ArticlePDF Available

Language control in bilinguals: The adaptive control hypothesis

Authors:

Abstract and Figures

Speech comprehension and production are governed by control processes. We explore their nature and dynamics in bilingual speakers with a focus on speech production. Prior research indicates that individuals increase cognitive control in order to achieve a desired goal. In the adaptive control hypothesis we propose a stronger hypothesis: Language control processes themselves adapt to the recurrent demands placed on them by the interactional context. Adapting a control process means changing a parameter or parameters about the way it works (its neural capacity or efficiency) or the way it works in concert, or in cascade, with other control processes (e.g., its connectedness). We distinguish eight control processes (goal maintenance, conflict monitoring, interference suppression, salient cue detection, selective response inhibition, task disengagement, task engagement, opportunistic planning). We consider the demands on these processes imposed by three interactional contexts (single language, dual language, and dense code-switching). We predict adaptive changes in the neural regions and circuits associated with specific control processes. A dual-language context, for example, is predicted to lead to the adaptation of a circuit mediating a cascade of control processes that circumvents a control dilemma. Effective test of the adaptive control hypothesis requires behavioural and neuroimaging work that assesses language control in a range of tasks within the same individual.
No caption available
… 
Content may be subject to copyright.
Language control in bilinguals: The adaptive control
hypothesis
David W. Green
1
and Jubin Abutalebi
2,3
1
Cognitive, Perceptual and Brain Sciences, Faculty of Brain Sciences, University College
London, London, UK
2
Faculty of Psychology, Vita-Salute San Raffaele University and San Raffaele Scientific
Institute, Milan, Italy
3
Division of Speech and Hearing Sciences, University of Hong Kong, Hong Kong
Speech comprehension and production are governed by control processes. We explore their nature and
dynamics in bilingual speakers with a focus on speech production. Prior research indicates that
individuals increase cognitive control in order to achieve a desired goal. In the adaptive control
hypothesis we propose a stronger hypothesis: Language control processes themselves adapt to the
recurrent demands placed on them by the interactional context. Adapting a control process means
changing a parameter or parameters about the way it works (its neural capacity or efficiency) or the way
it works in concert, or in cascade, with other control processes (e.g., its connectedness). We distinguish
eight control processes (goal maintenance, conflict monitoring, interference suppression, salient cue
detection, selective response inhibition, task disengagement, task engagement, opportunistic planning).
We consider the demands on these processes imposed by three interactional contexts (single language,
dual language, and dense code-switching). We predict adaptive changes in the neural regions and circuits
associated with specific control processes. A dual-language context, for example, is predicted to lead to
the adaptation of a circuit mediating a cascade of control processes that circumvents a control dilemma.
Effective test of the adaptive control hypothesis requires behavioural and neuroimaging work that
assesses language control in a range of tasks within the same individual.
Keywords: Adaptive control hypothesis; Bilingual; Cognitive control; Language control.
Specific forms of training such as learning to play
the piano (e.g., Bialystok & Depape, 2009) and
patterns of upbringing (e.g., Hedden, Ketay,
Aron, Markus, & Gabrieli, 2008) shape how
individuals perform in nonverbal tasks tapping
processes of cognitive (executive) control. Strik-
ingly, the use of more than one language appears
to be a further important factor shaping indivi-
dual performance on such tasks (e.g., Bialystok,
Craik, Green, & Gollan, 2009, for a review). Why
might this be? There is a two-step argument:
Increased cognitive demands associated with
language control in bilingual speakers lead to
enhanced skills in cognitive control and these
enhanced skills are deployed in performing
nonverbal tasks tapping such control. We focus
here on the first step: the cognitive demands of
language control in bilingual speakers.
We specify the language control processes
involved in three different real-world interac-
tional contexts (single language, dual language,
and dense code-switching). By an interactional
Correspondence should be addressed to David W. Green, Cognitive, Perceptual and Brain Sciences, University College London,
Gower Street, London WC1E 6BT, UK. E-mail: d.w.green@ucl.ac.uk
DG was supported by the Wellcome Trust.
Journal of Cognitive Psychology, 2013
Vol. 25, No. 5, 515530, http://dx.doi.org/10.1080/20445911.2013.796377
#2013 The Author(s). Published by Taylor & Francis.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited. The moral rights of the named author(s) have been asserted.
context we refer to the recurrent pattern of
conversational exchanges within a community of
speakers. Our thesis is that the control processes
themselves adapt to the demands imposed on
them by these different contexts. This thesis
entails theoretical work to identify the nature of
such adaptations. It also implies that experimental
tests take explicit account of the contexts of
language use.
On evolutionary grounds, we consider that the
processes involved recruit processes involved in
the control of action in general (e.g., Stout &
Chaminade, 2012). They are recruited then in the
speech of monolingual speakers too.
This evolutionary assumption poses a puzzle.
Given the vast range of skills that individuals
learn why, in principle, might learning to use a
further language make a difference to the
exercise of cognitive control and so affect the
cognitive control of nonverbal tasks? Each non-
verbal skill would seem to demand the kinds of
control processes relevant to language control:
For instance, individuals must monitor the con-
text, maintain the action goal and resist inter-
ference from other competing actions that may be
triggered by the situational context. Is language
control special in some way?
Conceivably there is a difference in the com-
plexity of what is controlled. Many nonverbal
activities, though complex (e.g., making a cup of
tea), involve stereotyped sequences. Communica-
tive actions involving language are arguably
relatively less stereotyped and so more demand-
ing in their planning and execution. However,
much everyday discourse involves simple linear
syntactic structures that are relatively undemand-
ing in terms of their planning and execution
(Frank, Bod, & Christiansen, 2012).
A further difference may be more crucial. In
the case of nonverbal actions, there can be
alternative ways to achieve a given goal. These
alternatives are equifinal and we may execute
the one that is available faster perhaps because its
planning is simpler given the context. For exam-
ple, we elect to switch off a kettle with our left
rather than the right hand whilst holding a cup in
the other. In the case of speech, alternative
(congruent) utterance plans are not invariably
substitutable. Speakers, whether bilingual or
monolingual, may need to select between differ-
ent ways to conceptualise an event and select
between different ways to express this conceptua-
lisation depending on their addressee. For bilin-
gual speakers, alternative utterance plans in
different languages are not invariably substituta-
ble as the addressee may not know the other
language or there may be social or topic reasons
for not using it. We suggest therefore that use of
language whether in monolingual or bilingual
speakers cannot be equated with the myriad of
equifinal nonverbal actions: It is a special kind
of action in which congruency is an insufficient
criterion for selection. Substantial experimental
evidence indicates that in bilingual speakers both
languages are active even when only one is being
used (see, for reviews, Bialystok et al., 2009; Kroll,
Bobb, & Wodniecka, 2006). On this view, selec-
tion follows activation of alternative possible
candidates for expressing a message. In bilingual
speakers, the demand to select an utterance
despite ‘‘equifinality’’ recurs in a repeated and
sustained fashion. Accordingly, we infer that, in
principle, language use in bilingual speakers
increases the demand on the processes involved
in utterance selection over and above those that
are imposed on monolingual speakers. If control
processes adapt to such demands then this argu-
ment provides a basis for expecting possible
advantages in the cognitive control of nonverbal
tasks though it leaves open the mechanism
involved. More critically, it requires us to specify
the precise processes that might be subject to
adaptation and how the contexts of language use
may shape such adaptation. Our goal is to identify
a set of language control processes that support
conversation in different interactional contexts,
articulate the relative demands of these contexts
on these processes, and spell out the neural bases
of adaptive changes.
