L3 Generativist assumption proved wrong

  • Created by: BKW
  • Created on: 10-01-20 18:18
View mindmap
  • l3 - looking into generativist claims of full productivity
    • children should be equally good at supplying all the different morphemes they know
      • Wilson(2003) naturalistic data study looking at children's use of present tense person/number agreement morphemes eg He's, You're, I'm
        • each child showed diff rates of correct production for 3 diff types of morphemes. children also showed diff rates of production for different morphemes that are diff forms of same AUX
          • Eg - Nina - He's supplied 82% of time , I'm only 9% but they're both singular form
        • Didn't control for knowledge of individual morphemes
        • Pine et al 2008 - replicated study but counting only from the time child first correctly used each morpheme, therefore 'knew' the morpheme found the same
          • English is actually bad to test this on because theyre aren't that many morphemes
            • Aguado-Orea&Pine - naturalistic data study in SPANISH which has 18 morphemes. child data compared to adults to ensure adult has full productivity
              • found child used significantly less different morphemes than the mother or father
                • Maybe children just like to talk about a narrower range than adults - cant control what child is saying unless using elicited production
    • Children shouldn't be better at applying morphemes to some verbs over others
      • Theakston et al  - elicited production study on 2y6m -3y/o. used novel verbs 12/24 children could produce 'blicks' after hearing 'blicked' however 23/24 produced the -s on a familiar verb
        • Rowland&Theakston 2009 found similar findings for Auxillaries 'BE, HAVE &DO) and for modal auxillarys (CAN,HAVE ect)
      • Rowland&Theakston 2009 found similar findings for Auxillaries 'BE, HAVE &DO) and for modal auxillarys (CAN,HAVE ect)
      • Rasanen, Ambridge &Pine -elicited production study 3yrs1mo-4yrs1mo - production of 3sg -s morpheme on 48 diff verbs
        • found correlation between proportion of 3sg forms in the input and correct production of 3sg-s with 100% getting it right if they heard if more often than if they heard it less often
          • this discredits generativists while evidencing constructivists
        • Dabrowska&Szxzerbinski / Krajewski et al did similar studies with nouns and noun case marking showing similar results
    • children should never supply the wrong agreement marking mophemes
      • Wexler 1998- children do not say "I like-s ice cream" claiming young german children do not make agreement mistakes
      • Deen 2004 - highlighted errors of emission eg mummy eat cake are very common whereas error of substitution are very rare eg I eat-s cake
      • Hoekstra &Hyams 1998 - when finite forms are used agreement is almost always correct - some children just think its optional
      • Harris &Wexler 1996 - agreement is always correct with main verbs
      • Hoekstra&Hyams 1998 naturalistic data from German, Italian, spanish and Catalan language found incorrect agreement to be only 2%
        • Rubino&Pine 1998 claimed if most of what children say is rote learned (constructivist) of course they will only make few errors, if they learn I like as a frozen phrase or "I like -NOUN" as a slot and frame there will only be few errors
      • If we look at forms that the child rarely attempts to say they produce much higher error rates
        • Aguado-Orea did a similar study to Rubino but in SAPNISH and only counted from the point at which 2 children produced first correct 3pl form
          • One child Juan still produced 75 errors when the 3sg was used instead
    • Why do constructivists think these agreement errors happen?
      • theyre incorrectly repeating the wrong form from adult utterances - rote learned phrases
      • They're just defaulting to the most frequent form
        • Rasanen, Ambridge, Pine 2014 - looked at English agreement errors found theyre defaulting even though it looks like simple emission
        • Rasanen, Ambridge &Pine - did study in Finnish observing same kind of errors, but it also relates these errors to input
        • Aguado-Orea&Pine found 1. error rates were highest in low frequency person/no contexts, 2 Errors predicted not only by frequency of person/no contexts also by frequency of individual ready inflected lexical targets
          • supports idea we pattern according to input - children make error when they haven't learned relevant low frequency form instead use well learned high frequency


No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Child language aquisition resources »