GRS LX 700
Language Acquisition
and
Linguistic Theory
Week 14.
Issues relating to bilingualism and
general wrap-up
Code-switching

Code-switching often occurs in
conversations between fluent bilinguals
“mixing up” the two languages.

Sometimes people distinguish between
code-mixing (intra) and code-switching
(inter).

We won’t distinguish here, but we’re mainly
talking about intrasentential mixing.
Spanish-English

No, yo sí brincaba en el trampoline when I was a senior.
‘No, I did jump on the trampoline when I was a senior.’

La consulta era eight dollars.
‘The office visit was eight dollars.’

Well, I keep starting some. Como por un mes todos los días
escribo y ya dejo.
‘Well, I keep starting some. For about a month I write
everything and then I stop.’
But it isn’t random…




*El viejo man
*The old hombre
*The viejo hombre
The old man
El hombre viejo
Certain mixes are not considered to be
possible by fluent bilinguals.
But it isn’t random…

The old man
*The buuRaa man
*The buuRaa aadmii

*She sees lo.

How can we characterize what mixes are
possible vs. impossible?


Vo buuRaa aadmii
Vo old aadmii
Vo old man
Prior efforts



Several proposals have been offered to account
for what are good mixes and what aren’t, but it
appears to be a hard problem. Very famous
attempt by Poplack (1980, 1981):
The equivalence constraint. Codes will tend to
be switched at points where the surface
structure of the languages map onto each other.
The free morpheme constraint. A switch may
occur at any point in the discourse at which it is
possible to make a surface constituent cut and
still retain a free morpheme.
Poplack

Looking at the constraints on code-switching of
this sorts can help us understand the nature of
(at least fluent) bilingual language
representation.

One odd thing about Poplack’s constraints is
that it implies that part of UG is dedicated to
mixing. The Free Morpheme Constraint and
Equivalence Constraint are only constraints on
mixing two grammars. Is UG built specifically for
bilinguals?
Problems for Poplack





Equivalence and Free Morpheme Constraints:
Accounts for *estoy eatiendo, but leaves
unexplained:
The students habian visto la pelicula italien.
*The student had visto la pelicua italien.
*Los estudiantes habian seen the Italian movie.
Motrataroa de nin kirescataroa n Pocajontas
Ref-treat-vsf about this 3s-3os-rescue-vsf in P.
‘It deals with the one who rescues P.’
Problems for Poplack?


*El no wants to go
*He doesn’t quiere ir.

*No nitekititoc
not 1s-work-dur (‘I’m not working’)

Amo estoy trabajando
not be.3s work-dur ‘I’m not working’
Problems for Poplack

*Tú tikoas
tlakemetl
2sg 2s-3Os-buy-fut garment-pl-nsf
(‘You will buy clothes’)

El kikoas
tlakmetl
he 3S-3Os-buy-fut garment-pl-nsf
‘He will buy clothes’
MacSwan 1999

Perhaps the most currently comprehensive
and promising account, building on recent
developments in syntactic theory.

One of the basic premises is that language
parameters are properties of lexical items
(not of a language-wide grammar). E.g.,
verb-movement is due to a property of the
tense morpheme in French, not shared by
the tense morpheme in English.
MacSwan 1999

The broad (“minimalist”) approach to
grammar takes language to consist of two
primary components.

Computational system (builds trees),
language invariant.
Lexicon, language particular. Functional
elements of the lexicon encode the
parameters of variation.

MacSwan 1999

MacSwan’s proposal is that there are no
constraints on code mixing over and above
constraints found on monolingual sentences.


(His only constraint which obliquely refers to code
mixing is the one we turn to next, roughly that within a
word, the language must be coherent.)
We can determine what are possible mixes by
looking at the properties of the (functional
elements) of the lexicons of the two mixed
languages.
MacSwan 1999

The model of code mixing is then just like
monolingual speech—the only difference
being that the words and functional
elements are not always drawn from the
lexicon belonging to a single language.