Competing representations may extend over
the entire speech pipeline from formulating
the message, selecting and sequencing relevant
lemmas and word forms, to retrieving, and
articulating relevant phonemes and monitoring
self-produced speech with respect to its predicted
acoustic/phonetic form. The targets of language
control then may differ in their linguistic level and
so the precise locus of control effects will vary
(Kroll et al., 2006). We recognise that comprehen-
sion processes in bilingual speakers are relevant
to the adaptive response. They may tune the
system to detect critical features that discriminate
one language from another (Krizman, Marian,
Shook, Shoe, & Kraus, 2012; Kuipers & Thierry,
2010) and adapt processes that control interfer-
ence between competing word meanings (e.g.,
Macizo, Bajo, & Martin, 2010). However, we
focus on speech production because the ability
516 GREEN AND ABUTALEBI
to formulate a relevant message is also vital to
comprehension. In listening to a speaker we
develop a forward model of what they may say
and this allows us to act collaboratively. We spell
out our proposal in terms of the adaptive control
hypothesis in the next part of the paper. In the
third part we review and discuss the hypothesis
before concluding.
THE ADAPTIVE CONTROL
HYPOTHESIS
We envisage that cognitive control processes
select competing representations in working
memory as individuals seek to achieve their
intended goals. The targets of these control
processes differ: they may be verbal or nonverbal
representations. In consequence, disruption of the
neural linkage between the regions involved in
control and the specific target of control can give
rise to dissociations in performance despite a set
of common control processes.
Prior research indicates that deviations from
required performance (e.g., overt errors or delays
in responding) trigger control processes that serve
to bring behaviour more in line with what is
required (e.g., Botvinick, Cohen, & Carter, 2004).
So, for example, individuals respond more slowly
to the direction of a target arrow when it is
flanked by arrows pointing in the opposite direc-
tion. However, this conflict effect is reduced when
an incongruent stimulus is presented on an
immediate subsequent trial. Neuroimaging work
implicates a feedback circuit in which signals from
a midfrontal neural region (i.e., the anterior
cingulate cortex) that detects conflict trigger a
response in a left inferior frontal region that
serves to suppress interference (e.g., Kerns
et al., 2004). We argue for the stronger hypoth-
esis, the adaptive control hypothesis, in which
the processes of control themselves adapt to the
demands placed upon them. For bilingual speak-
ers the interactional context in which they find
themselves drives the adaptive response. Why
might control processes adapt? One reason, we
suggest, is that there is an interactional cost in not
doing so. We look at this cost in a later section.
Conceptually then, and also by way of over-
view, we distinguish the interactional context, the
speech pipeline (that is, the conceptual-affective-
linguistic-sensorimotor representations entrained
in the production and comprehension of speech);
the processes that control these representations in
working memory to ensure the communicative
goal and the process that sets the parameters of
these control processes*the meta-process (see
Figure 1 for a depiction). It is these parameters
that are subject to adaptation and mediate
changes in skill. Neurally, adaptation may be
achieved in different ways: through a change in
structural resources or capacity (e.g., grey matter
density), through a change in regional efficiency
(e.g., through tuning neuronal populations or
changing the responsiveness of neuronal popula-
tions) or through a change in the connectivity of
the network (e.g., white matter connectivity).
Such parameters then capture the persistence or
flexibility of control, the efficiency of transmis-
sion across the network, and the coordination of
different control processes.
In the following three subsections, we describe
the interactional contexts, identify a set of control
processes used in these contexts, and then con-
sider how the demand on these control processes
varies as a function of the interactional context.
Three interactional contexts
We consider three interactional contexts (three
different recurrent patterns of conversational
exchange) as a way to contrast demands on
control processes. We use these contexts, rather
Interactional context Meta-control process
The speech pipeline Control processes
Figure 1. Architecture of the adaptive control hypothesis. Filled arrows depict internal processes of control.
THE ADAPTIVE CONTROL HYPOTHESIS 517
than more specialised contexts (e.g., simultaneous
translation, air-traffic control) because these
contexts reflect the everyday conversational use
of language.
(1) A single-language context in which one
language is used in one environment and
the other in a second distinct environment.
For example, a nonnative language may be
spoken exclusively in the work environ-
ment with colleagues, whereas the native
language is used exclusively with family
members at home. In such a context there
is no frequent switching between languages;
(2) A dual-language context in which both
languages are used but typically with
different speakers. Switching between lan-
guages may occur within a conversation but
not within an utterance.
(3) A dense code-switching context in which
speakers routinely interleave their lan-
guages in the course of a single utterance
and adapt words from one of their lan-
guages in the context of the other. For
example, in FrenchAlsatian code-switched
speech, a speaker may adapt French verbs
through the addition of a German particle
(-ieren) as in ‘‘choisieren’’ from the French
‘‘choisir’’ rather than switch to the German
word for chose, ‘‘wa¨hlen’’ (Edwards &
Gardner-Chloros, 2007). In EnglishTaga-
log code-switched speech too there is
morphosyntactic adaptation as in: ‘‘Wala
akong cash pang grocery ngayon, if you
want, bukas na lang, ipagdadrive pa kita!
[English translation: I do not have cash for
grocery today, if you want, tomorrow, I will
even drive you there!]. The phrase ‘‘ipag-
dadrive’’ [I will even drive] is a code-
switched stretch of speech comprising a
personal pronoun, auxiliary, modifier, and
verb.
Eight cognitive control processes
Previous proposals have captured broad differ-
ences in classes of control operations. In review-
ing accounts of research on bilingual advantages
in cognitive control tasks, Hilchey and Klein
(2011) distinguished accounts that emphasise
control processes that maintain a task goal and
monitor for conflict over those that emphasise the
need to inhibit competing representations. Of
course, the processes of goal maintenance, con-
flict monitoring, and interference suppression are
all needed for implementing a given task such as
speaking in one language rather than another.
One way to recognise the interplay of these
processes is to distinguish proactive control pro-
cesses that establish a task goal from reactive
control processes that limit interference with it
(Braver, 2012). This contrast is implicit in the
inhibitory control model (Green, 1998). In this
model, the selection of a language for speaking
requires the activation of the task schema for that
language (a language task schema). Its activation
is increased because the intended language is
specified in the conceptual representation. Sche-
ma selection arises through competition with the
schemas for other languages and also, potentially,
through the reactive inhibition of representations
that trigger selection of these competing task
schemas (see Morales, Gomez-Ariza, & Bajo, in
press, for explicit recognition of this contrast and
a novel experimental test of its implications for
bilingual performance). Our basic supposition
here is that the language task schemas are in a
competitive relationship in the single language
and dual language contexts but are in co-opera-
tive relationship in the dense code-switching
context. How the schemas are coordinated affect
the various control processes that are associated
with them.
In order to articulate the dynamics of control,
we refine the decomposition proposed by Miyake
et al. (2000). These authors distinguished between
maintaining and updating information such as
task goals; inhibiting competing representations
and switching between tasks.
We consider conversation in a dual-language
context as a way to achieve a plausible decom-
position or fractionation of control processes and
assess the likely interdependence of these control
processes. The left column of Table 1 lists the
eight control processes.
A speaker must establish and maintain a task
goal such as speaking in one language rather than
another. We refer to this process as goal main-
tenance in Table 1. A face-to-face conversation is
inherently multimodal and so a variety of cues
such as the voice, face, and gestures of the
addressee may support goal maintenance. How-
ever, other cues in the immediate environment
such as the voices of other speakers talking in the
other language may activate the goal of speaking
in the other language. Maintaining the current
goal requires processes that control interference.
518 GREEN AND ABUTALEBI
We identify two such control processes in line
with Kerns et al. (2004): one that monitors for
conflict (conflict monitoring) and a process that
suppresses interference (interference suppres-
sion). These two top-down processes of control
are needed to sustain the current language goal.
We take inhibitory processes to be central to the
control of interference for neurocomputational
reasons: to speed up the time course for ensuring
maintenance of the current goal, or for the
efficient selection of a new goal and to avoid
catastrophic increases in activation (see Abutale-
bi & Green, 2007). The precise locus of suppres-
sion in bilingual speakers will depend upon the
source of the interference. For example, it may be
at the level of the language task schema itself or
at the level of particular lexical or syntactic
competitors (e.g., Green, 1986, 1998; Kroll et al.,
2006). We also leave open the precise mechanism
of suppression. It may be one that directly inhibits
the competing representation. Alternatively, it
may be one in which the target representation
and competing representation are interconnected
via mutual inhibitory links and so increasing the
activation of the target leads to suppression of the
competitor indirectly.
