The Competition Model

1
The Competition Model

• Eva M. Fernández
• Queens College Graduate CenterCity University
  of New York

2
Historical backdrop

• Questions about how two languages coexistpinned
  against questions about how a second language is
  acquired
• Pre-Chomskyan era
• Language speech
• Language is a set of habits, learned by exposure
  and practice
• Operant conditioning rewards and punishments
• And practice eradicates bad habits (e.g., of L1)

3
Historical backdrop

• L1 acquisition research
• Developmental stages common to all children
• Developmental errors, even when ungrammatical
  forms dont occur in the environment
• L2 acquisition research
• Learner errors are highly suggestive of internal
  development
• Errors resemble those made by children
• Errors arent always traceable to L1

4
Historical backdrop

• Intellectual motivation Chomsky (and others)
• Language is not habit formation its implicit,
  mental, and biological
• Proof
• Competence performance
• Competence (grammar) very similar across
  languages, hugely complex, vastly
  under-represented in the stimulus
• Platos problem solved by the proposal that most
  of what you know about language is innate

5
Historical backdrop

• Chomskyan research program study competence via
  the idealized speaker/hearer, whose competence
  developed
• Instantly
• In a purely homogeneous speech community
• Without memory/performance limitations
• Huh?!

6
Historical backdrop

• Interlanguage, and attention returns to transfer
• During acquisition
• At the steady state
• At different levels of analysisphonology,
  syntax, semantics, lexicon
• Competition Model is one model set up to account
  for transfer effects, more sophisticated than
  most, because transfer isnt unidirectional, L1 ?
  L2

7
Competition Model (CM)

• Kathryn Bates, Brian MacWhinney1970s, 1980s-ff.
• Cues compete, and the processor weighs them, to
  arrive at the interpretation of sentences
• Cross-linguistic differences in how cues are
  weighed by speakers of different languages
• Such differences bear on the way bilinguals
  process their two languages

8
CM Data Collection

• Off-line decisions are optimal reflections of the
  structure of the language (MacWhinney, 2005, p.
  12)
• Measuring strength of cues to the selection of an
  agent, using a sentence interpretation procedure
• The canaries squashes the elephant.

9
CM Cues

• Word order
• Subject-verb agreement
• Object-verb agreement
• Case-marking
• Contrastive stress
• Topicalization
• Animacy
• Omission (say, of pronouns)
• Pronominalization

Designs cross two or more cues, e.g. word
order subject-verb agreement
10
CM Designs

• Whats stronger in Lx, word order (WO) or
  subject-verb agreement (SVA)?
• Target interpretation (driven by plausibility)is
  supported (?) or not (?) by a given cue

WO SVA
? ? The elephant squashes the canaries.
? ? The canaries squashes the elephant.
? ? The elephant squash the canaries.
? ? The canaries squash the elephant.
11
CM Cue weight studies

• If Lx and Ly have different cue weights(e.g.,
  Spanish relies on SVA, English on WO)what does a
  Spanish/English bilingual do,in Spanish and
  English interpretation tasks?
• Four possibilities
• Forward transfer (L1 ? L2)
• Backward transfer (L2 ? L1)
• Differentiation (L1 ?L2)
• Amalgamation (L1, L2)

12
CM data Steady-state bilinguals

• Kilborn, 1987, 1989
• German-English bilinguals
• Audio stimuli, German English (separate
  sessions)
• Outcome forward transfer, L1 ? L2
• Vaid Pandit, 1991
• Hindi-English bilinguals, Hindi at home, English
  at school
• Outcome highly variable!
• 7 forward transfer
• 19 partial forward transfer
• 17 amalgam in both
• 5 differentiation

13
CM data Interlanguages in flux

• McDonald, 1987, 1989 (see Figs. 2-3, MacWhinney)
• Late English-French billinguals, 1st-4th semester
• Clear forward transfer throughout, but by 4th
  semester, strategies look like for adult
  bilinguals
• Liu et al., 1992
• Chinese-English, English-Chinese, L2 acquired
  early or late
• Late acquirers forward transfer
• Early acquirers (L2 6-10) differentiation
• Very early acquirers (L2 lt4) backward transfer

14
Generalization

• bilinguals do not function with two
  independent language systems. Rather, there is a
  considerable amount of interaction between the
  two systems in the form of transfer (forward and
  backward) as well as, in some cases, an
  amalgamation of strategies.
• Hernández et al., in press

15
MacWhinneys Unified Model

• Beyond cue competition concepts that are core in
  CM, the Unified Model is meant to account for
• Language acquisition
• Childhood multilingualism
• Second language acquisition
• Adult monolingualism
• Is it meant to be a TOE?
• TOE Theory of Everything

16
Cues and competition

• PRODUCTIONCues (forms) compete to express
  functions
• PERCEPTIONFunctions compete based on cues from
  surface forms
• The outcome of such competition is determined by
  the relative strength of cues
• Akin to Optimality Theory?

