Two languages in mind: Bilingualism as a tool to
investigate language, cognition, and the brain
Judith F. Kroll
Department of Psychology
Center for Language Science
Pennsylvania State University
[email protected]
May 1, 2015
Acknowledgments
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
Teresa Bajo
Kinsey Bice
Susan Bobb
Cari Bogulski
Ingrid Christoffels
Dorothee Chwilla
Albert Costa
Annette De Groot
Franziska Dietz
Giuli Dussias
Melinda Fricke
Chip Gerfen
Tamar Gollan
David Green
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
Taomei Guo
Jason Gullifer
Noriko Hoshino
April Jacobs
Niels Janssen
Debra Jared
Sonja Kotz
Wido La Heij
Jared Linck
Fengyang Ma
Pedro Macizo
Mari Cruz Martín Rhonda McClain
Erica Michael
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
! 
Natasha Miller
Maya Misra
Jill Morford
Juliana Peters
Pilar Piñar
Eleonora Rossi
Rosa Sánchez-Casas
Ana Schwartz
Bianca Sumutka
Gretchen Sunderman
Natasha Tokowicz
Janet Van Hell
Zofia Wodniecka
Megan Zirnstein
• 
• 
• 
• 
• 
NIH Grants MH62479 and HD053146; NIH Fellowship F33HD055003
NSF PIRE Grant, OISE-0968369: Bilingualism, mind, and brain
NSF Grants, BCS-0111734, BCS-0418071, BCS-0955090 NSF Dissertation Grants to Sunderman, Schwartz, McClain, Hoshino, Bobb, Bogulski, & Gullifer
Open Project Grant at State Key Laboratory for Cognitive Neuroscience and Learning,
Beijing Normal University, China "  John Simon Guggenheim Memorial Foundation Fellowship, 2013-14
How to do research on bilingualism in Central PA?
Collaborate with others in locations where bilingualism is more prevalent: Urban centers in the US and elsewhere in the world
The Penn State Center for Language Science International Bilingualism Network
Support from NSF PIRE (Partnerships for International Research and
Education): 2010-2015: Bilingualism, mind, and brain: An interdisciplinary program in cognitive psychology, linguistics, and cognitive neuroscience More people in the world are bilingual than monolingual.
But until very recently, most research on language and cognition
examined only monolingual speakers of a single language and
typically speakers of English as the native language.
Bilinguals were considered to be a “special” population.
A word on terminology: Bilinguals and Second Language (L2) Learners
We adopt a broad definition of bilingualism to include all
individuals who use more than one language regularly.
We distinguish bilingual groups with respect to their
proficiency in the L2, their relative language dominance,
the age of acquisition, and the degree to which the context
of language use supports each of the two languages.
We focus on those speakers for whom the native language
and the L1 are the same.
Past research on language processing and its cognitive basis has
typically assumed that monolingual speakers are the model subjects
of study and that the native language alone provides an adequate
basis on which universal principles might be generalized. #  On this view, bilinguals have been considered a special group
of language users, much like brain damaged patients, children
with language disorders, or deaf individuals who use a signed
language to communicate. #  Each of these special groups holds genuine interest for the
field but their performance is not necessarily taken to provide the
primary source of evidence for the purpose of adjudicating the
classic debates about the representation and processing of
language in the mind and brain.
Why?
Learning an L2 past early childhood is a difficult task with mixed outcomes.
Even highly successful late L2 learners speak with an accent and appear to fail to acquire subtle aspects of the L2 grammar. Flege et al. (1995)
Johnson & Newport (1989) For these reasons, the evidence on bilingualism, particularly for late
acquirers of an L2 has been taken by many to suggest that the L2 is
fundamentally different and separate from the native language, with
properties that are enabled by domain-general cognitive processes but
constrained by the inability to access all of the linguistic representations
typically associated with the native language.
On this (traditional) view:
#  Late bilinguals may indeed be special, with a mixed language system that
includes a full native language and a funky L2
#  Bilinguals should be functionally monolingual in the native language
#  The L1 should transfer to the L2 but not much transfer would be expected
from the L2 to the L1
!
On the 125th anniversary of the journal Science, Kennedy and Norman (2005)
identified the biological basis of second language (L2) learning as one
of the top 125 questions to be answered in the next 25 years of research: !