The detection of salient cues is integral to
successful conversation and in a dual-language
context a salient cue such as the arrival of a new
addressee may require the speaker to switch to
their other language because they typically use
that language with that addressee. We therefore
consider salient cue detection to be a control
process. Experimentally, research on cue detec-
tion is tested in a go/no-go paradigm. But such a
process may also be recruited to inhibit a pre-
potent ongoing response so as to allow a more
task relevant response (e.g., Forstmann et al.,
2008). We term this latter process ‘‘selective
response inhibition’’. Here, it serves to stop the
person continuing to speak in the current lan-
guage and triggers disengagement from it (task
disengagement). Switching between languages
requires an individual both to disengage from
the prior task and to engage with the new one
(task engagement). The speed of switching from
one task to another depends on this disen-
gagementengagement cycle. Selective response
inhibition might reasonably be viewed as an
instance of interference suppression rather than
as independent control process but it is triggered
by the need to change the task goal and so
we retain it as a special case of interference
suppression.
Switching between tasks also involves conflict
monitoring and interference suppression as a
previously active task schema must be sup-
pressed and a new one activated. We consider
task disengagement and task engagement as
distinct processes because a change of task has
wide ramifications in terms of how other pro-
cesses are configured. Speaking in English as
opposed to Mandarin, for example, requires a
shift in vocabulary, syntax, and prosodic patterns.
In fact, experimental work looking at the se-
quential behavioural effects of interference and
task switching suggests that these are best
captured by distinct conflict-control circuits
(Brown, Reynolds, & Braver, 2007). Neuroima-
ging data also implicate distinct neural regions in
task switching and interference control (e.g.,
Cools & D’Esposito, 2011).
Table 1 lists one further control process that we
term ‘‘opportunistic planning’’. By this we mean
making use of whatever comes most readily to
hand in order to achieve a goal. Speakers in
TABLE 1
Demands on language control processes in bilingual speakers as a function of the interactional context relative to demands on the
processes in monolingual speakers in a monolingual context
Interactional contexts
Control processes Single language Dual language Dense code-switching
Goal maintenance
Interference control: conflict monitoring and interference suppression
Salient cue detection  
Selective response inhibition  
Task disengagement  
Task engagement  
Opportunistic planning 
indicates the context increases the demand on that control process (more so if bolded);indicates that the context is neutral in
its effects. Please see main text for explanation of the control processes.
THE ADAPTIVE CONTROL HYPOTHESIS 519
general may plan their speech opportunistically
but we have a specific sense in mind. In the case
of bilingual speakers, we mean adapting the
words of one language to fit into the syntactic
frame of another as in the preceding examples.
Less proficient speakers of a language may also
plan their utterances opportunistically by, for
example, recruiting gesture to convey meaning,
but the control demand in their case reflects the
absence of suitable linguistic means rather than
the flexible use of available means.
Interactional contexts and demands on
control processes
In Table 1 we indicate whether the specific
interactional context increases demand on a con-
trol processes () or is unlikely to affect it ()
compared to the demand experienced by a
monolingual speaker in a monolingual interac-
tional context. This assessment is also a judgement
about the differential effect of the interactional
contexts on control processes within bilingual
speakers.
How might the different interactional contexts
affect the demand on these control processes? If
both languages are active and compete for selec-
tion, then demand on processes associated with
goal maintenance, conflict monitoring, and inter-
ference suppression may be high across all con-
texts. The contexts differ though in the explicit
presence of other languages. Both languages are
present in the dual-language and dense code-
switching contexts but not in the single-language
context. The contexts also differ in how inter-
ference may be resolved. In the single-language
and dual-language contexts interference must be
resolved so as to avoid switching into the other
language. The language task schemas are in a
competitive relationship. By contrast, in the dense
code-switching context, opportunistic planning
can make use of alternative forms of expression
that would be competing in another context. The
language task schemas are in a co-operative
relationship. This does not mean that code-
switched speech is not cognitively demanding.
Our thinking is that dual-language activation in a
dense code-switching context creates opportu-
nities for morphosyntactic integration. However,
though such a context may circumvent the need
for the strong suppression of alternatives, it
imposes a demand on the fine temporal control
of morphosyntactic processes. As a first approx-
imation then, we suggest that the demand on the
control processes of goal maintenance, conflict
monitoring, and interference suppression, is high-
est in the dual-language context. By contrast
demand on opportunistic planning is highest in
the dense code-switching context.
In Table 1 we also indicate a differential
demand on the process of salient cue detection
and on the subsequent cascade of control pro-
cesses that concern selective response inhibition,
and task disengagement and engagement. The
basis of this claim lies in the recurrent demand for
speakers in a dual-language context to circumvent
a control dilemma.
By way of illustration of this dilemma, consider
a conversation in a single-language context.
Imagine FarsiEnglish speakers in Iran who
must communicate in their second language
(English) because their visiting colleague only
speaks English. In order to do so they must
establish and stabilise a particular control state
in which, for instance, any competing linguistic
representations from Farsi are suppressed (i.e.,
the language task schema for English is domi-
nant). However, a control state in which current
goals and actions associated with them are
stabilised, that is, a state resistant to change or
interference, is at odds with a state that favours
flexibility in switching to new goals in response to
new inputs. In this scenario, it is the arrival of a
colleague who speaks Farsi but not English. This
control dilemma, in which effective suppression
limits flexibility to respond to new cues, is not a
recurrent dilemma in the single-language context
but it is one in the dual-language context. The
dilemma can be circumvented by an independent
control process that we have already identified:
salient cue detection. Such a process can then
trigger processes (selective response inhibition,
task disengagement, and task engagement) that
leads to a fluent switch in language. We conjec-
ture therefore that this cascade of processes will
be subject to adaptation in the dual-language
context.
What drives the system to adapt its
control processes?
The prototypical use of language is conversation
and conversations are joint actions in which the
participants seek to minimise joint effort in
520 GREEN AND ABUTALEBI
achieving a shared situation model (e.g., Clark,
1996). Taking this perspective as our point of
departure, we consider what we term the ‘‘inter-
actional cost’’ as a factor that motivates adaptive
changes in control processes.
In a single-language context, a conversation in
which the speaker repeatedly switched into their
native language would disrupt the conversation.
Initially, addressees might be sympathetic and
attempt to complete utterances for the speaker.
Failure to control language use, together with low
proficiency in the second language, would also
preclude the speaker contributing effectively
when the other person speaks. They would fail
to develop a suitable forward model to complete
an utterance where necessary. The bilingual
speaker would fail to ensure that the joint effort
involved in the conversation is no greater than
necessary in order to build a shared situation
model. Repeated conversations on these lines put
the bilingual’s employment at risk. Nonnative
speakers need to change in two ways. They need
to increase their second-language proficiency. But
in order to increase proficiency in the second
language, given that both languages are active,
the speaker needs to adapt the processes that
control interference especially when using the
second language, otherwise there is a computa-
tional paradox: An increase in second-language
proficiency yields a concomitant increase in
interference that reduces fluency. Interactional
cost imposes a demand to adapt the control
processes of goal maintenance, conflict monitor-
ing, and interference suppression.
In the dual-language context the demands on
control processes are more complex and, so too, is
the adaptive response. To reduce interactional
cost, speakers must sustain attention to the
current language goal and suppress interference
but be in a position to switch languages on
detection of an addressee with whom they con-
verse in their other language. Reducing the
interaction cost specifically requires that they
manage the control dilemma that reducing inter-
ference also reduces sensitivity to relevant ex-
ternal cues.
Finally, in the dense code-switching context we
have supposed that a key demand is on opportu-
nistic planning. But where is the interaction cost
that drives the adaptive response? Consider a
speaker who does not engage in dense code-
switching in a community of speakers who do.
Such a speaker clearly marks themselves out as an
outsider. But their behaviour also imposes a cost
on those they talk to because it increases the
demand on such speakers to avoid code switching
(on grounds of reciprocity). Since such speakers
are putatively less practised in controlling inter-
ference this increases joint effort and so imposes
an interaction cost. Over time, such additional
cost may lead speakers in that community to
disengage from conversation with the outsider.
In the previous three subsections we have
described the basis of the adaptive control hy-
pothesis. We turn now to consider some predic-
tions that might be made in terms of overt
behaviour and the neural regions that mediate
cognitive control.
Testing the adaptive control hypothesis
A basic prediction of the hypothesis is that
speakers in the three different interactional con-
texts will show different patterns of adaptive
response. All else being equal, the hypothesis is
refuted if the interactional context proves irrele-
vant. Indeed, one might argue that finding any
evidence for the effects of the interactional
context on any of the control processes would
be highly surprising. After all, the need to switch
between languages is present in all contexts*only
the details vary. The detection of a salient
language cue may show little or no variation
because detection of salient cues is critical to
survival whatever the interactional context. Op-
portunistic planning may also be intrinsic to
everyday conversational practice. For example,
in casual speech, speakers make use of previously
primed phrases rather than formulate them anew
precisely because they are readily to hand.
On the other hand, the core premise that the
brain adapts to demands might also make the
hypothesis irrefutable because all speakers adjust
their behaviour during an experiment to the
specific control demands it imposes. However if,
as the hypothesis envisages, recurrent control
demands lead to specific adaptive changes, then
the control states they mediate should be trig-
gered more easily by relevant experimental con-
ditions. In other words, any learning effects during
the experiment should be observed earlier for
speakers from an interactional context that is best
suited to the experimental conditions. In fact,
under optimal conditions, the relevant control
state should be triggered immediately.
We also acknowledge the issue of circular
causation. Interactional context may shape
THE ADAPTIVE CONTROL HYPOTHESIS 521
adaptive response but individual differences (in
predispositions and genetic make-up) surely
constrain such effects. We consider this issue in
more detail in the later subsection on individual
differences. For now, we assume that individual
variation in executive control processes and, in
sociality, is normally distributed in speakers in the
three interactional contexts. In consequence,
adaptive effects of these contexts should be
observable. We consider behavioural predictions
first of all.
Behavioural predictions
We expect that control processes are predictive
of performance in conversational, dialogic tasks
where joint effort must be minimised (see
Festman, 2012, and Pivneva, Palmer, & Titone,
2012, for novel research examining the link
between executive control tasks and conversa-
tional performance in bilingual speakers). Sup-
pose individuals have to describe a depicted event
to an addressee. In one experimental condition
they are free to switch between languages at will.
In another they are required to switch on cue into
one of their languages or the other. For speakers
from a dense code-switching context, fluent
performance will be associated with the freedom
to use either language, whereas an imperative cue
requiring them to restrict their utterances to a
single language will impair performance. By
contrast, baseline fluency will be relatively great-
er for those bilinguals from the single-language
and dual-language contexts when only one lan-
guage is required. For speakers from these two
contexts, the adaptive control hypothesis makes
further predictions. A cue to switch languages
should trigger a set of processes where the
cascade is more integrated for bilinguals in the
dual-language context compared to those from a
single-language context. Two effects are pre-
dicted. The transitory cognitive load imposed in
response to the imperative cue might occur ear-
lier for speakers from a dual-language context
and so affect their current speech precisely
because the cue sets processes in train to switch
language more rapidly. However, as a corollary,
once the switch in language has occurred, hesita-
tion and speaking rate in such speakers should
return to baseline more rapidly.
The basis of these predictions lies in how the
language tasks schemas are coordinated in the
different contexts and in the planning processes of
speech. In a dense code-switching context speak-
ers establish a cooperative relationship between
the schemas. This permits opportunistic planning
in which the speech plan reflects the dynamic
accessibility of words and constructions regardless
of their language membership and incorporates
items contingent solely on them meeting current
syntactic constraints. In such circumstances there
is no basis for predicting a switch cost. The same
possibility exist for speakers from the other
contexts instructed to switch at will but they
have to overcome their default coordination in
which the language tasks schemas compete to
control output and so routinely restrict access to
the speech plan to items and constructions from
just one language.
Whereas switching language during sentence
production may incur no overt cost for speakers
from a dense code-switching context, switching
language in response to an imperative cue to
name a picture will continue to incur a cost even
for speakers in a dual language context adept at
switching between languages because in such
circumstances speakers must continue to establish
a competitive relationship between the language
schemas.
With respect to experimental tasks that tap
specific component control processes adaptive
effects should be evident in the analysis of
reaction time distributions for conflict tasks such
the colourword Stroop task. Interference effects
can be plotted as a function of response speed and
these plots show that interference effects decrease
for slower responses and more so for individuals
who are more proficient in inhibition. Our
analysis leads to the prediction that bilingual
speakers in the dual-language context will be
the more proficient in inhibition than those
in the single-language or dense code-switching
contexts.
A more critical prediction concerns how speak-
ers adapt to the control dilemma in which a
suppressive state limits flexibility to respond to a
cue indicating a new task. Experimentally we can
examine the relationship between response to an
interference trial and a response to an immedi-
ately following trial involving a language switch.
We can attempt to maximise sensitivity by ensur-
ing that the cue signalling a different language
coincides with a property that must be suppressed
(cf. Goschke & Dreisbach, 2008). In a colour
word Stroop task, the written word is the sup-
pressed property and so a printed cue on a
following (neutral) trial that signals a change
522 GREEN AND ABUTALEBI
in language should be less readily detected.
For speakers in the single-language and dense
code-switching contexts reduced verbal interfer-
ence on the interference trial will be associated
with increased language switching costs. By con-
trast, if speakers in the dual-language context
have adapted to circumvent the control dilemma
then these two effects should dissociate. We have
proposed a test using a visual Stroop task yet the
nature of any adaptive response may be specific
to the multimodal nature of the typical conversa-
tional exchanges. A cross-modal Stroop task is
perhaps more pertinent. Here the auditory input
is the suppressed dimension and the salient cue is
an auditory one (e.g., in the language to be
switched into). Indeed, testing the specificity of
the cue (to known voices vs. the voices of others)
provides a way to explore the precise tuning of
the process of salient cue detection.
Neuroimaging predictions
Empirical research (e.g., Luk, Bialystok, Craik, &
Grady, 2011) that has established a widespread
difference in the brains of adult monolingual and
bilingual speakers encourages the search for the
nature and origins of specific adaptive changes. We
consider predictions that relate specific control
processes to particular neural regions and net-
works. Figure 2 provides a schematic description of
the neural structures and their connections that we
associate with language control processes. In the
figure the networks involved in language control
are indicated with bidirectional continuous arrows,
whereas those indicating part of the speech pipe-
line are indicated with dashed lines.
For the speech pipeline we reference regions
based on the analysis of verbal fluency data
(Eickhoff, Heim, Zilles, & Amunts, 2009). We
treat these as representative of regions active in
self-generated speech and a subset of those active
in conversational speech. Unrepresented in the
figure is the monitoring of speech output that
involves anterior and posterior regions of the
temporal cortex. Also unrepresented are input
regions to the left cortex that contribute to the
conceptual content to be expressed. In previous
papers (e.g., Abutalebi & Green, 2007, 2008), we
reviewed work showing that both cortical and
subcortical structures are involved in language
control and language switching (see Luk, Green,
Abutalebi, & Grady, 2012, for a recent meta-
analysis). In this work we identified the anterior
cingulate cortex (ACC) and the presupplemen-
tary motor area (pre-SMA) with conflict monitor-
ing and acknowledged a role of the pre-SMA in
initiating speech in language switching (see Luk
et al., 2012, for further discussion). We associated
the control of interference with left prefrontal and
inferior cortex, parietal cortices with the main-
tenance of task representations, and one subcor-
tical structure (the caudate) in the basal ganglia
with the switching between languages (Abutalebi
et al., 2013). On one proposal, basal ganglia
circuits, more generally, serve to control or gate
access between prefrontal cortex and posterior
cortical regions that represent task information
Figure 2. Simplified language control network and speech production regions (see main text for explanation).
THE ADAPTIVE CONTROL HYPOTHESIS 523
(Cools & D’Esposito, 2011; see also Frank, 2011).
We extend the networks involved in language
control in Figure 2.
In Figure 2 we include a direct link between
frontal cortex and the right cerebellum as a
further circuit involved in language control
(Fabbro, Moretti, & Bava, 2000). Neuropsycho-
logical data and functional data indicate that
damage to the right cerebellum suppresses activa-
tion in left frontal cortex and elicits aphasic
symptoms typical of those shown by frontal lobe
patients. Reperfusion of the cerebellum reduces
such symptoms (Marien, Engelborghs, Fabbro, &
De Deyn, 2001). Neuropsychological data impli-
cate cerebellar structures specifically in the con-
trol of morphosyntax such that their damage leads
to substitution of inappropriate bound mor-
phemes (Silveri, Leggio, & Molinari, 1994). We
therefore consider that it will play a critical role
for speakers in dense code-switching contexts.
We also include a circuit involved in the
detection of salient cues (salient cue detection).
This circuit involves regions in right inferior
frontal cortex (e.g., Aron, Behrens, Smith, Frank,
& Poldrack, 2007) and the thalamus. The thala-
mus directly accesses regions of the basal ganglia
such as the caudate and putamen (Smith, Surme-
ier, Redgrave, & Kimura, 2011). We note that left
subcortical regions such as the caudate seem more
involved in the control of verbal interference
(e.g., Abutalebi et al, 2008; Ali, Green, Kherif,
Devlin, & Price, 2010). Finally, we suppose
reciprocal connections between basal ganglia
structures and the cerebellum (Bostan, Dum, &
Strick, 2010).
Adaptive effects should be expressed in the
neural regions and circuits that mediate the
control demands in each context. For example,
in order to implement the cascade of control
processes required in the dual-language context,
the hypothesis predicts changes in the network
comprising cortical, thalamic, and basal ganglia
regions. In addition, there is a demand on frontal
structures linked to conflict monitoring and
interference suppression (e.g., Kerns et al., 2004)
and to parietal regions (and frontal regions)
associated with task changes. By contrast, the
hypothesis predicts an adaptation in speakers in
dense code-switching contexts involving the con-
nectivity of right cerebellar and left inferior
frontal regions. In the single-language context,
the demand is to ensure efficient suppression of
the nontarget language over extended periods of
time. To the extent this is successful, there is no
additional demand on subcortical structures asso-
ciated with language switching.
Adaptive effects may also be revealed by the
patterns of neural activation as speakers from
the different contexts perform the same task, by
the distinct correlations of behavioural perfor-
mance with measures of structure or connectivity
and in neuropharmacological assessments. We
present an illustrative prediction for each type
in the following paragraphs.
Recent work has distinguished different inhi-
bitory mechanisms in the control of language:
those involved in globally suppressing language
representations (relevant perhaps when a lan-
guage is used over an extended period of time)
and those involved in more local suppression
when participants switch on cue between two
languages. One study (Guo, Liu, Misra, & Kroll,
2011) contrasted a condition in which picture
naming was blocked by language and one where
participants switched between their two lan-
guages. Guo et al. (2011) observed that the
ACC (and supplementary motor area) was more
active during language switching (consistent with
local inhibition) and the dorsal left frontal cortex
and parietal cortex were more active during
language blocking (consistent with global inhibi-
tion). Indeed in a context where just one language
is used for naming, frontal regions but not
subcortical regions, show increased activation in
bilingual relative to monolingual speakers and
such activation is linked to the control of inter-
ference (Parker Jones et al., 2012). One inter-
pretation is that frontal regions in such a context
select the target name from an activated set of
competing names at an early stage in the produc-
tion process (i.e., before release of the utterance
plan). By contrast, subcortical regions play a more
critical role during switching consistent with a
later selection of the target name (Green, 2011).
The adaptive control hypothesis predicts that the
precise pattern of blocked naming effects will
depend on the interaction context. The difference
in neural response between blocked naming and
language switching should be greater for those in
the single-language interaction context who are
unused to language switching compared to those
in the dual-language context. Left caudate activa-
tion should show a marked increase reflecting the
increased demand on processes involving late
selection. Speakers from a dense code-switching
context may show a distinct profile in which they
use the cerebellar-frontal circuit to mediate lan-
guage switching.
524 GREEN AND ABUTALEBI
The adaptive control hypothesis predicts that
the interactional context differentially affects
processes that circumvent the control dilemma
in which a suppressive state, induced to control
interference, makes the system less response to a
cue signalling a change in task. How might this
prediction be tested? Prior neuroimaging research
(Luk, Anderson, Craik, Grady, & Bialystok, 2010)
has combined an interference suppression task
with a go/no-go task, but we need a paradigm in
which individuals respond to a salient cue despite
a suppressive control state. We can use the
experimental task described previously that com-
bines a verbal interference task with language
switching. A common pattern across the three
contexts should be found for the interference
trials: increased activation in left inferior frontal
regions predictive of reduced interference (e.g.,
Ali et al., 2010; Parker-Jones et al., 2012). Of
specific interest are the data in which a language
switch trial follows an interference trial. On these
language switching trials, there should be distinct
effects of interactional context. The required
cascade of control processes should be better
synchronised for speakers from the dual-language
context compared to speakers from the single-
language and dense code-switching contexts. One
possible correlate is a stronger coupling of activa-
tion in right inferior, thalamic and basal ganglia
regions and frontal-parietal regions. Probing the
cascade more precisely requires studies that relate
behavioural indices, from tasks that tap specific
control processes, to the neural response.
Individual differences
We have illustrated how the three interactional
contexts lead to adaptive changes in control
processes. The proximal cause of adaptive change
is a person’s engagement with the recurrent forms
of conversational exchange in that context. It
follows that we can recast our proposal at the level
of the individual speaker. We can ask about the
extent to which they engage in the type of exchange
suited to a given context. We can also ask about the
extent to which they experience different interac-
tional contexts, that is their individual behavioural
ecologies. We consider this question first of all.
Our three interactional contexts are defined by
a single type of exchange but, in principle, speakers
may experience all three contexts and may experi-
ence these contexts to different extents. We can
envisage a space of possible speakers characterised
by the distribution of the types of conversational
exchange in which they engage. For example, a
speaker may have a preponderance of single-
language context exchanges, some dual-language
context exchanges, and no dense code-switching
exchanges. Another may have a preponderance of
dense code-switching exchanges, some single-lan-
guage context exchanges, and no dual-language
context exchanges. According to the adaptive
control hypothesis, the precise pattern of adaptive
changes to the control processes and their neural
basis will reflect the actual pattern of exchanges.
Such variety in the types of exchange is not just a
theoretical possibility. At least in terms of single-
language versus dual-language exchanges it fits the
results of an interesting questionnaire-based study
of language use in SpanishCatalan bilingual
university students (Rodriguez-Fornells, Kra¨mer,
Lorenzo-Seva, Festman, & Mu¨ nte, 2012). In this
study, although two-thirds of the sample used both
Catalan and Spanish at university, two-thirds used
either one or the other language at home. For
simplicity’s sake, though we endorse our recasting,
we will continue to refer to the interactional
contexts as these define the specifics of the
adaptive response.
We turn now to other factors that may con-
strain how individuals respond in a given interac-
tional context. We argued that speakers adopt
conversational practices suited to that interac-
tional context because there is an interaction cost
in not doing so. One constraint on their ability to
avoid such a cost is their proficiency in the two
languages. The relationship between proficiency
and specific adaptive changes as a function of
interactional context is unlikely to be straightfor-
ward and we have implicitly assumed that our
speakers are highly proficient in both languages
and so examined adaptive changes under condi-
tions of relatively stability from a proficiency
point of view. We simply note here that for
speakers in single- and dual-language contexts
an increase in proficiency is most likely associated
with increased skill in the control of interference.
The same may only be true for those in dense
code-switching contexts until they can begin to
use their knowledge of the two languages oppor-
tunistically.
A different constraint is that individuals may
vary in their sensitivity to this interaction cost*
perhaps because they differ in sensitivity to social
cues in general. In consequence they may not
engage in the conversational practices that im-
pose the recurrent demands needed to entrain
THE ADAPTIVE CONTROL HYPOTHESIS 525
adaptive changes. Accordingly, a corollary of the
hypothesis is that sensitivity to interaction cost
will predict the extent to which individuals engage
in exchanges typical of that interaction context.
Testing this prediction requires an analysis of
individual conversational practice. A further issue
is the quantitative relationship between the num-
ber of recurrent exchanges and adaptive changes.
We presume the relationship is nonlinear and
reflects the typical relationship between practice
and performance.
Sensitivity to interaction costs is not the only
factor that may constrain adaptive change. In-
dividuals vary in their capacity for cognitive
control. Scores on tests of executive function are
known to predict cross-language intrusion errors
in which words from a person’s first language
intrude on their speech in their second language
(e.g., Festman, 2012; Rodriguez-Fornells et al.,
2012). In the single and dual-language contexts,
the integrity of left frontal structures will predict
cross-language intrusion errors. Potentially, cross-
language intrusions can be used opportunistically
in dense code-switching contexts. If such oppor-
tunistic use requires the integrity of the right
cerebellar-left frontal circuit, then indices of its
integrity (e.g., regional variations in grey matter
density or white matter connectivity) will predict
the extent to which a speaker engages in dense
code-switching and their facility and perhaps
pleasure in doing so.
Examining the role of individual differences in
adaptive response is a discovery procedure. It
provides a way to identify different circuits that
may mediate the same control process. Suppose,
by way of example, that, as predicted, the right
cerebellar-left frontal circuit showed adaptive
changes in speakers in a dense code-switching
context. Impairment to this circuit should then
preclude dense code-switching if that circuit was
necessary. The identification of speakers with
such impairment whose code switching was nor-
mal from a behavioural point of view immediately
implies that there must be at least one other
circuit that can fulfil the function. Individual
differences therefore provide a way to explore
such degeneracy in control processes (Green,
Crinion, & Price, 2006).
Preexisting neuropharmacological differences
are also relevant to the adaptive response of the
control systems. Individuals with increased dopa-
mine receptors in dorsal striatal regions (the
caudate and putamen) stop more quickly in
response to a stop signal (a salient cue) and their
response profile is associated with increased
inhibition-related activation in frontal-striatal
pathways (Ghahremani et al., 2012). In turn,
dopamine receptor values depend on versions of
the COMT gene. Current data suggest that
individuals with one allele (Met) are better at
tasks that require interference suppression
whereas individuals with the other variant (Val)
are better at switching to a new task (Cools &
D’Esposito, 2010). It would be of great interest to
see how the interactional context shapes perfor-
mance when these genetic differences are taken
into account. The adaptive control hypothesis
predicts that processes associated with salient
cue detection are targets for adaptation. One
speculative prediction is that receptor density will
change as function of control demands. The
adaptive control hypothesis then promotes long-
itudinal studies charting adaptive changes in
bilingual speakers in the light of genetic variation.
REVIEW AND DISCUSSION
We have proposed the adaptive control hypoth-
esis. We identified interaction cost as a factor that
drives control processes to adapt with the precise
adaptation shaped by the interactional context.
The processes of cognitive control adapt by
changing the parameters of their operation in-
cluding their coordination with other control
processes. Speakers in a single-language seek to
maintain the current language goal and avoid
cross-language intrusions. We associated effective
suppression of the nontarget language with left
inferior regions. Speakers in a dense code-switch-
ing context opportunistically use joint language
activation to create novel mixed-language utter-
ances. On the basis of neuropsychological data,
and the requirement for fine temporal control in
this context, we associated adaptive change with
the left frontal and right cerebellar circuit. Speak-
ers in the dual-language context are faced with a
control dilemma: A suppressive state that limits
interference from the current nontarget language
also restricts the speed of response to a cue
signalling a change to that language. We proposed
that they circumvent this dilemma by linking a
region that detects salient cues to those involved
in selective response inhibition, task disengage-
ment, and engagement. This circuit includes a
right inferior frontal region, the thalamus, and
basal ganglia structures. We have provided illus-
trative behavioural and neuroimaging predictions.
526 GREEN AND ABUTALEBI
Individual differences are also pertinent to
testing the hypothesis. The proximal cause of
adaptive change is the recurrent exchanges typical
of that context. However, individuals may vary in
the extent to which they engage with typical
recurrent exchange associated with a given inter-
actional context and over what period of time.
Differences in sensitivity to interaction cost,
differences in the capacity for cognitive control
and differences in circuit neuroanatomy may all
constrain engagement and affect the degree of
adaptive change. Individuals may also experience
different interactional contexts (single language,
dual language, dense code-switching) and so the
adaptive response will reflect the distribution of
exchanges typical of those contexts.
It is possible too that individuals differ in how
they respond to control demands. We have sup-
posed, as a first approximation, that goal main-
tenance, conflict monitoring, and interference
suppression will all show adaptive changes in
speakers in a dual-language context. But it may
be objected that this pattern depends on precisely
how individuals manage the relative activation of
their languages (i.e., the extent of the competitive
relation between the language schemas). Concei-
vably, careful balancing of the relative activation of
the two languages (e.g., through proactive control)
may limit the need to control between-language
interference (and use reactive inhibition) and so
reduce the need to circumvent the control dilem-
ma described earlier. If so, the ability to switch
rapidly between languages (involving selective
response inhibition and task disengagement and
engagement processes) may dissociate from the
skilled control of interference and salient cue
detection. However, in our view, speaking in one
language to the exclusion of another is intrinsically
linked to a suppressive state. This state of affairs
does not arise in a dense code-switching context.
Exploring control states experimentally requires
an increased focus on the pattern of performance
within individuals on a range of tasks that tap
different processes of language control.
Other specific factors may be relevant to
particular contexts. For example, in the dense
code-switching context, the extent of opportunis-
tic planning will vary with the variety of locally
adjusted forms that the speakers use. Speakers
will also differ in the novelty of their own code-
switched forms (see Wei, 2013). Such variations
may be critical to the adaptive response. Other
types of code-switching exchange (see Deuchar,
Muyksen, & Wang, 2007) may impose somewhat
different demands. In the dual-language contexts,
speakers may vary in their styles of conversa-
tional management. Conversations are multimo-
dal. Face and hand movements are known to
facilitate comprehension and to configure the
regions involved in language comprehension
(e.g., Skipper, Goldin-Meadow, Nusbaum, &
Small, 2009). They also allow a nonverbal channel
(a smile, a handshake) to signal acknowledge-
ment of a new addressee. In everyday conversa-
tional exchanges, such expressions and gestures
are integrated with the processes of language
switching. An exploration of adaptive changes
therefore requires a shift in emphasis to more
ecologically valid forms of multimodal exchanges.
Generalisations to other tasks
In keeping with focus of this paper, we have
suggested tests of the adaptive control hypothesis
using language-based tasks. The adaptive control
hypothesis may also be used to make predictions
about the performance on tasks that are not
directly tied to language control. At a general
level, others have pointed to the relationship
between control states required in bilingual speak-
ers and those required for certain types
of thinking (e.g., Hommel, Colzato, Fischer, &
Christoffels, 2011). The basic idea is that control-
ling the interference of two language leads to a
control state conducive to convergent thinking but
inimical to divergent thinking. Using different
verbal materials to assess convergent and diver-
gent thinking, Hommel et al. (2011) found that
bilinguals were better at convergent thinking and
monolinguals at divergent thinking. Such an out-
come, we suggest might not hold for speakers from
a dense code-switching context. Instead, those
adept at dense code-switching might be skilled at
a form of mental synthesis (Green, 2011). They
might, for example, be faster to envisage how a pair
of letters might depict an object (e.g., an umbrella
from the letters J and D or a chair from the letters I
and N; cf. Pearson, Logie, & Gilhooly, 1999).
The adaptive control hypothesis can also be
used to ground more precise predictions about
performance of speakers on nonverbal tasks.
Existing data point to associations between
performance on language switching tasks and
nonverbal switching tasks (Prior & Gollan, 2011;
Soveri, Rodriguez-Fornells, & Laine, 2011),
between intrusion errors in a single-language
conversational context and cognitive measures
THE ADAPTIVE CONTROL HYPOTHESIS 527
of executive functioning (Festman, 2012) and
between measures of language switching and
the control of nonverbal interference (Linck,
Schwieter, & Sunderman, 2012). The adaptive
control hypothesis makes predictions about spe-
cific control processes that are contingent on the
interactional context or, more generally, the
distribution and nature of recurrent exchanges.
In that sense it provides a rationale for the
detailed characterisation of bilingual speakers if
robust and replicable findings are to be obtained.
The hypothesis also envisages that it is the pattern
of performance on a range of tests that is critical.
One basic prediction is that where a task
matches the control demands of that context,
bilingual speakers will show more rapid adapta-
tion. Compared with monolingual Spanish speak-
ers, CatalanSpanish speakers, two-thirds of
whom, we may infer from Rodriguez et al.
(2012), operate in a dual-language context, dis-
played better interference suppression in a non-
verbal flanker task during early trial blocks
(Costa, Herna
´ndez, & Sebastian-Galle
´s, 2008).
Speakers from a single-language context, but not
those from a dense code-switching context, may
also show some benefit. The benefit for speakers
in a dual-language context may be linked to the
increased efficiency of an anterior cingulate
region (see Abutalebi et al., 2011).
In terms of the pattern of performance, superior
skill in resolving nonverbal interference should be
associated with superior skill in circumventing the
control dilemma for those in the dual-language
context. Goschke and Dreisbach (2008) showed
that individuals were less likely to detect a cue that
signalled a different task and required a different
response when that cue coincided with a property
that had to be ignored in the primary task. The
adaptive control hypothesis predicts that relative
to monolingual speakers, speakers from the dual-
language but not from other bilingual interactional
contexts would be more adept at background
monitoring and so more frequently detect the
cue and make the different response required.
Such associations leave open the precise
mechanisms. An advantage may arise because
control is exercised by a common mechanism,
that is, a common network of neural regions
or a common pool of resources. Alternatively,
language control may involve a distinct control
network that emerges from a more general system
mediating action control. Bilingual speakers
routinely recruit this specialised language control
system to handle nonlinguistic tasks. Monolingual
speakers do so too*they also use language to label
objects and events and to sequence actions. How-
ever, because the language control network is
shaped by different demands in the two cases
(depending, for example, on the interactional
context) different effects arise. In consequence,
monolingual and bilingual speakers may display
different patterns of relations linking verbal
and nonverbal performance (e.g., Blumenfeld &
Marian, 2011). Lastly, the network for language
control and that for the control of nonlinguistic
actions may be jointly active. In this case, differ-
ences between bilingual and monolingual speakers
may arise because of synchronisation between the
two networks with neural response reflecting their
joint but distinct operation. Discriminating these
alternatives in terms of functional imaging data
requires analysis of the causal dynamics of control.
Our analyses point to the need for further dis-
criminating tests linked to a detailed profile of the
speakers’ use of their languages.
CONCLUSION
We have concentrated on the suggestion that the
interactional context (the typical interactional
exchange in those contexts) is important in leading
bilingual speakers to adapt their cognitive control
processes and to tune the networks of control. We
have exemplified the adaptive control hypothesis.
Our illustrative predictions may prove false. There
may be no systematic behavioural or neural
differences as a function of the interactional
contexts but we hope to have established that
there is value in identifying the demand on control
processes and how the precise contexts of use
shape their properties, coordination, and cascade.
Exploration of the adaptive control hypothesis
requires a continued shift to studies that examine
the pattern of performance (behavioural and
functional) on a range of tasks that tap specific
processes of language control. From a practical
point of view this shift requires the development
of efficient testing protocols. These would also be
of clinical relevance. This approach, linked to a
detailed characterisation of the behavioural ecol-
ogy of the bilingual speaker, will play a vital role in
the development of neurocomputational models
of speech production in such speakers.
528 GREEN AND ABUTALEBI
REFERENCES
Abutalebi, J., Annoni, J. M., Seghier, M., Zimine, I.,
Lee-Jahnke, H., Lazeyras, F.,... & Khateb, A.
(2008). Language control and lexical competition
in bilinguals: An event-related fMRI study. Cerebral
Cortex,18, 14961505.
Abutalebi, J., Della Rosa, P. A., Ding, G., Weekes, B. S.,
Costa, A., & Green, D. W. (2013). Language
proficiency modulates the engagement of cognitive
control areas in multilinguals. Cortex,49, 905911.
Abutalebi, J., Della Rosa, P. A., Green, D. W.,
Hernandez, M., Scifo, P., Keim, R., ... & Costa, A.
(2011). Bilingualism tunes the anterior cingulate
cortex for conflict monitoring. Cerebral Cortex,22,
20762086.
Abutalebi, J., & Green, D. W. (2007). Bilingual
language production: The neurocognition of lan-
guage representation and control. Journal of Neu-
rolinguistics,20, 242275.
Abutalebi, J., & Green, D. W. (2008). Control mechan-
isms in bilingual language production: Neural evi-
dence from language switching studies. Language
and Cognitive Processes,23, 557582.
Ali, N., Green, D. W., Kherif, F., Devlin, J. T., & Price,
C. J. (2010). The role of the left head of caudate in
suppressing irrelevant words. Journal of Cognitive
Neuroscience,22, 23692386.
Aron,A.R.,Behrens,T.E.,Smith,S.,Frank,M.J.,
& Poldrack, R. A. (2007). Triangulating a cogni-
tive control network using diffusion-weighted
magnetic resonance imaging (MRI) and func-
tional MRI. Journal of Neuroscience,27, 3743
3752.
Bialystok, E., Craik, F. I. M., Green, D. W., & Gollan,
T. H. (2009). Bilingual minds. Psychological Science
in the Public Interest,10,89129.
Bialystok, E., & Depape, A. M. (2009). Musical
training, bilingualism, and executive functioning.
Journal of Experimental Psychology: Human Per-
ception and Performance,35, 565574.
Blumenfeld, H. K., & Marian, V. (2011). Bilingualism
influences inhibitory control in auditory comprehen-
sion. Cognition,118, 245257.
Bostan, A. C., Dum, R. P., & Strick, P. L. (2010). The
basal ganglia communicate with the cerebellum.
Proceedings of the National Academy of Sciences,
107, 84528456.
Botvinick, M. M., Cohen, J. D., & Carter, C. (2004).
Conflict monitoring and anterior cingulate: An
update. Trends in Cognitive Sciences,8, 539546.
Braver, T. S. (2012). The variable nature of cognitive
control: A dual mechanisms framework. Trends in
Cognitive Sciences,16, 106113.
Brown, J. W., Reynolds, J. R., & Braver, T. S. (2007). A
computational model of fractionated conflict-
control mechanisms in task switching. Cognitive
Psychology,55,3785.
Clark, H. H. (1996). Using language. Cambridge:
Cambridge University Press.
Cools, R., & D’Esposito, M. (2011). Inverted U-shaped
dopamine actions on human working memory and
cognitive control. Biological Psychiatry,69, e113
e125.
Costa, A., Herna
´ndez, M., & Sebastian-Galle
´s, N.
(2008). Bilingualism aids conflict resolution: Evi-
dence from the ANT task. Cognition,106,5986.
Deuchar, M., Muyksen, P., & Wang, S.-L. (2007).
Structured variation in code-switching: Towards an
empirically based typology of bilingual speech
patterns. International Journal of Bilingual Educa-
tion and Bilingualism,10, 298340.
Edwards, M., & Gardner-Chloros, P. (2007). Compound
verbs in code switching: Bilinguals making do?
International Journal of Bilingualism,11,7391.
Eickhoff, S. B., Heim, S., Zilles, K., & Amunts, K.
(2009). A systems perspective on the effective
connectivity of overt speech production. Philoso-
phical Transactions of the Royal Society,367A,
23992421.
Fabbro, F., Moretti, R., & Bava, A. (2000). Language
impairments in patients with cerebellar lesions.
Journal of Neurolinguistics,13, 173188.
Festman, J. (2012). Language control abilities of late
bilinguals. Bilingualism: Language and Cognition,
15, 580593.
Forstmann, B. U., Jahfari, S., Scholte, H. S., Wolfen-
steller, U., van den Wildenberg, W. P. M., &
Ridderinkhof, K. R. (2008). Function and structure
of the right inferior frontal cortex predict indivi-
dual differences in response inhibition: A model-
based approach. Journal of Neuroscience,28, 9790
9796.
Frank, M. J. (2011). Computational models of moti-
vated action selection in corticostriatal circuits.
Current Opinion in Neurobiology,21, 381386.
Frank, S. L., Bod, R., & Christiansen, M. H. (2012).
How hierarchical is language use? Proceedings of
the Royal Society: Biological Sciences,279B, 4522
4531. doi:10.1098/rspb.2012.1741
Ghahremani, D. G., Lee, B., Roberstson, C. L., Tabib-
nia, G., Morgan, A. T., De Shetler, N., ... London,
E. D. (2012). Striatal dopamine D
2
/D
3
receptors
mediate response inhibition and related activity in
the fronto-striatal neural circuitry in humans. Jour-
nal of Neuroscience,32, 73167324.
Goschke, T., & Dreisbach, G. (2008). Conflict-triggered
goal shielding: Response conflicts attenuate back-
ground monitoring for prospective memory cues.
Psychological Science,19,2532.
Green, D. W. (1986). Control, activation and resource.
Brain and Language,27, 210223.
Green, D. W. (1998). Mental control of the bilingual
lexico-semantic system. Bilingualism: Language and
Cognition,1,6781.
Green, D. W. (2011). Language control in different
contexts: The behavioural ecology of bilingual
speakers. Frontiers in Psychology,2(103), 14.
Green, D. W., Crinion, J., & Price, C. J. (2006).
Convergence, degeneracy and control. Language
Learning,56,99125.
Guo, T., Liu, H., Misra, M., & Kroll, J. F. (2011). Local
and global inhibition in bilingual word production:
fMRI evidence from Chinese-English bilinguals.
NeuroImage,56, 23002309.
THE ADAPTIVE CONTROL HYPOTHESIS 529
Hedden, T., Ketay, S., Aron, A., Markus, H. R., &
Gabrieli, J. D. E. (2008). Cultural influences on
neural substrates of attentional control. Psychologi-
cal Science,19(1), 1217.
Hilchey, M. D., & Klein, R. M. (2011). Are there
bilingual advantages on nonlinguistic interference
tasks? Implications for the plasticity of executive
control processes. Psychonomic Bulletin and Review,
18, 625658.
Hommel, B., Colzato, L. S., Fischer, R., & Christoffels,
I. (2011). Bilingualism and creativity: Benefits in
convergent thinking come with losses in divergent
thinking. Frontiers in Psychology,2(273), 15.
Kerns, J. G., Cohen, J. D., MacDonald, A. W., III, Cho,
R. Y., Stenger, V. A., Aizenstein, H., & Carter, C. S.
(2004). Anterior cingulate conflict monitoring and
adjustments in control. Science,303, 10231026.
Krizman, J., Marian, V., Shook, A., Shoe, E., & Kraus,
N. (2012). Subcortical encoding of sound is en-
hanced in bilinguals and relates to executive func-
tion advantages. Proceedings of the National
Academy of Sciences of the USA,109(20), 7877
7881. doi:10.1073/pnas.1201575109
Kroll, J. F., Bobb, S. C., & Wodniecka, Z. (2006).
Language selectivity is the exception, not the rule:
Arguments against a fixed locus of language selec-
tion in bilingual speech. Bilingualism: Language and
Cognition,9, 119135.
Kuipers, J. R., & Thierry, G. (2010). Event-related brain
potentials reveal the time-course of language change
detection in early bilinguals. NeuroImage,50, 1633
1638.
Linck, J. A., Schwieter, J. W., & Sunderman, G. (2012).
Inhibitory control predicts language switching per-
formance in trilingual speech production. Bilingual-
ism: Language and Cognition,15, 651662.
Luk, G., Anderson, J. A. E., Craik, F. I. M., Grady, C.,
& Bialystok, E. (2010). Distinct neural correlates for
two types of inhibition in bilinguals: Response
inhibition versus interference suppression. Brain
and Cognition,74, 347357.
Luk, G., Bialystok, E., Craik, F. I. M., & Grady, C. L.
(2011). Lifelong bilingualism maintains white matter
integrity in older adults. Journal of Neuroscience,31,
1680816813.
Luk, G., Green, D. W., Abutalebi, J., & Grady, C.
(2012). Cognitive control for language switching in
bilinguals: A quantitative meta-analysis on func-
tional neuroimaging studies. Language and Cogni-
tive Processes,27, 14791488.
Macizo, P., Bajo, T., & Martin, M. C. (2010). Inhibitory
processes in bilingual language comprehension: Evi-
dence from Spanish-English interlexical homographs.
Journal of Memory and Language,63,232244.
Marien, P., Engelborghs, S., Fabbro, F., & De Deyn, P. P.
(2001). The lateralized linguistic cerebellum: A
review and a new hypothesis. Brain and Language,
79, 580600.
Miyake, A., Freidman, N. P., Emerson, M. J., Witzki, A.
H., Howerter, A., & Wagner, T. (2000). The unity
and diversity of executive functions and their
contributions to complex frontal lobe tasks: A
latent variable analysis. Cognitive Psychology,41,
49100.
Morales, J., Gomez-Ariza, C. J., & Bajo, B. M. (in
press). Dual mechanisms of cognitive control in
bilinguals and monolinguals. Journal of Cognitive
Psychology. doi:10.1080/20445911.2013.807812
Parker Jones, ’O
¯., Green, D. W., Grogan, A., Pliatsikas,
C., Filippopolitis, K., Ali, N., ...Price, C. J. (2012).
Where, when and why brain activation differs for
bilinguals and monolinguals during picture naming
and reading aloud. Cerebral Cortex,22, 892902.
Pearson, D., Logie, R. H., & Gilhooly, K. J. (1999).
Verbal representations and spatial manipulation
during mental synthesis. European Journal of Cog-
nitive Psychology,11, 295314.
Pivneva, I., Palmer, C., & Titone, D. (2012). Inhibitory
control and L2 proficiency modulate bilingual lan-
guage production: Evidence from spontaneous
monologue and dialogue. Frontiers in Psychology,
3(57), 118.
Prior, A., & Gollan, T. H. (2011). Good language-
switchers are good task-switchers: Evidence from
Spanish-English and Mandarin-English bilinguals.
Journal of the International Neuropsychological
Society,17, 682691.
Rodriguez-Fornells, A., Kra¨ mer, U. M., Lorenzo-Sava,
U., Festman, J., & Mu¨ nte, T. F. (2012). Self-assess-
ment of individual differences in language switching.
Frontiers in Psychology,2(388), 115.
Silveri, M. C., Leggio, M. G., & Molinari, M. (1994).
The cerebellum contributes to linguistic production:
A case of agrammatism of speech following right
hemicerebellar lesion. Neurology,44, 20472050.
Skipper, J. L., Goldin-Meadow, S., Nusbaum, H. C., &
Small, S. L. (2009). