Title: 1st A' GUIORA ANNUAL ROUNDTABLE CONFERENCE IN THE COGNITIVE NEUROSCIENCE OF LANGUAGE: THE COGNITIVE
11st A. GUIORA ANNUAL ROUNDTABLE CONFERENCEIN
THE COGNITIVE NEUROSCIENCE OF LANGUAGE THE
COGNITIVE NEUROSCIENCE OF L2ANijmegen, September
20, 2005 Age and L2AAn Overview
- David Birdsong
- University of Texas
- Visitor MPI-Nijmegen
2ACKNOWLEDGMENTS
- Organizers and Steering Committee of the
Roundtable Conference Series - Language Learning
- Max Planck Institute for Psycholinguistics
3Preliminaries
- Overview (selective) of facts and issues relating
to age and L2A - Theory Evidence Methodology
- L2A end-state data
- Behavioral brain-based evidence
- L2 attainment L2 processing
4Organization
- I. Age and Nativelike Attainment in L2A
- II. Interpreting Age-Related Effects
- III. The Aging Brain and L2A
- IV. The Nature of Age Effects
5Terminology
- gt Ultimate attainment ? nativelikeness
- attainment at end state/asymptote
- gt Nativelikeness experimental performance like
that of native controls - gt AoA Age of Acquisition/Arrival (immersion in
L2 context)
6I. Age and Nativelike Attainment in L2A
7Age of Acquisition/Arrival (AoA) and
nativelikeness received views
- Bley-Vroman (1989 44) Insignificant incidence
of nativelikeness in late L2A ineluctable
failure Fossilization Fundamental Difference
Hypothesis - Johnson Newport (1989 255) for AoA gt 15
later AOA determines that one will not become
nativelike or near-nativelike in a language
CPH/L2A
8Age of Acquisition/Arrival (AoA) and
nativelikeness received views
- 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 learner - Deficient language-learning mechanism
9Nativelikeness in late L2A
- 1. Not typical, but not rare
-
- 20 studies / 13 years gt3 - 45 of subjects
perform like native controls (median gt 11) - morphosyntax pronunciation
-
- most learner samples not composed of pre-screened
L2 high-proficients -
10Nativelikeness in late L2A
- 2. Not restricted to a narrow or single domain
- Birdsong (2003) Marinova-Todd (2003) Multiple
tasks covering segmental global pronunciation,
vocabulary knowledge, morphosyntactic
production/judgment, language use, etc. - Marinova-Todd 3/30 subjects indistinguishable
from native controls across 9 tasks. - Birdsong 3/22 like natives on 5/7 tasks
11Nativelikeness in late L2A
- 3. Heuristic No limits on what is learnable
- There is no task involving L2 linguistic
knowledge (or pronunciation accuracy) where
nativelikeness is out of reach for all late L2
learners at end state.
12Nativelikeness in late L2A
- 4. Heuristic Limits on what is processable
- L2 processing tasks where all late learners fall
short of nativelikeness (e.g., Clahsen Felser,
in press Dussias, 2004 Dupous Peperkamp,
2002 Frenck-Mestre, 2005 Juffs, 2004
Papadopoulu Clahsen, 2004) - gt lexical retrieval
- gt structural ambiguity resolution
- gt detection of acoustic distinctions in syllable
stress, vowel length, consonant voicing - gt sentence processing - eyetracking
-
- Quantitative differences speed accuracy
- Qualitative differences shallow/deep
processing local /non-local parsing mishearing -
- ? Trainability
13Nativelikeness (L1-likeness) in late L2A
- 5. Neurofunctional similarities
- Abutalebi et al. (2005) Green (2005) Stowe
Sabourin (in press) In terms of where (fMRI)
and when (ERP), similarities with L1 in brain
activation among L2 high-proficients, even late
learners. - to be continued
14Nativelikeness in late L2A
- 6. Incidence of nativelikeness derivable from
slope of AoA function (Birdsong, 2005) -
-
15Slope predicts incidence of nativelikeness
shallow slope
16Slope predicts incidence of nativelikeness
steeper slope
17Nativelikeness in late L2A
- 7. Incidence of nativelikeness affected by
exogenous factors - Bialystok Hakuta (1999) Flege et al. (1999)
inter alia Slope of the age function (gt rate of
nativelikeness) depends on L2 input / interaction
/ use / training education task etc.