Second Language Acquisition
Bilingualism
1200
1000
Number of
Papers
800
600
400
200
0
1985-89 1990-94 1995-99 2000-04 2005-09
Time Period
Research articles published on Second Language Acquisition and
Bilingualism since 1985 (Web of Science)!
There has been a virtual explosion of research on bilingualism:
Figure 1. Results of search for topic ‘‘bilingualism’’ on ThompsonReuters Web of Science for (a) number of papers published and
(b) number of citations of those papers for years 1993 to 2012.
(From Kroll & Bialystok, 2013, Journal of Cognitive Psychology)
Second Language Acquisition
What have we learned in the recent!
upsurge of research?!
Bilingualism
1200
1000
Number of
Papers
800
600
400
200
0
1985-89 1990-94 1995-99 2000-04 2005-09
Time Period
!
#  Recent neuroscience evidence has called into question the
presence of hard constraints on L2 learning; proficiency in L2
may be more important than age of acquisition (e.g.,
Abutalebi et al., 2005; Steinhauer et al., 2009) and the brain!
may outpace behavior in revealing L2 learning.!
#  But there are consequences: proficient bilinguals are not
monolingual-like in their native language, suggesting that the
native language is open to change and to the influence of the
L2 (e.g., Ameel et al., 2009). Competition across the two languages
may reshape the networks that support each language. !
!
Take the recent evidence on bilingualism to ask the question of why is it so difficult for even smart people to acquire a second language as an adult? And who succeeds in this task?
Different ways to think about constraints on late adult acquisition of the L2
#  Critical period effects: Constraints on plasticity past early childhood
restrict the full acquisition of the L2 grammar and phonology
Constraints on representation (e.g., Shallow Structure Hypothesis,
Clahsen & Felser, 2006)
Constraints on processing (e.g., the Declarative/Procedural Model, Ullman, 2001; demands on working memory resources) #  Cross-language competition: Alternative forms and structures are active and compete
Implications for reading and speaking (e.g., Dijkstra, 2005; Kroll et al., 2006)
Implications for behavioral and neural development (e.g., Hernandez et al., 2005)
Factors that may be involved in successful late L2 learning:
1.  A high level of cognitive resources, particularly working
memory? Executive function skills?
2.  A high level of phonological awareness? Some people seem
to have the other language in their ear.
3. Language learning aptitude? Motivation?
4. Courage: Some people are willing to make utter fools of
themselves and attempt to speak the L2 even when they know few words and almost no grammar!
Each of these factors accounts for something but not enough.
A new hypothesis that is guided by research on proficient
bilingualism and informed by research on learning and
memory: Successful L2 learners may be individuals who are
able to tolerate change to the native language. That change may involve processing costs that initially slow the
native language and make native language performance more
error prone, make learners less sensitive to some features of the
native language, and that open the native language to the
influences of the L2. High levels of cognitive resources and immersion in the L2 may
enhance this process but what is hypothesized to be fundamental
is change to the first language (L1) that functionally allows the
L2 to develop as part of the language system. Three discoveries about bilingualism:
1.  Both languages are always active and competing.
2.  The native language changes in response to learning and using an L2.
3.  The consequences of bilingualism are not limited to language but reflect
a reorganization of brain networks that hold implications for the ways
in which bilinguals negotiate cognitive competition more generally. Kroll, Bobb, & Hoshino (2014). Two languages in mind: Bilingualism as a tool to investigate language, cognition, and the brain. Current Directions in
Psychological Science.
How do the mind and brain accommodate the presence of two
languages?
The bilingual is a mental juggler: Both languages are
active
regardless of the requirement to use one language alone: Dutch-English speaker
“bike”
“fiets”
How does a bilingual select a given language to be used at any moment?
An illustration of language nonselectivity: The phonology of the
language not in use modulates the time for bilinguals to read
words in each language.
Cognates with identical/similar orthography but similar or different phonology:
English
Spanish
Cross-language
phonology
piano
piano
Similar [+p]
base
base
Different [-p]
620
600
+P
-P
Mean 580
Naming
Latency 560
(ms) 540
520
500
Type of Cognate
Schwartz, Kroll, & Diaz (2007): Bilinguals are faster to name cognates in
L2 when the phonology converges from L1 to L2. But the same result for
reading in the dominant L1. The cross-language effects are bidirectional. These cross-language interactions are persistent. We see them
even when bilinguals are processing words in sentence context, even when they are not required to use one of the two languages at all, even when the bilinguals are highly proficient in the L2, and even for language pairings that are highly dissimilar.