Where requirements conflict between
languages is where mixing will be
prohibited.
Clitics, bound morphemes

Some lexical items in some languages are
clitics, they depend (usually
phonologically) on neighboring words.
Similar to the concept of bound
morpheme.
John’s book.
I shouldn’t go.

Clitics essentially fuse with their host.


Clitics, bound morphemes

Clitics generally cannot be stressed.
*John’S book
 *I couldN’T go.


Clitics generally form an inseparable unit
with their host.
Shouldn’t I go?
 Should I not go?
 *Should I n’t go?

Spanish no





It turns out that Spanish no appears to be a
clitic (despite spelling conventions).
¿Qué no dijo Juan? ‘What didn’t J say?’
*¿Qué sólo leyó Juan? (‘What did J only
read?’)
*¿Qué meramente leyó Juan?
(‘What did J merely read?’)
*Juan no ha no hecho la tarea.
(‘J hasn’t not done the task.’)
Nahuatl amo

In Nahuatl, amo ‘not’ does not appear to
be a clitic.

Amo nio amo niktati nowelti.
Not 1s-go not 1s-3Os-see my-sister
‘I’m not going to not see my sister.’
Spanish-Nahuatl mixing



*No nitekititoc
not 1s-work-dur (‘I’m not working’)
Amo estoy trabajando
not be.3s work-dur ‘I’m not working’
Now, we can begin to make sense of the
difference in possible mixes at the point of
negation between Spanish and Nahuatl.
MacSwan 1999

MacSwan proposes essentially that it is not possible to
code-mix within a (word-like) phonological unit.
Essentially a restriction on what are “pronouncable”
trees.



Idea: phonology operates as a set of ordered rules which are
ordered differently in different languages—you can’t run both
sets of rules at once, hence the result if you tried would be
unpronounceable.
Since Spanish no fuses with the following verb, it can’t
be followed by a Nahuatl verb.
Since Nahuatl amo does not fuse with the following verb,
it is free to be followed by a Spanish verb.
English-Spanish


This also explains Spanish-English (well,
Spanish-anything)
*El no wants to go

What about English-Spanish?
*He doesn’t quiere ir.

*He doesn’t wants to go.

Agreement

In languages that code agreement between subject and
verb, it also appears that mixing is only possible where
the agreement relationship is not disrupted.

*He doesn’t quiere ir.

English negation: agreement appears on do.
Spanish negation: agreement appears on the verb.


You can’t have extra agreement: one subject, one
agreement. They need to match.
Agreement



*Yo nikoas tlakemetl
I 1s-3Os-buy-fut garment-pl-nsf
(‘I will buy clothes’)
*Tú tikoas tlakemetl
you 2s-3Os-buy-fut garment-pl-nsf
(‘You will buy clothes’)
Él/Ella kikoas
tlakemetl
He/She 3s-3Os-buy-fut garment-pl-nsf
‘He/She will buy clothes’
Agreement



Ni-k-koa-s ‘I will buy’
Ti-k-koa-s ‘You will buy’
Ø-k(i)-koa-s ‘He/she wlll buy’

Also relevant: Spanish marks and agrees with gender but
Nahuatl does not distinguish masculine from feminine.

Spanish pronouns have gender specification. The Nahuatl
verb does not. They can only be compatible (match) if
there is no Nahuatl agreement morpheme.
Spanish-Catalan-Greek



Spanish and Catalan both have two
genders, masculine and feminine.
Greek has three genders, masculine,
feminine, neuter.
Predicts: Mixing subjects and verbs
between the three languages is only
possible between the gender-compatible
languages.
Spanish-Catalan-Greek





Yo vull mengar el dinar (S-C)
Jo queiro comer la cena (C-S)
*Ego vull mengar el dinar (G-C)
*Ego queiro comer la cena (G-S)
…
Mixing and L2A?

Code mixing as discussed so far is generally
a property of the speech of fluent bilinguals
(often native bilinguals) and reflects properties
of universal language knowledge.