18Nativelikeness in late L2A
- 8. Incidence of nativelikeness affected by
linguistic factors - Flege et al. (1999) Slope of age function (gt
rate of nativelikeness) steeper for - - knowledge of ungrammaticality than
grammaticality -
- - knowledge of lexical / morphosyntactic
irregulars than regulars -
19Flege et al. (1999)Korean L1 / English L2 (n
240) subset of Johnson Newport (1989) items
grammaticality judgment accuracy
20Nativelikeness in late L2A
- 8. (cont) Incidence of nativelikeness affected
by linguistic factors - Geometry of age function varies by L1-L2 pairing
-
21Johnson Newport (1989) (broken lines
squares)vs. Birdsong Molis (2001) (solid lines
circles)JN Chinese/Korean L1 BM Spanish
L1English L2 grammaticality judgment accuracy
22Non-Nativelikeness
- Bilingualism effects ? learning deficits
23Bilingualism effects
- L1 -gt L2
- L2 -gt L1
- Grosjean (1989) A bilingual ? 2 monolinguals in
1 person - Flege (2002) No perfect bilinguals
24Bilingualism effects
- L2 effects in the L1
-
- Pronunciation VOT (Flege Hillenbrand,1984)
- Lexis Collocations (Laufer, 2003)
- Syntax Middle voice (Balcom, 2003)
- Syntactic processing (Cook et al., 2003)
- Lexical decision (Van Hell Dijkstra, 2002)
- gt Grammaticality judgments Ungrammatical L1
sentences that are grammatical in the L2 are
judged more acceptable than sentences that are
ungrammatical in both languages. (Altenberg
1991 Pelc 2001)
25Bilingualism effects
- L2 effects in the L1
-
- The L1 of a bilingual is not the language of a
monolingual, but differences lt? shortcomings in
language learning ability. L2 effects in L1 are
routine and normal in bilingualism (early
late).
26Bilingualism effects
- L1 effects in the L2
-
- The L2 of a bilingual is not the language of a
monolingual, but differences lt? shortcomings in
language learning ability. L1 effects in L2 are
routine and normal in bilingualism (early
late).
27Bilingualism effects
- L1 effects in the L2
-
- GENERAL L1 effect Cognitive repercussions as a
function of entrenchment maintenance of L1 Ex.
Slower processing (segmentation, parsing)
grammatical convergence lexical
interpenetration - GENERAL L1 effect High L1 use gt lower L2
proficiency than high L2 users
28Bilingualism effects
- L1 effects in the L2
-
- SPECIFIC L1 effect Slope of age function (gt
rate of nativelikeness) varies by L1-L2 pairing
signatures of non-nativelikeness (types of
accent) vary by L1 -
- Ex. Spanish L1 / English L2 shorter VOT lags
than monolingual English English L1 / Spanish L2
longer VOT than monolingual Spanish (Flege
Eefting, 1988)
29Bilingualism effects
- Conceptual/theoretical implications
- All non-nativelikeness lt? learning failure
- (Some non-nativelikeness lt nature of
bilingualism) - ? Utility of nativelikeness (resemblance to
monolingual) as a criterion for falsification of
CPH/L2A - X Utility of non-nativelikeness in support of
CPH/L2A
30L2 Dominance nativelikeness
- ? Upper limits of late attainment seen in
L2-dominants - Flege, MacKay Piske (2002). Italian L1/English
L2 bilinguals (1) L1-dominants, (2) balanced
bilinguals, (3) L2 dominants. Detectable accents
among 1 2 3 like native controls
31L2 Dominance nativelikeness
- Upper limits of late attainment seen in
L2-dominants ASYMMETRY - Golato (2002) English L1, late L2 French
English dominants parse French words like French
natives and English words like English natives.
French dominants (L2 dominants) parse both
English and French words like French monolinguals
(open-syllable segmentation routine)
32L2 Dominance nativelikeness
- Conceptual/theoretical implications
- Age effects and L1 effects are confounded. If we
could minimize L1 effects (i.e., study L1
attriters or L2-dominants), what would the
residual age effects be?
33L2 Dominance nativelikeness
- Conceptual/theoretical implications
- ? Dominance a predictor of attainment (a better
predictor than AoA) - H Degree of dominance (continuous construct,
psycholinguistically operationalized see Flege,
Grosjean, Golato, etc.) predicts degree of
monolingual-like experimental performance.
34L2 Dominance nativelikeness
- Conceptual/theoretical implications
- ? AoA function (timing slope of decline) for
those whose L1 is not dominant - H Reduced L1 dominance / reduced L1 use predicts
later offset and shallower slope of decline.
35L2 Dominance nativelikeness
- Conceptual/theoretical implications
- ? Brain-based data for L2-dominants
- H More similarities with native monolinguals
than observed for high L2-proficients.
36L1-likeness in late L2A
- Neurofunctional similarities
- Abutalebi et al. (2005) Green (2005) Stowe
Sabourin (in press) In terms of where (fMRI)
and when (ERP), similarities with L1 in brain
activation among L2 high-proficients, even late
learners. - continuation
37Age/Proficiency/Task L1-likeness
- Imaging studies
- Production tasks word repetition Klein et al.,
1994 cued word generation Chee et al., 1999
sentence generation Kim et al., 1997 cognate-non
cognate naming De Bleser et al., 2003 - gt L2 proficiency, not AoA, predicts similarity in
areas of activation between late L2 and L1 - gt Degree of similarity varies among studies
- gt Exposure differences and degree of proficiency
differences
38Age/Proficiency/Task L1-likeness/early
bilingual-likeness
- Imaging studies
- Comprehension tasks Perani et al. (1996) PET
Dehaene et al. (1997) - gt Differential activation between L1 and L2
low-proficients - Comprehension tasks Perani et al. (1998) PET
- gt Overlapping patterns for high-proficiency late
and early bilinguals
39Age/Proficiency/Task L1-likeness
- Imaging studies
- Comprehension judgment tasks Chee et al. 2001
fMRI - gt Late high proficients had relatively reduced
brain activity in left prefrontal parietal
areas - Comprehension judgment tasks Wartenburger et
al. 2003 fMRI DESIGN It/Ger early biling/hi
prof late biling/hi prof late biling/low prof - gt Activation for grammar judgments related to
AoA, among high proficients (though groups may
have actually differed in proficiency).
Activation for semantic judgments similar for the
high proficiency groups but more activation in
bilateral BA 47 insula for late vs. early high
proficients
40Age/Proficiency/Task L1-likeness
- Imaging studies
- Word-level meaning reference Chee et al.,
(2000) Ding et al., (2003), Xue et al., (2004) - gt Similar activation in L1 and L2 even in
relatively low proficient, latish learners (Xue
et al., 2004) (fusiform gyrus, Brocas, left
parietal)
41Age/Proficiency/Task L1-likeness
- Imaging studies
- Degree of activation More voxels in a given
area or more signal change for same voxels in L2
versus L1, both early and late bilinguals.
Increased neural activity interpreted as more
effortful processing in L2 (Stowe Sabourin, in
press).
42Age/Proficiency/Task L1-likeness
- NB Methodology
- Potential confounds with chronological age.
Speed, accuracy, effort (Park, 2000) and
ambiguity resolution (Kemper Kemptes, 2000)
affected by age. Activation varies by
chronological age Older adults may activate
less, more, or even different neural structures
(Park Gutchess, 2005 219).
43Age/Proficiency/Task L1-likeness
- ERP studies
- General observation Timing components of
high-proficient L2 use are similar to those of L1
use, AoA gt 12, (e.g., Ojima et al., 2005
Proverbio et al., 2002 Stowe Sabourin, in
press.) - ? ELAN for word category violations not observed
for late L2A (Hahne, 2001 Hahne Friederici,
2001 Weber-Fox Neville, 1996) - ? P600 effects found in some studies (Hahne,
2001 Sabourin, 2003) but not in others (Ojima et
al., 2005)
44Age/Proficiency/Task L1-likeness
- ERP studies
- L2-L1 similarities appear early in the course of
L2A. - McLaughlin et al. (2004) P600 effects for
syntactic violations after 4 months of L2
learning. - Osterhout et al. (2004) Word vs. pseudo-word
N400 effect after 14 hours of instruction.
