Outline of my talk
Critical period hypothesis and Fundamental
Difference Hypothesis in SLA
How critical is the critical period in second ERPs in L1 + L2
language acquisition? Evidence from event-
Recent ERP studies from our lab ((an)d others)
rellatedd brraiin poottenttialls (ERRPPss)) AoA, L2 proficiency and L1 transfer
Karsten Steinhauer Phonological processing: ‘appy’ or ‘happy’ ?
McGill University Implicit and explicit L2 acquisition
Morpho-syntax at very low L2 proficiency
Psych-341 12 March 2013
Age of acquisition (AoA) Effects in SLA
(Second Language Acquisition)
AoA-dependent differences between L1 and (late) L2
Critical Period Hypothesi(Penfield, 1957; Lennebeand 1964)
Fundamental Difference Hypothesis (FDH;Bley-Vroman, 1989):2003 L1 (early acquired first language)
“Due to loss of brain plasticity in childhood/puberty, Fast
Automatic,language-specific (modules, UG,LAD)
L1 and late L2 are learned and processed differently” Implicit (!), procedural[ized]
2 types of claims (controversial): Late L2 post CP second lagug)age)
1. Limited L2 attainment (fossilization) (Slow)
(Birdsong, 1999; Hyltenstam & Abrahamsson, 2003; 2008, 2009) Less automatic, domain-general cognition
2. Neurobiological dimension of Critical period hypothesis: Different Explicit, declarative(even for simple structures)
neurocognitive processes underlying L1 and L2
Age of acquisition (AoA) is critical t(around puberty: ~12-16 y)
Especially in phonology and morpho-syntax (less so in conceptual semantics)
(JoLong, 2003; Neville, 2007 etc.), 1999; Weber-Fox et al., 1996; Doughty &
(Bley-Vroman, 1989; DeKeyser, 2000, 2010; M. Paradis, 2002; Pinker, Ullman, 2001; Hyltenstam &
Support for the ‘critical period’ in SLA Counter-evidence
Birdsong and Molis (2001): Replication of J&N89 with Spanish
Johnson & Newport (1989):Chinese/Korean learners of English. learners of English.
Age of arrival (~ age of L2 exposure) predicts final L2 attainment only up AoA > 17 years does predict final L2 attainment.
to 17 years of age. Later exposure is not correlated with final L2
attainment (‘end state’). Native-like L2 attainment is not rare.
Late learners never reached ‘native-like’ performance levels. Relationship L1/L2 (transfer effects) may be crucial
White & Genesee (1996) found native-like competenceand
performance in late Francophone learners of English even for
stuccuresshatdd notalow forranssffeeubacenccyy)).
1 Alternative accounts for L2 age effects Current theories: L2 = L1 or L2 ≠ L1?
Three possibilities for the relationship between brain
Neurocognitive ‘entrenchment’ in L1
mechanisms for early L1 and late L2ll under di:cussion)
Socialization in L1 (+ L1 culture)
A. L2 ≠ L1 (“fundamental difference hypothesis” etc)
Motivation to acquire an L2 (what is ‘good enough’) completely compatible with critical period hypothesis
B. L2 = L1 (same mechanisms; but difficult to explain differences
HoourrsoffL2 exxpossuree(asolue nd elate o L1ÆÆ in attainment; Æ entrenchment, motivation, other factors)
Type of L2 exposure (e.g., class room vs. immersion) C. initially L2 ≠ L1, then L2ÆL1 … until L2 = L1 (exception)
(with increasing L2 proficiency, the L2 processing
changes and converges on L1 mechanisms
(Birdsong, 1999; Marchman, 1993; Hakuta/Bialystok, 2000; Morgan- ! Importantly: Brain research can test these 3 hypotheses:
Short et al, in press)
We can directly compare L1 and L2 brain mechanisms!
Are L1/L2 differences inevitable? What does Neuroscience add ?
Do highly proficient late L2 learners rely on
the same (automatic/implicit) neuro-cognitive
mechanisms as native speakers?
SLA researchers reported difficulties
denttiiffyiing qualliitdiifffferences iin
processing (e.g., implicit vs. explicit)
(e.g., Doughty, 2004) and suggested
Æ fMRI and EEG/ERPs
What does Neuroscience add ? What does Neuroscience add ?
WHERE in the brain WHERE in the brain WHEN in the brain
Brain surgery Brain surgery Disorders (even coma patients!)
