1.21k likes | 1.35k Views
The End State in L2A: Factors, Facts & Fallacies. Cognitive Science Seminar University of Texas 9 October 2009 David Birdsong Dept. of French & Italian birdsong@austin.utexas.edu [paper handout to accompany slides] [slides to be posted on blackboard]. L2A research Sample of journals.
E N D
The End State in L2A: Factors, Facts & Fallacies Cognitive Science Seminar University of Texas 9 October 2009 David Birdsong Dept. of French & Italian birdsong@austin.utexas.edu [paper handout to accompany slides] [slides to be posted on blackboard]
L2A research Sample of journals Cognition JML JCN Nature Neuroscience BBS Brain & Language Language Applied Psycholinguistics TICS Bilingualism: Language and Cognition Studies in Second Language Acquisition Second Language Research
L2A research Programs/Labs McGill Georgetown Illinois MPI-Nijmegen Essex CNRS Paris Amsterdam Groningen Heidelberg
L2A research Recent Ph.D’s at UT - F&I Robert Reichle (December 2008) “Syntactic focus structure processing: Behavioral and electrophysiological evidence from L1 and L2 French” Elenor Shoemaker (May 2009) “Acoustic cues to speech segmentation in spoken French: native and non-native strategies”
Context: the ‘Deficit Model’ tradition Bley-Vroman (1989: 44): Insignificant incidence of nativelikeness in late L2A L2A: “ineluctable failure” fossilized non-nativeness => Fundamental Difference Hypothesis
Context: the ‘Deficit Model’ tradition Johnson & Newport (1989: 255):for AoA > 15 “later AOA determines that one will not become native[like] or near-native[like] in a [2nd] language” <= maturationally-based CPH/L2A
Context: the ‘Deficit Model’ tradition Long (1989: 280):“The ability to attain native- like phonological abilities [in an L2] is beyond anyone beginning later than age 12, no matter how motivated they might be or how much opportunity they might have. Native-like morphology and syntax only seem to be possible for those beginning before age 15.”<= Maturational constraints
Context: the ‘Deficit Model’ tradition Hyltenstam & Abrahamsson (2003: 575):If we look at “overall L2 proficiency” we will find that “perfect proficiency” and “absolute nativelike command of an L2 may in fact never be possible for any [late L2] learner” <= Deficient language- learning mechanisms NB: B-V, J&N, Long, H&Acriterion = monolingual native NB: All approach deficit from end state (= ultimate attainment) perspective
Context: the ‘Deficit Model’ tradition W/r/t L2 processing => representation by late learners Tsimpli & Dimitrakopoulou (2007: 217): Uninterpretable features (e.g. +/- case) “difficult to identify and analyse in the input” <= “persistent, maturationally-based L1 effects on adult L2 grammars” => role of detection in unlearnability of uF; cf.Lardiere (2009) NB:criterion = monolingual native; S’s not always at end state
Context: the ‘Differences Model’ W/r/t L2 processing by late learners Clahsen & Felser (2006: 564): L2 processing nativelike in some areas of grammar; however even in high-proficient L2ers “differences persist in the domain of complex syntax” i.e, in “real-time computation of complex hierarchical representations” => consider explanations:deficient L2 grammar (representational deficit vis à vis native); L1 transfer; cognitive resource limitations; maturational constraints NB: C&F criterion = monolingual native; not all S’s at end state
Context: the ‘Differences Model’ W/r/t L2 processing by late learners Cutler (2003, inter alia): In segmental, subsegmental and suprasegmental perception, L2ers listen through L1 ears; see also Peperkamp, Sebastian-Galles, Dupoux, etc.