45Age/Proficiency/Task L1-likeness
- ERP / fMRI
- ? Non-native behavioral results nativelike
brain-based results gt amount of L2 learning
understated in behavioral measures - ERP Osterhout colleagues P600 N400
- fMRI Indefrey et al. (2005)
- gt However, Ojima et al. (2005) Late learners
(Japanese L1/English L2) detected S-V agreement
errors in GJT but did not show P600 effect
46II. Interpreting Age-Related Effects
47The age function
- Negative correlation of AoA and attainment
measures - Over AoA span linear function (other models not
much improvement) - Correlations between -.45 and -.77 median -.64
48The age function
- Disaggregation analyses yield mixed results
- JN sig overall sig early, random late
- BM sig overall ceiling early, sig late
49The age function
- Disaggregation analyses yield mixed results
- DeKeyser (2000) sig overall ns early, ns late
50The age function
- Summary of meta-analysis (Birdsong, 2005)
- gt Pooled data age effects persist indefinitely
(not bounded, no period) - gt Late AoA Typically significant
postmaturational declines - gt Early AoA Inconsistent effects, some flat,
some random, some declining
51Age function in L2A critical period function?
- Definitional logic if (maturationally-based)
critical period, then - gt Age effects should be finite (1) if confined
to a period (2) maturation is a discrete
process/phase within aging - gt Age function should look different pre- vs.
post-maturationally (discontinuity synchronized
with end of maturation) gt
52Age function in L2A critical period function?
53Maturationally-Based Critical PeriodGeometric
and Temporal Features
- 1 peak sensitivity 2 beginning of offset
- 3 end of offset 4 baseline
sensitivity - gt 3 coincides with end of maturation
- gt Age effects do not persist past 3
1
2
STRETCHED Z (Johnson Newport, 1989) (Pinker,
1994)
4
3
54STRETCHED Z (Newport, 1991)
55JN89 Early vs. Late AoA
56Age function in L2A critical period function?
- Stretched L illusory
- Stretched 7 elbow as late as 27.5
57Age function in L2A critical period function?
- X Conceptual Age effects end at end of
maturation (L geometry) age effects begin at
end of maturation (7 geometry) - X Typical linear function over span of AoA
- X Persistent age effects
- v Evidence of 7 geometry / window of
opportunity age non-effect - X But with 7 geometry the window of opportunity
extends past end of maturation - X Not much evidence of L geometry
58Age function in L2A critical period function?
- Minimal necessary feature of gradient
Discontinuity at some point in the overall age
function. (Stevens, Bialystok colleagues,
Flege, inter alia)
59Discontinuity via disaggregation
60Single linear model vs. elbow modelRe-analysis
of results of Flege et al. (1995) (Jan-Pieter de
Ruiter)
61III. The aging brain and L2A
62The aging brain
- Neurocognitive organizational analytic levels,
relevant to language use/learning - Functional/processing lexical encoding/retrieval
processing speed/depth concatenation/coordinatio
n of grammatical units in real time, etc. - Functional/learning Hebbian learning
explicit/explicit learning declarative/procedural
memory etc. - Brain structure hippocampus, striatum, etc.
- Cellular neurotransmission, regional volumes,
etc.
63Cognitive aging
- Mechanisms of cognitive aging (Park, 2000
Rogers, 2000) - gt Decreases in processing speed
- gt Deficits in working memory
- (Decreases in suppression, i.e, focus on relevant
material--possibly tied to working memory)
64Cognitive performance, by decade (Park, 2000)
65Cognitive aging
- General patterns (Bäckman, Craik, Park,
Salthouse, inter alia) - gt Tasks involving working memory or episodic
memory performance declines start in young
adulthood - gt Associative memory incremental learning
declines start in young adulthood - gt Tasks involving priming, recent memory,
procedural memory, semantic memory age effects,
if observed, are mild - gt Implicit memory in tasks involving lexical
recall milder effects than for explicit memory - All related to L2 use / L2 learning
66Cognitive aging
- General patterns, cont
- gt Fluid intelligence involving speeded solutions
to novel problems (e.g. Digit Symbol) Steady
decline over adulthood - gt Crystallized intelligence Declines begin in
late adulthood performance increase between
early and middle adulthood
67L2A and cognitive aging
- NB Language processing
-
- Vocabulary knowledge (in high proficient L2 and
L1) is crystallized intelligence serving language
comprehension gt low effort, little error less
so in low proficient L2 - Assumption gt conclusion In the typical case
(low-L2 proficient / L1 dominants) the proportion
of fluid intelligence involved in L2 use is
greater than in L1 use gt impact of age in L2 use
is greater than in L1 use
68L2A and cognitive aging
- NB Vocabulary knowledge vs. retrieval of
phonological form in real time
69L2A and cognitive aging
- NB Where speed efficiency demands are made
- (A) declines begin in early adulthood
- (B) declines are typically linear and all are
continuous Bäckman Farde 2005 - (?) Patterns consistent with L2 processing
declines - The early onset and gradual nature of the
age-related cognitive decline could inform
attempts to determine its biological origins.
