the grammar they are hearing?
  • Semantic bootstrapping
  • Can word meanings help in grammar learning?
  • Syntactic bootstrapping
  • Can syntax knowledge help in word learning?

38
Semantic Bootstrapping
  • Problem How do you learn the syntax of your
    native language? e.g., even if the notion of a
    count noun is grammatically marked in every
    language, how is it grammatically marked in your
    language? (as in a dog.
  • Solution Take advantage of a one-way mapping
    from cognition to syntax. For example
  • Not all count nouns are objects
  • a dog vs a problem, a day
  • But all objects are count nouns.
  • Not all verbs are actions (kick vs seem)
  • But all actions are verbs.

39
Syntactic cues to word meaning
  • Test 17 month olds, only 2 words in productive
    vocabulary
  • Cue Word order
  • Big Bird is washing Cookie Monster
  • Cookie Monster is washing Big Bird

40
Syntactic cues to word meaning
  • Cue transitive /intransitive distinction (24
    months)
  • Rabbit duck are both moving their arms in
    circles. Rabbit is pushing duck up and down
  • Intransitive The rabbit and duck are gorping
  • Transitive The rabbit is gorping the duck
  • Wheres gorping now? Find gorping!

41
Syntax of Verbs, Count Nouns, Mass Nouns
  • Picture is of hands in a bowl, kneading confetti
  • Verb sytax Do you know what it means to sib?
    In this picture, you can see sibbing.
  • Count noun syntax Do you know what a sib is?
    In this paicture, you can see a sib.
  • Mass noun syntax Have you seen any sib? In
    this picture, you can see sib.
  • Show me another picture of
  • Sibbing (choose hands kneading something else)
  • A sib (choose bowl)
  • Sib (choose confetti)

42
Do we learn grammar by word analogy? (i.e.,
ignoring the grammatical category)
  • Jane eats chicken
  • Jane eats fish
  • Jane likes fish (note fish is a noun)
  • Jane eats slowly (note slowly is an adverb,
    not a noun)
  • Jane might fish (note fish is a verb, might
    is an auxillary)
  • WORD analogy would make you think
  • eats, likes, might can all go in position 2,
    following subject (Jane)
  • chicken, fish, and slowly can all go in position
    3
  • (i.e., no distinction between ordinary and
    auxillary verbs in position 2 or between
    adverbs, nouns, and verbs in position 3)
  • So you would think the following were
    grammatical
  • Jane might slowly WRONG!
  • Jane likes slowly WRONG!
  • Jane might chicken WRONG!

43
Language uses two tricks Words Rules
  • Memory Store words
  • Regular forms store stems
  • Irregular forms store word
  • (ran means runpastness)
  • Expensive storage, cheap computation
  • Use rules
  • Add ed to verb stem to make past tense of verb
  • Add s to noun stem to make plural
  • Requires expensive computation, cheap storage

44
Rules applied Evidence from over-regularization
of past tense
  • Time course for over-regularization of past tense
  • 12 months isolated words
  • 18 months 2 word microsentences
  • See baby More milk
  • 24 months (2 years) Longer, more complex
    sentences, with function morphemes
  • -ing, -ed, -s, auxiliaries (have, must)
  • After 1.5 and before 3 years Children make
    over-regularization errors
  • Goed (went) Runned (ran)
  • Mean 4.2, median 2.5

45
How does child learn these rules?
  • Not through negative evidence e.g., child who
    was unable to talk but was able to understand
    language, could judge perfectly whether a
    sentence was grammatical or not. But he cannot
    have had any negative evidence because he could
    not produce wrong sentences where he could get
    feedback.
  • Not imitation or analogy
  • Note some verbs are irregular no rule generates
    their past tense form. These are the most common
    verbs (to be, to do, to go, to make, to have, to
    hold...)
  • held, went, heard past tense first stored as
    dictionary items, along with regulars played,
    tripped pastness is stored as part of these
    words meaning
  • Then the add -ed to stem rule comes on line,
    becomes obligatory

46
Why over-regularize?
  • Rate of over-regularization very low.
  • Sporadic malfunctioning, not qualitative
    difference is system. Same low rate ages 2-5
  • Still use irregulars Why not always regularize?
  • Blocking rule an irregular idiosyncratic form in
    mental dictionary that corresponds to a
    grammatical modification BLOCKS applicatin of the
    past tense rule
  • Memory retrieval the more something is heard,
    easier to retrieve. If slow in retrieving
    irregular form, will sometimes regularize it
  • older adults with memory loss over-reg

47
Qualitative Change in Learning?
  • Are children now learning by analogy on the basis
    of SOUND?
  • Test Auxiliary verbs sound like ordinary verb
    forms
  • They SOUND identical
  • Does child treat them differently even so?
  • I.e., Does child respect grammatical category in
    applying rules?

48
Qualitative Change?
  • e.g., Look what he doed (do ordinary verb)
    but not Look what he doed make (do auxiliary
    form)
  • Are children now learning by analogy on the basis
    of sound?
  • No.
  • Auxiliary ordinary forms SOUND identical, but
    child treats them DIFFERENTLY!
  • Past tense rule applied to ordinary verb form
  • But not to identical sounding auxiliary verb form
  • So what is going on with over-regularization?

49
Over-regularization is driving out a different
kind of error bare stem errors
  • A bare stem error (no inflection tacked on)
  • An over-regularization error
  • Note No backsliding or regression
  • Over-regularization is replacing another kind of
    error, not correct useage

50
Words Rules So why over-regularize?
  • Memory retrieval for irregular past tense forms
    not as good when new language learner
  • When retrieved, Blocking rule prevents over-reg
  • When not, -ed rule applied to bare stem
  • Learning by sound analogy not the explanation
  • I doed it, not I doed make it
  • Creation of new verbs by adults Problem of
    inducing past tense the same as for children
  • What happens?

51
Words Rules Evidence from adults
  • Creation of new verbs by adults
  • I wug, I wugged (no irregular form stored)
  • What about Verbs derived from a noun stem?
  • No past tense for a noun stem
  • Hockey stick he was high-sticked (not
    high-stuck)
  • It is not a way of getting stuck
  • A way of getting hit (by a stick)
  • Baseball. A fly. He flied out (not he flew out)
  • See same patterns in changes in language over
    time
  • As irregular past tense becomes rare, hard to
    retrieve
  • Did you chide him? I choad him I chided him

52
Language
  • Does the child come factory equipped with any
    knowledge of the world?
  • What is the difference between studying natural
    competences and side-effects?
  • What does learning mean?
  • How many learning processes are there?
  • Is instinct the opposite of learning?
  • What is the design of the instinct that causes
    learning in a given domain?

53
Questions to think about throughout 142...
  • What does the child know about the world?
  • How does the child come to know what she knows?
  • Is the childs mind different from the adults
    mind, or does the child just know less?
  • Does the child come factory equipped with any
    knowledge of the world?

54
Questions to think about throughout 142...
  • How does the environment affect development?
  • How does maturation affect development?
  • Why did scientists underestimate how much infants
    know?
  • What is the competence/ performance distinction?
  • Can one part of the brain know something that
    another part of the brain does not know?

55
Questions to think about throughout 142...
  • What is the difference between studying natural
    competences and side-effects?
  • What does learning mean?
  • How many learning processes are there?
  • Is instinct the opposite of learning?
  • What is the design of the instinct that causes
    learning in a given domain?
Write a Comment
User Comments (0)
About PowerShow.com