We can now return to our old question and
ask: Does the knowledge of second language
learners also have the restrictions on code
mixing? To the extent that this is “part of UG”,
is this aspect of UG active for L2’ers?
Toribio & Rubin

Beginning, intermediate, and advanced learners
of Spanish (English L1), asked to imitate codemixed utterances.




Beginning: Processing errors everywhere.
Intermediate: Repeated everything equally fluently.
Advanced: Repeated good mixes fluently, tripped up or
unknowingly corrected improper mixes.
Looks like the constraints emerge, but
intermediates are probably translating to L1 and
doing any judgments there.
Bhatia & Ritchie (1996)







±Us ne kahaa that he will go there.
Us ne kahaa ki he will go there.
*Us ne kahaa that vo vahãã jaay-egaa.
±He said ki vo vahãã jaay-egaa.
He said that vo vahãã jaay-egaa.
*He said ki he will go there.
Conclusion was that intermediate and advanced learners
do have (access to) the constraints. Beginning learners
showed very little sensitivity to contrasts.
So…

Code switching/mixing is quite systematic,
and moreover quite normal behavior for
fluent bilinguals.

It also gives us hints about how languages
are represented, to what degrees they’re
kept separate.
Anomic aphasia

Loss of ability to retrieve words from
memory, happens with almost any kind of
aphasia (often with other more severe
consequences), early dementia, healthy
tired individuals…

Words are not equally susceptible—
there’s a systematicity to the errors.
Anomic aphasia




Content words (boat) are more susceptible
than function words (the).
Infrequent words are more susceptible
than frequent words.
Proper nouns are more susceptible than
common nouns.
Sometime semantic classes go as a block
(color words, letters, numbers).
Regression

There’s an old idea sometimes referred to as
“the rule of Ribot” (dating back to the 1880s) that
the newest things learned are the most
vulnerable to attrition (Freud adopted this view,
calling it “regression”, a return to earlier stages,
Jakobsen claimed the same thing).

But an aphasic who says Baby cry is not as
happy with it as an infant.
Aphasias
Broca’s
Wernicke’s
conduction
nonfluent
cmpr ok
rep poor
fluent
cmpr poor
rep poor
fluent
cmpr ok
rep poor
anomic
transcortical
sensory
transcortical
motor
nonfluent
cmpr poor
rep poor
nonfluent
cmpr ok
rep ok
fluent
cmpr ok
rep ok
How about multiple
languages?


What about a second language?
Are the same brain areas used for both L1
and L2? Or are they different? Or do they
overlap?
Recovery from aphasia



When a bilingual suffers from an aphasia, several
things can happen during recovery (assuming
recovery)
Parallel recovery
Differential recovery



L1 recovers faster (“Ribot’s law”—old before new)
L2 recovers faster (“Pitres’ law”—frequent first)
Recovery generally implies that the actual
language centers haven’t been destroyed, only
either cut off or inhibited.
Recovery from aphasia


The fact that L1 and L2 can recover independently implies
that they are at least in part differentially represented in
the brain.
Case: Dimitrijevic (1940): Woman grew up in Bulgaria,
Yiddish home language, moved to Belgrade at 34 and
spoke Serbian (and Yiddish) from then on, “forgetting”
Bulgarian. A brain injury at 60, after two months for
recovery, resulted in her only being able to speak
Bulgarian and Yiddish; she could no longer speak Serbian
(though she could understand it), despite it having been
her dominant language for 25 years.
Second language recovery


Almost 1/3 of reported multilingual aphasics do not
recover their L1, but their L2 (L3, …).
Case: Minkowski (1928). Patient’s L1 was Swiss German,
learned standard German in school, moved to France for 6
years, became fluent in French, then moved back to
Switzerland (using SG, though still reading French). 19
years later, had a stroke. After 3 days for 3 weeks spoke
only (increasingly fluent) French, then started recovering
German, but for 6 months was incapable of using SG.
Around Christmas, suddenly SG returned (to the detriment
of French).
Factors involved in L2
recovery?