Specifically, whatever the proposed origin may
be, it would strengthen the case if the causative
factor(s) would show a similar onset and
trajectory as the behavioral data (68)
70Age and brain volume
- General patterns (Raz, 2005) in vivo MRI
- gt Brain volume decreases with age, with degree
and timing of declines varying by structure - gt Typically linear, always continuous no
leveling off
71Age, brain volume L2A
- General patterns, cont
- gt Gray matter volume linear decline beginning in
childhood - gt White matter volume linear increase up to
early 20s plateau into 60s linear decline
into old age - gt Degree of decline (slope) varies with
cardio-vascular health other biographical /
lifestyle factors
72Age Effect Size by Region of Interest
(ROI)cross-sectional studies 20-80 yoa
magnitude of effect as age-volume rmean, median
interquartile range of r (Raz, 2005)
73Age Effect Size by ROI (Raz, 2005)
- Magnitude of effects (approx. coefficients)
cross-sectional observation - Prefrontal cortex r -.54
- Putamen r -.46
- Caudate r -.42
- Hippocampus r -.36
- Temporal cortex r -.32
74Age Effect Size by Region of Interest
(ROI)longitudinal studies volume decline per
year(Raz, 2005)
75Age Effect Size by ROI(Raz, 2005)
- Median annual declines (approx.)
- Entorhinal cortex 1.4
- Hippocampus 1.2
- Caudate nucleus 1.1
- Frontal lobe 1
76Striatum volume decline per year(Raz et al.,
2003)
- Median annual decline (approx.)
- Healthy adults (n 53) 20-77 yoa
- Caudate nucleus .83
- Putamen .73
- Globus palidus .51
- Shrinkage
- gt begins in young adulthood
- gt linear declines (same rate of decline for
younger and older subjects) - gt parallels dopamine declines in these areas
77Age, brain volume L2A
- NB Correlative w/ L2 behavioral data,
volumetric declines not bounded - Also Timing geometry by ROI
- Caudate, cerebellum cortical structures linear
decline starting in adolescence, throughout
lifespan - Entorhinal cortex hippocampus greatest annual
shrinkage, but not linear decline controversy
w/r/t timing of onset (probably midlife)
78Age, brain volume L2A
- NB Timing cognitive aging
- In neocortical areas, timing of volume declines
// timetable of associated cognitive task
declines however, not well synchronized with MT
declines - episodic/associative memory declines
(Reuter-Lorenz, 2000) - gt Neural resources (for which volume is a proxy)
in some brain regions are better predictors of
performance than in others
79Dopamine systems language
- Generalities
- Nigrostriatal dopamine (DA) role in efficient
motoric function - D1 D2 receptors diffusely distributed
throughout the neocortex, esp. dense in caudate
putamen areas of striatum. - Bäckman Farde (2005) DA role in higher-order
cognitive functions, some of which are implicated
in language learning processing switching
between attentional targets, motoric sequencing
(verbal fluency), working memory modulation.
NB Role of each in bilingualism
80Dopamine systems language
- DA in basal ganglia functions
- BG in syntax, lexicon phonology production
(Crosson et al., 2003) and processing (Friederici
Kotz, in press Moro et al., 2001 Newman et
al., 2001), processing of rules vs. words
(Teichmann et al., in press)
81Dopamine systems L2A
- DA in basal ganglia functions
- BG in L2A motivation reinforcement (Schumann
1997, 2001, 2004) contribution to
proceduralization (Lee, 2004), defossilization of
automatized procedures (Lee, 2004) gt minimizing
L1 influence / intrusion
82Age dopamine systems
- DA decrements with age when where
- Li et al., (2001) D2 receptor declines begin in
early 20s and continue over lifespan. D2
declines in BG hippocampus, amygdala, FC, ACC.