Context: Meisel (2009) Meisel (2009: 8)“Changes in the L[anguage] M[aking] C[apacity] in the course of childhood development make it impossible for the L2 learner to acquire a complete native grammatical competence of the target language” Meisel (2009: 13) FDH “does not entail that L2 acquisition becomes totally or partially impossible. Rather, the claim is that L2 knowledge is of a different kind” <= source of knowledge is different
Current aL2A work under ‘DM’ Methodologically: (Usually) studied: - L2 high-proficients Understudied: - L2 dominants (Often) not considered: - facilitating / inhibiting external factors - individual info processing differences - reciprocal L2 <-> L1 influence - AoA as comparison condition or control variable - assurance of L2A asymptote - incidence (#’s) of nativelike processers => Incomplete empirical picture of nature + extent of native / L2 differences
Current aL2A work under ‘DM’ Programmatically: Data: - varied behavioral & brain-based studies - varied methods & tasks Domains: - range of processing and knowledge domains Focus: - non-nativelike processing at group level Orientation: - deterministic differences Goal: - theory of qualitative L1-L2 differences in knowledge / processing => Empirical gaps w/r/t upper limits of (late) L2A
‘Upper Limits Model’ Complementing the DM perspective: Programmatically & methodologically Distinguish: - what adult L2ers do - what adult L2ers don’t do - what adult L2ers can do - what adult L2ers can’t do Populations: - L2 dominants (two types) - L1 attriters - (L2 high-proficients) - incentivized L2ers: functional need for L2 nativelikeness - L2ers desiring socio-cultural integration, L2 identity
‘ULM’ Programmatically: Integrate: - individual differences w/r/t internal factors e.g. WM components Rationale: - claims that only freakish aptitude => nativelikeness [more later] - WM correlates with level of ultimate attainment in some tasks Integrate: - facilitating / inhibiting external conditions Rationale: - in comparing L1A and L2A end state, level the playing field in terms of facilitating conditions [more later] - ‘normal’ adult L2A conditions = abnormal for LA generally Integrate: - AoA / AoT as control, predictor variables Rationale: - we know the effects of age and +/- their sources [more later] - to see what can L2ers do in spite of age influences
‘ULM’ Programmatically: Goals: - Establish end state processing profiles: - L2 dominants - defined by L2 vs L1 relative use - defined psycholinguistically (independent processing measures) - late vs. early dominants - unstable dominants - Establish upper bounds of L2ers’ processing & knowledge: - vis à vis natives - vis à vis early bilinguals / early L2ers - in their own right (v. comparative fallacy) - keeping in mind L2<->L1 influence (L1 of a monolingual ≠ L1 of a bilingual) - Integrate above into theory of L2 knowledge & processing at the limits
‘ULM’ Perspectives: Analogy: Track team with ankle weights: What would happen if we took off the weights? Analogy: Ghetto vs. suburbs HS exit exam scores: We know that the achievements will differ, but the story shouldn’t stop there.
‘ULM’ Perspectives: X freakish talent (WM capacity, LTM, musical ability) X freakish accomplishments (Hale, “Christopher”) √ normals working within their limits √ give acquisition the same chance it has in younger populations… - one can’t take away age effects - one can minimize L1 effects - one can provide benign external conditions …and see what happens => relevance to claims, assumptions; CPH/L2A, FDH, ‘access’
UT Cog Sci Presentation L2 end state attainment: Approach ultimate attainment neutrally: deficits & differences alongside upper limits FACTS about end-state attainment, as mediated by age of acquisition and age of testing FACTORS that constrain vs. enable L2 acquisition, and their nature FALLACIES w/r/t end state, e.g. CPH/L2A Q: Under DM: Evidence for maturation-based differences? Q: Under ULM: What can (late) L2-dominants do?
Distinctions: AoA ≠ AoT ≠ maturational state AoA = Age of Acquisition (Immersion, Onset) = macro-variable, encompassing inter alia: - degree of L1 entrenchment - L1 proficiency - state of system plasticity - state of cognitive development - degree of (neuro-)cognitive decline (adults) - suite of neurobiological variables AoA = proxy for initial state of L2A NB: AoA-related effects NB: “The age factor” = convenient but underspecific label
Distinctions: AoA ≠ AoT ≠ maturational state AoT = Age at Testing = macro-variable, encompassing inter alia: degree of L1 entrenchment L1 proficiency L2 proficiency state of system plasticity state of cognitive development degree of (neuro-)cognitive decline [see Supplements] suite of neurobiological variables also: socio-psychological identification w/ L2 & L1 also: frequency of L2/L1 use AoT = proxy for current state of L2 knowledge and processing NB: AoT-related effects
AoT Advanced AoT: - More profound impact in L2 processing than in L1 processing.L2 processing more vulnerable because: - greater reliance on fluid intelligence than in L1 use - less-routinized procedures than in L1 use Open question: - AoT (& AoA)-related effects = less severe among L2 dominants?
Distinctions: AoA ≠ AoT ≠ maturational state Maturational state: > (vs. // vs. X) // degree of L1 entrenchment // L1 proficiency // L2 proficiency > state of system plasticity > state of cognitive development > degree of (neuro-)cognitive decline > suite of neurobiological variables X socio-psychological identification w/ L2 & L1 X frequency of L2/L1 use // experientiallycorrelates with aging AoA & AoT > biologicallyaging-related, pre-/post- ‘maturation’ AoA & AoT X intrinsically unrelatedto aging AoT only
Distinctions Ultimate Attainment in L2 = end state (asymptotic) knowledge and processing ≠ only nativelikeness =any level at end state, up to & including nativelike