Minkowski’s idea is that the languages are not really
spatially separated, but that they exert mutual inhibition in
a fairly delicate balance. A lesion will disrupt that balance
and can suppress a language (including L1).
In support, often “lost” languages can be recovered faster
than usually required to “learn from scratch”.
Also, autopsy studies don’t seem to reveal a larger extent
to Broca’s area in polyglots (Sauerwin, spoke 54
languages both at poetry and prose level; normal extent
and development in Broca’s area)
Factors involved in L2
recovery?






Familiarity often is the determining factor.
Conscious vs. unconscious knowledge.
Psychological, emotional factors.
Language spoken to patient in hospital.
Domain-specific (rote) language
Higher inhibition levels between closelyrelated languages.
Recovery of non-communication
languages

Case: Grasset (1884). Patient knew only French
(never studied other languages), but then had a
stroke and after a few days, began speaking
only Latin (single words only, primarily prayerrelated).

Case: Pötzl (1925). Professor who knew several
modern languages as well as classical Greek
and Latin. After a stroke, he was only able to
express himself in the dead languages, which he
only knew through reading.
Paradoxical recovery

Case: Paradis & Goldblum (1989). L1 Gujarati, from
Madagascar (spoke Malagasy), learned French in school.
After brain surgery, tested fine in French but was having
trouble with Gujarati at home—fairly classic Broca’s
aphasia symptoms. Malagasy was fine. Over following
months, Gujarati was recovered, but at the expense of
Malagasy. 2 years later, Gujarati was fine, Malagasy was
impaired. 4 years later, both were fine.

Suggests differential inhibition (rather than localization).
Switching and mixing



Healthy bilinguals speaking to other bilinguals
will often code-mix or code-switch.
Aphasic bilinguals sometimes mix unconsciously
without regard to the normal conversational
triggers of code-mixing (often using multiple
languages in conversation with monolingual
speakers).
Or, they will show fixation on one language,
responding only in one language regardless of
the language in which they are addressed.
Alternating antagonism



More dramatic cases reported where patients switch week by
week or day by day between near-total control and nearabsent control of one language, in complementary distribution
to another.
Case: Bruce (1895): Welsh/English (Welsh, left handed,
demented, docile; English, right handed, restless and
destructive). Alternated sometimes several times per day.
Bruce proposed this was due to differential hemispheric
dominance; later supported by studies of subjects with
severed corpus callosum. Suggested left hemisphere was
home of abstract (instructable) capacities.
Translation



Aphasic deficits in translation capabilities
suggest translation might be a separate system.
Reported cases of loss of ability to translate (but
retaining some abilities in each language).
Other reported cases of loss of ability not to
translate; Case: Perecman (1984): patient would
always spontaneously translate German (L1)
sentences uttered into English (L2) immediate
afterward, yet could not perform translation task
on request.
Translation

Sometimes this can happen even without
comprehension; Case: Veyrac (1931):
patient (English L1, French dominant L2),
could not understand simple instructions in
French, but when instructed in English
would spontaneously translate them to
French and then fail to carry them out.
Paradoxical translation

Case: Paradis et al. (1982). Patient
switched (by day) between producing
Arabic and producing French. When
producing only Arabic, she could only
translate from Arabic into French; when
producing only French, she could only
translate from French into Arabic.
Bilingual representation

A number of dissociated phenomena in bilingual
aphasia studies.