- Kaasinen et al., (2000) Extrastriatal D2
binding losses TC, PC, OC, hippocampus, thalamus
83Age dopamine systems
- DA decrements performance
- Volkow et al. (1998) PET study (adults 24-86
yoa) of striatal D2 binding potential behavioral
measures on motoric, perceptual executive
speed D2 receptor decline with age in caudate (r
-.62) putamen (r -.7) similar correlations
for performance measures - Prull et al., (1999) Declining nigrostriatal DA
system gt impoverished input to frontal lobes gt
reduced executive capacity of working memory
also attention inhibitory processes (Braver et
al, 2001) - Bäckman Farde (2005) With DA loss neural noise
increases gt less distinctive neural
representations gt declines in working memory
executive function
84Age, dopamine systems L2A
- DA role in a variety of cognitive functions
underlying L2 processing and learning.
Age-related DA declines candidate mechanism
underlying age effects in L2 learning
processing - Age and other biochemical culprits (later life)
- gt Age-related declines in other
neuromodulators/receptors, esp. ACh - gt Age/Stress gt increases in cortisol levels gt
hippocampal atrophy (Lupien et al., 1997, 2001) - gt Aging and estrogen / testosterone mediating
role in verbal production, memory, processing
(Resnick Maki, 2001 Kimura, 1995)
85IV. The Nature of Age Effects
86Causal mechanisms mediating factors
- MacWhinney (in press) 10 concrete proposals
relating AoA to ultimate attainment in L2 - Singleton (in press) 14 versions of CPH/L2A
- Birdsong (1999) 6 major variants of CPH/L2A,
plus experiential endogenous factors in
attainment
87Causal mechanisms mediating factors
- Classes of explanation
- Neurobiological
- Neurocognitive
- Cognitive-developmental
- Affective
- Experiential
88Causal mechanisms mediating factors
- Most factors/mechanisms are consistent with
research - All could be at work in some fashion in L2A, some
accounting for more variance than others in the
aggregate - also individual differences
- Multidimensional, multifactorial, idiosyncratic
nature of age effects
89Summary
- Behavioral evidence
- 1. Unbounded L2 attainment declines ? period
- 2. Bounded age non-effects up to AoA 27.5
- 3. Timing of declines lt? maturational effects
- 4. Slope of function lt external factors
- 5. Non-(monolingual)nativelikeness lt
bilingualism effects - 6. Decline in cognitive processes underlying L2A
use synchronized with behavioral declines - 7. Despite 1,2,5,6 non-trivial incidence of
nativelikeness among late L2 learners
90Summary
- Brain-based evidence
- 8. L1-like where when of brain activation
usually better predicted by L2 proficiency than
by AoA - 9. Linear, unbounded volumetric declines in ROI
comparable to gradients for L2 use/learning - 10. Volume declines synchronized with cognitive
function declines for some but not all ROI - 11. Dopamine system declines synchronized with
decrements in cognitive function underlying L2
use/learning
91Summary
- Further study
- 12. Non-nativelike processing vs. nativelike
attainment in late L2A - 13. L2 attainment processing among L2-dominants
gt subtraction of general specific L1 effects - 14. Linkage of cognitive aging to neural aging
linkage of both to L2 declines over age - 15. Factors/mechanisms of age-related declines
92 93(No Transcript)
94Causal mechanisms mediating factors
- Some factors/mechanisms not consistent with
research - Maturational accounts
- Implicit knowledge not acquirable past puberty
- Less is More (mixed support)
- Biological accounts predicting zero
nativelikeness
95Dekeyser figure
96(No Transcript)
97(No Transcript)
98(No Transcript)
99Specific L1 effectsEx. Van Boxtel (2005)
- L1 French LATE
- L1 German LEARNERS OF
- L1 Turkish DUTCH L2
-
- Sentence preference tasks and imitation tasks
dummy subjects in Dutch -
- - L1 French L1 German scores more often in
native range than L1 Turkish scores -
- - more French German L1ers at near-nativelike
levels
100Age function in L2A critical period function?
- Finite effects elbow at puberty gt
-
101Young vs OldImplicit vs Explicit Memory(Park,
2000)