Morford, Wilkinson, Villwock, Piñar, & Kroll (2011)
Deaf signers judge the semantic relatedness of two words in English. The ASL translations of the English words have a “phonological” form relation or not. Deaf signers are faster to judge the English when the ASL converges and slower when it conflicts. Monolingual English speakers do not show these effects.
movie
paper
Phonologically related in ASL
lion
baby
Phonologically unrelated in ASL
Three discoveries about bilingualism:
1.  Both languages are always active and competing.
2.  The native language changes in response to learning and using an L2.
3.  The consequences of bilingualism are not limited to language but reflect
a reorganization of brain networks that hold implications for the ways
in which bilinguals negotiate cognitive competition more generally. Kroll, Bobb, & Hoshino (2014). Two languages in mind: Bilingualism as a tool to investigate language, cognition, and the brain. Current Directions in
Psychological Science.
Illustrate the effects of L2 on L1 in a program of research on
bilingual speech planning.
Evidence for inhibition of the L1 to enable speech production in the L2. We see suppression of the L1 in the earliest measures of brain activity when bilinguals prepare to speak words in either language,
in their behavior when they begin to speak, and in late acoustic
measures of produced speech.
Misra, Guo, Bobb, & Kroll (2012)
Use ERPs to examine the earliest time course of cross-language
activation in bilingual speech planning.
The effect of language blocking in picture naming in the L1 and L2.
Relatively proficient Chinese-English bilinguals but dominant in L1
Chinese.
Group 1: Name pictures in L1 then L2 Group 2: Name pictures in L2 then L1
Name in
L1
Name in
L2
Name in
L2
Name in
L1
The pictures were the same for both languages; two blocks per
language: L1, L1, L2, L2 or L2, L2, L1, L1
Blocked Picture Naming: Early indices of inhibition L1
L1 First
L1 Following L2
L2
L2 First
L2 Following L1
Inhibitory pattern for L1 and facilitatory pattern for L2:
If it were a matter of recovering from momentary inhibition
following naming in L2, then later in the L1 naming blocks we
should see this recovery but the pattern persists, suggesting the
presence of global inhibition.
Does behavior also reflect this early inhibitory pattern for the L1?
Moriyasu (2014): examined simple picture
naming for Japanese-English bilinguals
who were highly proficient in English as
the L2 and living in the US but still very
dominant in L1 Japanese
Measure
Japanese (L1)
English (L2)
Self rating
proficiency
(1-7 scale)
6.5 4.5
Category fluency
(in 30 seconds)
48.3
38.6
Moriyasu (2014)
1200
L1
L2
L2
L1
Japanese (L1)
English (L2)
Mean 1100
Picture
Naming
Latency
(ms) 1000
900
L1 then L2
L2 then L1
Order of picture naming
When L1 is named first, we see the expected
pattern of faster naming latencies for L1
than L2.
When L1 is named after L2, they are slower
to speak Japanese than English! A reversal
of their normal language dominance.
But can we see this inhibitory effect at the latest stages of production during articulation? Three picture naming blocks with an interpolated block in the other language:
L1 – L2 – L1
L2 – L1 – L2
Effects of language blocking on articulatory duration: Are there
late inhibitory effects?
Name pictures in three blocks: L1 Chinese- L2 English- L1 Chinese
Name L1
Name L2
Name L1
460
450
440
Before L2
After L2
Mean 430
Duration
420
(ms)
410
400
Mean Duration to Speak L1
Articulatory duration is longer in L1 following picture naming in L2. These data are similar to the conditions that produced extended negativity in the
ERPs and longer RTs in the naming. The effect is present even for identical
tokens that should produce repetition priming, suggesting that there is inhibition
of the L1 following naming in the L2. Abutalebi & Green (2007): Different loci of cognitive control in the bilingual brain: different components of inhibition?