Sometimes only one language returns, not always L1
production and comprehension and translation seem to
be separable, and even by language.
Monolingual aphasia studies seem to correlate lesion
localization with function.
Not much evidence for localization differences
between multiple languages per se.
Some evidence for localization differences between
types of learning? (written, conscious vs. unconscious,
implicit vs. explicit memory?)
Bilingual representation



Given the postmortem studies showing no real
morphological differences between monolinguals and
polyglots, the most consistent picture seems to be one of
shared neural architecture with inhibition between
languages.
Choice of language A inhibits access to grammar,
vocabulary of language B during production.
Comprehension is often spared even in the face of
production inability, suggesting that the same kind of
inhibition does not hold of comprehension.
Bilingual representation

Many of the aphasic symptoms in
production can be described in terms of
changing inhibitions; the lesion disrupts the
balance of inhibition and excitation between
neural structures, leading to:
loss of inhibition (pathological mixing)
 heightened invariant inhibition (fixation)
 shifting inhibition (alternating antagonism)
 psychological inhibition (repression)

Subsystems

There also seem to be several subsystems
which can be individually impaired.






Naming, concepts
Fluency of production
Ability to retain and repeat
Translation from L1 to L2
Translation from L2 to L1
Some of these seem to correlate with
localization differences.
More modern methods and
results

Recording electrical activity in the brain can also
help us see which parts are used in language
tasks




Electroencephalogram (EEG)
Event-related potentials (ERP).
Magnetoencephalogram (MEG)
Functional brain imaging



Computer axial tomography (CT) (X-rays)
Positron emission tomography (PET)
Functional magnetic resonance imaging (fMRI)
MEG
ex. Pylkkänen, Stringfellow, Kelepir, & Marantz (2000)
stimulus
BELL
M350: The first MEG component sensitive to
manipulations of stimulus properties affecting lexical
activation. Working hypothesis: this component
reflects automatic spreading activation of the lexicon –
at signal maximum all the competitors are activated.
M180: A visual response
unaffected by stimulus
properties such as frequency
(Hackl et al, 2000), repetition
(Sekiguchi et al, 2000, Pylkkänen
et al 2000) and phonotactic
probability/density. Clearly
posterior dipolar pattern.
M250: A component between
the M180 and M350. Also
insensitive to variations in
stimulus properties that affect
lexical access. Clearly distinct
from the M350 as these two
responses have opposite
polarities. Processing of
orthographic forms?
RT
Postlexical processes
including the word/nonword
decision of the lexical
decision task.
More modern methods and
results



Wada test. Sodium amytal causing temporary
neural paralysis can simulate a possible aphasia
(in order to avoid it during neurosurgery).
Electrical stimulation. Similar but shorter term,
more localized.
Results are mainly in line with other knowledge,
but the problem with these tests is that a)
electrical stimulation is hard to repeat
(imprecise), b) both methods can only be used
on people waiting for neurosurgery who may
have abnormal brains.
Ojemann & Whitaker 1978
•Dutch inhibited
•English inhibited
•Both inhibited
•Neither inhibited
•For what
it’s worth…
Differences: bilingual and
monolingual representations



Best guess at this point is that there is overlap—
the several languages make partial use of
physiologically distinct areas of the brain, but
also share a lot in common.
Some evidence that second language has a
right-hemisphere component, more diffuse than
first language, although directly contradictory
findings have also been reported.
The state of things is actually a little bit
disappointing—but it turns out to be hard work..!
End of the semester
Wrap up
&
Take-Home Points
Some major views on
L1A/syntax




Radford/Guilfoyle/Noonan: kids lack
functional elements initially, have only
lexical elements.
Wexler: kids have access to all the same
grammatical elements that adults do.
Rizzi: kids have “truncated trees”
Vainikka: kids “grow trees”
L1A: Case errors

Kids will sometimes make case errors with
the subject (until around 2).
Me got bean.
 In English, accusative (me) is the “default.”
 Very often taken to indicate a subject not in
SpecIP (a.k.a. SpecAgrSP). No IP? (Radford)
Sometimes IP and above (Rizzi, Vainikka)?
No AgrSP? (Wexler)

L1A: Null subjects

Kids will also often drop out subjects, even in
languages where null subjects are not allowed.