17
Other models of L2 acquisition

• An internal (mental) grammar (Chomskyan
  tradition)
• Does it develop like L1?
• Is it subject to transfer effects?
• Do the two codes mix?
• Can L1 attrite? If so, how/when/why?
• Input drives acquisition, by triggering internal
  reorganization, controlled by
• Universal Grammar (competence)
• Acquisition strategies (performance)
• Working memory limitations (gral. cognitive
  arch.)

18
Unified Model

• Differs somewhat from more mainstream models of
  L2 acquisition
• No mention of role of Universal Grammar
• Focus on acquisition (learning?) strategies not
  specific to language analogies, learning from
  item-based constructions
• Incorporates notions of
• Chunking
• Transfer (codes)
• Resonance

19
Chunking

• Unanalyzed wholes chunks
• Definitely play a role in acquisition L1 L2
• I gotta go uttered by child
• Não falo português uttered by adult
• Can improve developing fluencye.g., linking
  chunks
• muy buenos días muy buenos días

20
Transfer

• Whatever can transfer, will
• Early term interference
• Nowadays
• Positive transfer pro-drop in Spanish
  Portuguese
• Negative transfer pro-drop in Spanish not
  French
• Transfer of training too many present perfects
  because of overdrilling sparragus
  (hypercorrection)
• Errors of avoidance

21
Transfer Audition

• In bilingual acquisition, Lx and Ly prosodies are
  recognized as different early on
• Until 18 months, infants have superb phoneme
  recognition abilities by 18 months, their
  phonemic repertoire is locked (Janet Werker and
  colleagues)
• Comprehension in very young children is massively
  sophisticated, even before a productive
  vocabulary has developed
• Early stages of L2 acquisition listening
  routines, with L1 bias

22
Transfer Articulation

• Much harder than audition!
• Involves multiple muscles
• Emerged late in evolutionary timeline
• Yet by age 5, most L1 acquirers have it
• For L2 (children and adults)
• Early on, massive transfer of L1 patterns,
  leading to short-term gains, but long-term
  liabilities
• Age effects Neuronal flexibility? Input? Affect?
• Training and rehearsal could help

23
Transfer Lexical learning

• In L2 acquisition, early on, massive transfer of
  conceptual structures from L1
• chair is just another way of saying silla
• Lots of lexical transfer is positive and
  therefore goes unnoticed
• Negative transfer can sometimes be suppressed
  can it?
• Errors minimized when two L1 words map onto one,
  not so when one L1 word maps onto two in L2

24
Transfer Sentence comprehension

• Evidence discussed earlier
• Studies of steady-state bilinguals
• Studies of language acquirers
• learning sentence processing cues in a second
  language is a gradual process that begins
  with L2 cue weight settings that are close to
  those for L1. Over time, these settings change
  in the direction of the native speakers settings
  for L2 (p. 23)

25
Transfer Pragmatics

• Greetings, leave-takings, promises, turn-taking,
  honorifics, terms of endearment
• Mostly very language-specific
• Cooperative principle language universal?
• Not much research on L2 pragmatics! (Brazilians
  acquirers of English Fernando Naditch, recent
  NYU dissertation)

26
Transfer Morphology

• Transfer close to impossible?
• L1 Chinese cant use knowledge about classifiers
  to learn, say, Spanish as L2
• L1ers of languages without determiners (Chinese,
  Russian) have a hard time learning determiners in
  L2
• If a morphological feature is structurally
  mapable from L1 to L2, perhaps
• my computer, shes very slow
• die Mond (ltla luna)

27
Resonance

• Covert inner speech, used to
• process new input
• relate new forms to other forms
• repeated coactivation of reciprocal
  connections. As the set of resonant connections
  grows, the possibilities for cross-associations
  and mutual activations grow and the language
  starts to form a coherent co-activating neural
  circuit (p. 31)

28
Resonance

• Might account for delays in behavioral measures
• Lx, if more frequently used internally than Ly,
  is in a higher state of activation than
  Ly(recall Frenck-Mestre Pyntes quantitative
  differences in eye movements between L1 and L2)
• Might account for intuition that practice makes
  perfect strategic resonance facilitates
  encoding new forms

29
Age effects

• repeated use of L1 leads to its ongoing
  entrenchment which operates differentially
  across linguistic areas, with the strongest
  entrenchement occurring in output phonology and
  the least entrenchment in the area of lexicon,
  where new learning continues to occur in L1 in
  any case (p. 37)
• Learning is highly strategic, therefore high
  variability in L2A
• but why not also in L1A?