Guo, Liu, Misra, & Kroll (2011): fMRI evidence for global inhibition Chinese-English bilinguals named pictures in three blocks: Name in
Name in
Name in
Chinese (L1) – English (L2) – Mixed
L1
L2
L1 or L2
English (L2) – Chinese (L1) – Mixed
Name in
Name in
Name in
L1 or L2
L2
L1
The comparison between blocked and mixed picture naming performance was defined as local
switching, while the comparison between blocked naming in each language was defined as
global switching. Distinct patterns of neural activation were found for each of these comparisons.
Guo, Liu, Misra, & Kroll (2011): fMRI evidence Distinct patterns of neural activation were found for local inhibition as compared to
global inhibition in bilingual word production: The dorsal anterior cingulate cortex (ACC) and the supplementary motor area (SMA)
appear to play important roles in local inhibition, while the dorsal left frontal gyrus and
parietal cortex appear to be important for global inhibition.
Does inhibition of L1 depend on the context of language use?
Linck, Kroll, & Sunderman (2009): Losing access to
the native language while immersed in a second language
Examined the performance of L2 learners who were all native
English speakers at an intermediate level of studying Spanish
as the L2.
One group was studying abroad in Spain and immersed in the
L2 environment whereas the other group was studying in the
classroom in very monolingual Pennsylvania.
L1 activity was reduced in both comprehension and production
when learners were immersed in the L2 environment.
Linck et al. (2009): Semantic fluency in immersed learners vs.
classroom learners 60
50
40
Number of
Exemplars 30
20
Controls
Immersed
10
0
English
Spanish
Language of Production
The immersed learners had lower semantic fluency in English, the L1.
Taken together with their performance on a translation recognition task
showing that they were insensitive to lexical distractors in the L1, these
data suggest that they were suppressing the L1 while living in the L2
context (and see Baus et al., 2013, for related findings and Levy et al.,
2007, for a similar finding in a retrieval induced forgetting paradigm).
The native language is not the Rock of Gibraltar These
studies of cross-language interaction suggest that not only does the L1
influence the L2, but the L2 has persistent effects on the L1, these influences even for highly proficient bilinguals. Negotiating
has been hypothesized to confer some of the observed consequences to cognition in the realm of executive function.
The effects of bilingualism on the L1 are not just about words. We see related phenomena at every level of language processing, including the grammar and the phonology. L2 comes to affect L1.
Is this story about the effects of cross-language activation from L2 to L1 only about the lexicon? No. It’s also about the grammar.
Many studies of the L2 grammar demonstrate persistent transfer of the L1 in using the L2 but they also show that the L1 becomes sensitive to the influence of the L2.
Dussias (2003): How do the structural commitments of one language influence
the processing of the other language?
Peter fell in love with the daughter of the psychologist who studied in California.
Who studied in California?
Native English speakers: the psychologist
Native Spanish speakers: the daughter
Critical result: Native Spanish speakers immersed in an English dominant
environment begin to parse sentences in Spanish, their native language, like
English, their L2!
Three discoveries about bilingualism:
1.  Both languages are always active and competing.
2.  The native language changes in response to learning and using an L2.
3.  The consequences of bilingualism are not limited to language but reflect
a reorganization of brain networks that hold implications for the ways
in which bilinguals negotiate cognitive competition more generally. Kroll, Bobb, & Hoshino (2014). Two languages in mind: Bilingualism as a tool to investigate language, cognition, and the brain. Current Directions in
Psychological Science.
What is the consequence of parallel activity and competition across the bilingual’s two languages? Juggling may tune brain networks that
enable control and build cognitive reserve.
Life experience as a bilingual changes the mind and the brain
1.  Bilingualism changes the efficiency of the brain networks responsible for
resolving competition and conflict in non-linguistic tasks.
2.  These changes are sometimes observable in behavior but even when they are
not, they may be evident in structural and functional changes in the brain.
3.  The consequences of bilingualism are more evident for older bilinguals than
for young adult bilinguals. Bilingualism provides protection against
cognitive decline.
4.  The regulatory processes that are engaged by bilingualism can also be
trained outside of language experience, suggesting that they are domaingeneral cognitive and neural mechanisms that underlie executive function
more generally.
5.  Some of these control processes can be caught “on the fly” as language
processing is ongoing and others are likely to reflect longer term language
use.