Hyams (1986): Mis-set parameter; they’re speaking
Italian initially.
Kids who are learning null subject languages drop
more subjects than kids who are learning non-null
subject languages.
Bloom: Long sentences are harder, drop what you
can. The beginning of a sentence is more susceptible.
Wexler/Hyams: Kids drop more subjects with
nonfinite verbs. PRO. Sometimes topic drop with finite
verbs, where “topic” isn’t yet grasped.
L1A: Optional Infinitives


In many languages, kids will allow
nonfinite verbs in root clauses sometimes,
early on (up to a little after 2).
NS/OI? Wexler (1998) suggests that
there’s a strong correlation between lack
of OI’s in 2-year-old speech and being a
null subject language.

True? Or are OI’s just extra-rare in null
subject languages (correlation with more
elaborate inflection?).
L1A: Finite vs. nonfinite



During Optional Infinitive stage, kids with
OI’s treat finite verbs like finite verbs and
nonfinite verbs like nonfinite verbs.
German (Poeppel & Wexler): V2 for finite
verbs, final V for nonfinite verbs.
French (Pierce): Verb before pas for finite
verbs, verb after pas for nonfinite verbs.
Some stories about OIs




Rizzi: until maturation of Root=CP, trees
truncated sometimes below tense.
Wexler/Schütze: Syntax intact, but something
prohibits the same (subject) DP from licensing
both TP (finite tense) and AgrP (Nom case).
Radford: Kids don’t use functional categories at
this point (yet, leaves the “finite verbs act finite”
data unexplained).
Legendre et al: Kids minimize the number of
functional projections, basically same outcome
as Schütze & Wexler.
L1A: Principles B and P



Even older kids seem to allow coreference in apparent violation of Principle
B: Mary saw her.
Chien & Wexler, then Thornton & Wexler,
show that when quantifier binding is
available (and thus requires coindexation),
Principle B is respected.
Principle P is slow in coming (matures?),
which says coreference --> coindexation.
L1A: A-chains, passives



Kids are also purportedly slow to master
passives and unaccusatives.
Borer & Wexler (1987): This is maturation of the
ability to represent “A-chains”—more specifically,
the ability to move an object-type thing into a
subject-type position (non-local assignment of qroles).
Babyonyshev et al. (1998) show kids have
trouble with the genitive of negation.
L1A: A-chains etc.

Some possible reasons for skepticism on
this:

Snyder, Hyams, Crisma (1994): French kids
get auxiliary selection right with reflexive
clitics:

Le chienj si’est [ ti mordu tj ].
VP-internal subjects
 Korean negation misplacement seems to
differentiate unergative/transitive from
unaccusatives.

L1A: Negation outside of IP

Kids for a while seem to have trouble with
negation outside the IP, and repair their
utterances so that it remains inside
(usually in an adult-ungrammatical way).
What kind of bread do you don’t like?
 Where he couldn’t eat the raisins?

L1A: Syntax


In general, the errors kids are making seem to
be very systematic.
They seem to know many aspects of the
grammatical system, allowing us to pinpoint (if
we look closely enough and ask the right
questions) what parts don’t seem to be working.




A-chains (or “dethematization of an external arg.”).
Using a [D] feature twice to check functional features.
Allowing negation in C.
Requiring coreference to imply coindexation.
L2A: What can we say?


Certain things are required to explain L1A.
Kids don’t get negative evidence



or if they do, it is inconsistent, it is noisy, and moreover
sometimes when we try to give them negative evidence, they
ignore it.
The kids must be able to learn a system that assign * to
some sentences, based only on positive evidence.
Conclusion: Universal Grammar constrains the kinds of
languages there can be, those languages cannot
generate certain kinds of sentences (hence: *).
L2A: What can we say?





L1A: Languages differ from one another.
Something needs to be learned from the environment.
Yet much of the grammatical system seems common
across languages.
Languages can be thought of as varying not in the system
(the principles) but in the parameters.
The kids, who learn their native language so fast, must
have some help setting the parameters. A Language
Acquisition Device (LAD) designed to choose among the
options made available by UG.
L2A: What can we say?




