Matthew Saxton

1
(No Transcript)
2

Testing Assumptions about the InputEmpirical
Evidence on Negative Evidence

• Matthew Saxton

January 29th 2008
3
Errors in language acquisition

• defining feature of a language learner
• all (typical) children retreat from error
• but how?

4
Negative evidence

• evidence that a given structure is ungrammatical
• parental correction of child errors

5
No negative evidence problem

• longstanding assumption
• parents do not correct their childrens errors
• no negative evidence

6

• A basic premise of almost all work on language
  acquisition in a generative framework is that
  learning must progress without the aid of overt
  correction ? that is, the learner will not
  receive "negative evidence," in the form of adult
  feedback telling the child that his or her
  utterances do not conform with those of the adult
  grammar.
• Weissenborn, Goodluck Roeper (1992, p.9)

7
Does it matter?

• no negative evidence assumption
• ..... one of the most important discoveries in
  the history of psychology (Pinker, 1988, p.104)

8
Empirical support for APS

• parental Approval and Disapproval
• Eve Mama isnt boy, he a girl.
• Mother Yes, thats right.
• (Brown Hanlon, 1970)

9
There is not even a shred of evidence that
approval and disapproval are contingent on
syntactic correctness. Brown Hanlon (1970,
p.201)
10
Forms of correction

• signal of Disapproval
• informant
• meaningful look or pause
• explicit grammar lesson
• differential responding
• clarification requests
• direct contrast between child and adult forms

11
Beyond disapproval

• repeats of ill-formed utterances usually
  contained corrections and so could be
  instructive.
• Brown Hanlon (1970, p.197)

12
Direct Contrast hypothesis

• Child He was the baddest one.
• Adult Yeah, he sounds like the worst.
• juxtaposition of erroneous and correct forms
• unique discourse context
• child may perceive adult form as being in
  contrast with their own

13
Immune to correction?

• Anyone who has attempted to correct a
  two-year-olds grammar will know that it cant be
  done
• Jackendoff (1993, p.22)

14
McNeill (1966, p.69)

• Child Nobody dont like me.
• Mother No, say nobody likes me.
• Child Nobody dont like me.
• repeated 8 times
• Mother No, now listen carefully.
• Say NOBODY LIKES ME.
• Child Oh, nobody dont likes me.

15
Diary study

• Matthew with Alex (aged 4 years)
• aim deliberately correct childs errors and
  gauge effect

16
A That .... that ... that says you cant go
there. M Hmm. A That says you cant go
there. M Why cant you go there? A Cos thats
the part who you / l / .... who you see
.... M Its the .... A .... over. M Its the
part where you what? A Where you look over.
17
A Im easy to eat you up. M You can eat me up
easily? A Yeah. M What? A I can eat you up....
bang M I bet you cant. A I bet you I ....
I, I, I can. I bet you cant eat me up easily.
18
M What you doing? A Im rolling
about. M Youre spinning round, are you? A Im
rolling .... Im spinning around .... .... on
your chair. M Hmm.
19
M You have to shut the doors / w / in
winter. A Yeah, but I dont want to. Its too
bored if I shut the door every day. M Its not
boring. A It is. M What do you
mean? A What? M Why do you say that? A Because
its .... because its .... too.... Its too
boring.
20

• A I drawed a lovely picture for you,didnt I?
• M You drew a picture?
• Where?
• A I drew lots of lovely pictures.

21

• A I dont like Marmite.
• M Mm, yummy. Course you like Marmite. You
  always have Marmite.
• A But I dont ... but I gone off it now.
• M You have not gone off it.
• A I have. I have gone off it. I have.
• M Well, youre a terror.

22
Effects of direct contrasts

• switch from error to correct
• Farrar (1992) 12 - 45
• Morgan et al. (1995) 23 58
• Saxton (2000) 8
• Strapp Federico (2000) 11

23
An experimental approach

• compare effects of positive versus negative input
• control over input information via novel verbs
• irregular past tense forms

24
Positive input

• any linguistic form modelled by an adult

25
Novel verbs

• longstanding paradigm (Berko, 1958)
• aim isolate the effects of input

26
(No Transcript)
27
Supplying negative evidence

• Adult What happened?
• Child He pelled his leg.
• Adult Oh yes, he pold his leg.

28
Supplying positive input

• Adult Look, he pold his leg.

29
Negative gt positive

• production of correct form
• negative 43
• positive 0
• 81 of children produced at least one correct
  form following negative evidence

30
(No Transcript)
31
Empirical support I

• experimental and observational(Farrar, 1992
  Saxton, 1997)
• mother, father, siblings (Strapp, 1999)
• working class (Post, 1992)
• immediate and longer-term effects(Saxton, 2000
  Saxton et al., 2005)

32
Empirical support II

• beyond L1 English French Japanese Korean
  (Chouinard Clark, 2002 Izumi, 2002 OGrady
  Lee, 2006)
• L2 acquisition (Mackey et al., 2003)

33
Theoretical status

• universality
• inevitability
• necessity

34
Mother eased out

• in many communities of the world, parents do not
  indulge their children in Motherese
• Pinker (1994, p.40)
• motherese is not a universal part of L1
  acquisition
• Ayoun (2003, p.51)

35
Trackton

• Now just how crazy is dat? White folks uh hear
  dey kids say sumpn, dey say it back to em, dey
  aks em gain n gain bout things, like they
  posed to be born knowin. You think I kin tell
  Teegie all he gotta know? Aint no use me tellin
  him learn dis, learn dat. Whats dis? Whats
  dat? He just gotta learn, gotta know
• (Heath, 1983, p.84).

36
Haggan (2002)

• the way people say they talk to children
• versus
• the way people actually talk to children

37
Myth of non-universality

• selective focus on anthropological data
• absent features of CDS ? absence of CDS in
  toto
• critical features that are present have been
  ignored

38
Universality

• understanding of language is made easier by the
  habit that mothers and nurses have of repeating
  the same phrases with slight alterations Jesperse
  n (1922, p.142)
• random affection for repetitiousness makes an
  excellent atmosphere in which the child acquires
  speech Mead (1930, p.35)

39
Universal negative evidence

• Arabic, Danish, French, Hebrew, Japanese, Korean,
  Mandarin, Manus, Kiche Mayan, Samoan
• and English
• every single child and every single structure
  examined so far (gt 20 studies)

40
Inevitability

• recasts (including negative evidence)
• an artefact of conversation between a linguistic
  sophisticate and a cognitively naive learner
• adults naturally follow the childs lead

41
Necessity

• facilitative, yes
• necessary?
• onus on nativists to find even one deprived child

42
APS revisited

• no empirical support for no negative evidence
  assumption
• of little value in specifying principles of
  Universal Grammar