Distinctions Nature, causes & domains of non-nativelike ultimate attainment (Hopp, 2007) Nature: REPRESENTATIONAL * COMPUTATIONAL / \ / | \ Cause: impairment * L1 impairment * inefficiency * L1 [FT/FA] / \ | | | Domain: module * interface parsing * info * inter- routes integration ference [FFF] [uF] [DP] [lim cap] [CM] [SS] [MSI]
Distinctions Age, end state, upper limits: Computational (in)efficiency • => default to lexis/plausibility in complex computations L1 influence • developmental & at AoA / AoT • as alternative to / complement to impairment Perceptual components of processing <-> grammar • detection of uF •as precondition for interpretation (e.g. in French liaison) Nature: REPRESENTATIONAL * COMPUTATIONAL / \ / | \ Cause: impairment * L1 impairment * inefficiency * L1 [FT/FA] / \ | | | Domain: module * interface parsing * info * inter- routes integration ference [FFF] [uF] [DP] [lim cap] [CM] [SS] [MSI]
Distinctions Heuristics: Universal Learnability versus Selective Processability - Universal Learnability: anythingcan be learned by someone - Selective Processability: some thingscan’t be processed by anyone
Distinctions The AoA function, shapes: straight line stretched ‘7’ inverted ‘V’ stretched ‘Z’ The AoA function, timing of deflection: coinciding with known maturational epochs coinciding with ages unrelated to maturation The AoA function, steepness => # of nativelike-ers
Range of Scores of Native Controls AoA 20 Years Slope predicts incidence of nativelikeness: shallow slope => high rate of nativelikeness
Range of Scores of Native Controls AoA 20 Years Slope predicts incidence of nativelikeness: steep slope => low rate of nativelikeness
FACTS of Biological Aging& their relationship to L2 processingby late L2ers
Cognitive Aging Effects Effects in language processing: • processing speed • working memory capacity • lexical retrieval • linear over AoT/AoA, starting at 20 years of age • prior to AoT/AoA: increase then decline (inverted “V”) plateau then decline (stretched “7”) • linkage of processing behaviors to biological aging (Bäckman & Farde, 2005) => L2 processing constraints on input processing at AoA/AoT output performance decrements at AoT depressed levels of processing at L2 end state
Constraining FACTORS 1-Neuro-biology/anatomy/chemistry/cognition: post-puberty, with increasing age (AoA & AoT): • ‘use it then lose it’ (Pinker, 1994) • pathological increases in cortisol levels • declines in neurotransmitter levels: ACH dopamine, etc. Supplement II •declining regional brain volumes (Raz, 2005) Supplement III
Biological Aging Effects Declines in neurotransmitter levels: ACH, dopamine, etc. Supplement II Neurotransmitter declines in L2 processing: • variety of cognitive functions underlying L2 processing - working memory capacity & executive function - attention & inhibitory processes (L1 suppression) - coordination/proceduralization in syntax • linear over AoT/AoA, starting at 20 years of age • prior to AoT/AoA: increase, then decline (inverted “V”) plateau, then decline (stretched “7”)
Biological Aging Effects Declines in regional brain volume Supplement III Effects in L2 processing: • variety of cognitive functions underlying L2 processing - executive function - LTM - coordination/proceduralization in syntax • linear over AoT/AoA, starting at 20 years of age • prior to AoT/AoA: increase then decline (inverted “V”) possible plateau (stretched “7”) • not reliable linkage of behavior to biological sources -effects more associated w/neurochemistry than w/structure -uncertain timing of decrease thresholds => associated cognitive decrements
Constraining FACTORS 2-Cognitive development: • Adult working memory bandwidth lets in too much linguistic information to process at once => incomplete processing of, e.g., sequences of morphemes: “less is more” (Newport, various)
Constraining FACTORS 2-Cognitive development: post-concrete operations • Analytic/metalinguistic/explicit input & learning • Literacy => L2 vis à vis L1 learning & processing: ? effortful ? inefficient √ different [cognitive, neural] resources
Constraining FACTORS 3-L1 entrenchment: L1 representations are increasingly defined with use => • developing representations/categories assimilate to old (Flege, various) • competition between old and developing representations (MacWhinney, various) • [Hebbian] learning inhibits related new learning (Elman, various)
Facilitating FACTORS Enabling via 1- Subtraction of constraint • minimize L1 influence via L1 attrition or L2 dominance [More to come]
Facilitating FACTORS Enabling via 2- Offsetting the effects of limiting factors • training on L2 perception • training on L2 pronunciation
Facilitating FACTORS Enabling via: 3-Individual variation • aptitude components (phonological) working memory • health / genetic / lifestyle hypertension: inverted “U” function ACH, testosterone, estrogen, dopamine, cortisol [early & late] • attitude & affect motivation; L2 ‘identity’; ‘passing for’ • L2 use / practice / rehearsal / education
Facilitating FACTORS NB: Some factors trump others e.g., neurobiology = irrelevant if low desire for attainment; v. • passing for a native • ID with L2 culture & speakers
Facilitating FACTORS NB: Distinguish necessary from sufficient conditions: A given factor may be necessary but not sufficient for nativelike attainment
FACTS of Upper Limits: can do Uncontroversially a given late (AoA > 12) L2 learner can: perform like monolinguals across multiple complex behavioral measures of grammatical knowledge & lexical knowledge & global pronunciation can: perform like monolinguals on a range of brain-based measures of L2 processing e.g., ERP components, regional brain activity [more to come]
FACTS of Upper Limits: cannot do Uncontroversially late learner GROUPS cannot: perform like monolinguals on ‘challenging’ online and offline tasks/structures/items; various processing tasks involving parsing, suprasegmental perception, etc.
FACTS of Upper Limits: cannot do? Controversiallya given late L2 learner cannot: perform like monolinguals on certain on-line processing tasks involving sentence parsing, suprasegmental perception, etc. (quantitative & qualitative differences)