Brain scans / (f)MRI Brain scans / (f)MRI EEG (electro-encephalography)
Pretty pictures, but slow Pretty pictures, but slow Excellent time resolution (ms)
1 pic / ~ 6 seconds 1 pic / ~ 6 seconds 500 data points / 1 second
2 What does Neuroscience add ? Research Questions + Approach:
Do late second language (L2) learners process
Speech/language is fast and dynamic !
their L2 differently than native-speakers?
~ 3 words/second (even faster in reading)
Sound recognition, word recognition, meaning, grammar,
integration Do L1 background and type of L2 exposure
Processing dynamics in L2 learners may be
Approach: various experimental designs using
event-related brain potentials (ERPs)
Æ EEG (electro-encephalography)
is a good technique to test this
(focus of this talk) Main focus on (morpho-)syntax
EEG / ERPs ERPs – a measure of the brain’s electrical
Electroencephalography / Event-Related Potentials
activity time-locked to stimulus onset
• measure the brain’s neural
activity time-locked to
A happy participant • continuous online measure
(real-time without delays)
• excellent time resolution for a single target word
(1ms)[but poor spatial resolution]
• no task required; recordable at
any age (newborns, coma
• Distinct ERP Negative
components for specific polarity is
cognitive processes plotted
EEG Æ Event-related brain potentials (ERPs)
Violation paradigms and ERP differences
Conceptual Semantics: EEG
John ate broccoli at dinner.
He spread the warm bread with butter. • N400 electrodes
He spread the warm bread with *socks. (sensors)
The ce-cream waaseaten. • Left-anterior -5μVV N400
The ice-cream was in the *eaten. negativity
The children play in the garden.
• P600 EEG amplifier
The children *plays in the garden.
3 EEG Æ Event-related brain potentials (ERPs) EEG Æ Event-related brain potentials (ERPs)
John ate broccoli at dinner. John ate broccoli at dinner.
John ate democracyat dinner. John ate democracyat dinner.
- VVμ - VVμ
N400 N400 effect
EEG amplifier EEG amplifier amplit4u0d0esef(fNe4c0t0
+ 5μV + 5μV integrating word
300 – 500 ms
For example … (slower than normal) +
4 was too
drink. Good or bad?
5 lexical/semantic processing difficulties: N400
How about late L2 learners?
Graded amplitudes of
ERP components: the
more difficult, the
larger the N400
Kutas, Lindamood & Hillyard, 1984
ERPs in L1 and L2
(Weber-Fox & Neville, 1996; Hahne & Friederici, 2001)
How about ERPs for morpho-syntax?
Lexical-semantic processing in late L2 learners seems to be
very similar to that of native speakers:
- slightly delayed, but involving the same brain mechanisms
L1 profile for grammar (here: word order):
two brain waves
Correct: He hoped to enjoy the meal with friends
Violation: He hoped to meal the enjoy with friends
-5 μV And in late L2 ?
Early left negativity:
Assumed to reflect
highly automatic Eaarrllyy lleefftt
grammar processing + 5μV ngaattiivviittyy
CPH: Should be P600
found only in native
speakers, not in late
6 ERPs in L1 and L2
(Weber-Fox & Neville, 1996; Hahne & Friederici, 2001) State of the art … in 2001
•No ‘automatic’ LAN components in SLA
(Potentially no access to Broca’s area (LAD? UG?)
in late L2 , instead compensatory strategies [pragmatics,
delayed semantics] that elicit other, N400-like ERP components
Same claims even in 2009, 2011 by Helen Neville !!
Confound of AoA and proficiency in most studies
(ELAN (Perani et al., 1998: fMRI evidence !)
+P600 • In addition: baseline and other problems !
No effect or Æ Teasing apart AoA and proficiency: artificial
small P600 language learning !
Artificial Language and ERPs (Friederici, Steinhauer & Pfeifer, 2002) Grammar of ‘BROCANTO’ (Friederici, Steinhauer & Pfeifer, 2002)
Miniature Language ‘BROCANTO’
Small set of grammar rules, small vocabulary (14 words)
Sentences refer to moves of a computer game
Grammar rules are independent of game rules
No baseline/prosody confounds!
Trained group (N=28): Trained in both grammar and
vocabulary Lexical tasks: 89%, SD 14
Syntactic tasks: 93%, SD 4
Untrained control group (N=31): Just vocabulary training Lexical tasks: 86%, SD 5
Syntactic tasks: 58%, SD 5
“Native-like” anterior negativity and P600 in proficient
No effects at all in the control group late L2 learners (even for new rules)
rather than AoA
CPH / FDH
7 L2 profi