6.  Not all bilingual experience produces the same consequences. To illustrate:
The flanker task: non-linguistic task to assess executive function
What is the neural basis of the bilingual effect in resolving conflict?
The use of two languages may impose processing demands that
then create distinct profiles of bilingual cognition. Different
forms of bilingualism may have the consequence of differentially
tuning the neural networks that support language use (e.g., Green
& Abutalebi, 2013). Some bilinguals code switch frequently and
others not at all. Some languages share similar form and others
to do not. But in all cases, bilinguals have to potentially negotiate
a higher level of competition in their everyday use of language
than monolinguals. The ability to acquire these regulatory mechanisms and to use them to effectively control the L1 may be a crucial component
of successful L2 learning. The new hypothesis is guided by research on proficient
bilingualism and informed by research on learning and
memory: Successful L2 learners may be individuals who are
able to regulate change to the native language. That change may involve processing costs that initially slow the
native language and make native language performance more error
prone, make learners less sensitive to some features of the native
language, and that open the native language to the influences of the
L2. Research on learning and memory provides a framework for
understanding how initial conditions of study that are more
difficult, induce errors, and that require greater elaborative
processing may benefit learning in the long term (e.g., Bjork et al.,
2013; Healy & Bourne, 2013). There may also be individual
differences in the ability to self-regulate initial learning.
To illustrate: Jacobs, Fricke, & Kroll (under review) What facilitates learning initially may hinder language
processing later.
The story about cognates: translations that have
similar word form.
A trip to the Dutch super market:
aardappel
erwten
ui
wortel
selderij
paddestoel
kool
tomaat
spinazie
bieten
paprika
sla
Cognates are easier to
learn than non-cognates:
Creates an illusion that even
native English speakers can understand Dutch!
But learners need to be careful because even when the spelling
is identical, the pronunciation is never identical, and there are also
false friends that look alike but have different meanings and
can guesses can get you into trouble:
Back to the supermarket: cream for your coffee? It’s room in Dutch!
room is a false friend or an interlingual homograph Jacobs, Fricke, & Kroll (under review): Do cognates facilitate
planning and producing speech in the L2?
Compared 3 groups of native English speakers with Spanish
as an L2:
1. Intermediate classroom learners of Spanish
2.  Intermediate learners of Spanish immersed in a domestic
summer program at Middlebury College: no English allowed
3.  Advanced proficiency speakers of Spanish: graduate students
in the Spanish Department
Task: Name words in Spanish: The words were either cognates
or matched control words
Mean Naming Latency (ms)
The planning effect: Facilitation for naming cognates: Everyone shows it!
700
680
660
Cognates
Controls
640
620
600
580
Learners
Learners
Immersed
Advanced
Activation of the native language facilitates word naming.
The production effect: for the learners, facilitation becomes interference
Mean VOT (ms)
38
34
Cognates
Controls
30
26
Voice
Onset Times
(ms)
Cognates are easier
to learn but harder
to produce as
Spanish words for
classroom learners
22
Learners
Learners
Immersed
Advanced
The immersed and advanced groups both exhibit mean VOTs of under 30
ms, the general category boundary for English. The more advanced the learner, the more Spanish-like their speech.
The classroom learners are more English-like, with cognates having a subtle
but significantly greater VOT duration than their phonetic controls. The
activation of the L1 may be an impediment to acquiring the L2 phonology.
They may have to learn to inhibit L1 to acquire the L2 fully.
It is of interest that proficient bilinguals are better language learners
than monolinguals, particularly when they learn new information via
the native language. We consider the lessons that the bilingual
advantage may hold for all learners.
High levels of cognitive resources and immersion in the L2 may
enhance this process but what is hypothesized to be fundamental is
change to the L1 that functionally allows the L2 to develop as part of
the language system. The hypothesis is that late L2 learners must be able to regulate the
activity of the L1. In ongoing research, we are examining adult L2
learners at early stages of learning a new language and before,
during, and after language immersion in the L2. We use converging
behavioral and neurocognitive measures to ask whether there are
individual differences in the native language that predict success in
the L2 for the lexicon, the grammar, and the phonology.
But there are some language students who simply do not learn. For
them, there is no lexicon; no grammar, and absolutely no inhibitory control!
Thank you!