L2A is generally much harder, more
conscious, slower, less successful.
What’s different about L2A? Did UG
disappear? Did the LAD disappear?
Question: What is the state of the L2’ers
knowledge about the L2?
Does this conform to what UG would
allow?
L2A: UG-accessibility


In general, it seems that the evidence
points to the interlanguages being
allowable human languages. This could
either be influence from UG
(constraining possible languages) or
because the IL is a variation on L1.
Can we tell? Look at parameter
settings: Does IL represent a different
option from L1?
L2A: Transfer

If the IL is UG-constrained, what is the initial
starting assumption?
Is it some kind of general default setting for all the
parameters (likely to be a “subset” grammar from
which all other grammars can be learned via
position evidence alone)?
 Is it just carrying over the parameter settings from
L1?
 Some combination of these?

L2A: Tricks



In order to look properly at parameters, we need
to know what they are. And what a “default”
setting might be. This turns out to be hard.
Pro-drop parameter. Default: Drop subjects?
Subset learnable? Correlated with anything else?
Binding Theory Governing Category? Default?
Language-wide? Strictly predictable from
morphology?
L2A: Interlanguage =
L1+prescriptive rules?




Is the IL just L1 plus some prescriptive rules
(LLK)? (Fundamental Difference)
Or does the IL actually show resetting of
parameters?
Resetting should entail: cluster of properties
comes with new value (again requires that we
know what the parameters, values, clusters are)
If we can find a non-L1, non-L2, but UGavailable option in the IL, that also suggests
parameter setting.
Pro-UG


MacLaughlin (1998) and Japanese to
English via Russian anaphors.
Kanno (1996) and JSL learners seeming
to know how to drop case markers without
instruction.
UG?


White (1991), ESL kids coming from
French don’t seem to learn that the verb
doesn’t raise (at least over adverbs).
Hawkins et al. (1993), FSL people seem to
be “faking” French—early stage treating
negation as part of the verb, start to allow
SVAO in addition to SAVO (recruiting HNP
shift).
L2A: Is there a difference
between kids and adults?






L2A is harder as you get older.
L1A is quite possibility bounded in time.
Evidence for CPs seem to point to different CPs for
different subsystems…
CPs exist in vision, maybe we can find a brain
correlate?
Yet some people may manage to overcome this and
become indistinguishable from a native speaker.
Some plasticity remains?
What disappears/deteriorates? UG? LAD?
L2A: Negative evidence
useful?



L1A doesn’t use negative evidence.
If there is parameter transfer into IL from L1,
logical subset relations might require
negative evidence to reach correct
parameter setting.
Providing people with negative evidence
seems to help—but only in the short term
(without prolonged practicing), it may not
yield any permanent “parameter resetting.”
L2A: Markedness?



Are “unmarked” things easier/quicker to
learn than “marked” things? Does teaching
the “marked” things give you the
“unmarked” things for free?
What are the marked and unmarked
things?
Why do we see generalization beyond the
marked (e.g., in Doughty’s NPAH
experiment)
OI’s in adults? No, L2A≠L1A



Almost no finite (inflected) verb forms in non-finite contexts.
When verbs are marked with inflection, they systematically
(overwhelmingly) appear before negation (i.e., they move).
Many of nonfinite forms used in finite contexts (used finitely,
moved).
—Prévost & White
A(F)
Z(F)
A(G)
Z(G)
Oblig. Fin
+Fin
-Fin
767
243
755
224
389
45
434
85
Oblig. Nonfin
-Fin
+Fin
278
17
156
2
76
7
98
6
Bottom line:

Especially with respect to L2A, there are a
lot of things left to discover because
careful and theoretically informed
experiments still need to be done.

Many of the experiments that are in the
literature rely on misleading simplistic notions
(a monolithic UG subsuming the LAD, a single
once-and-for-all CPH, a one-stage-at-a time
view of acquisition, a subset relation for
adverb placement or binding domain
definitions)…



















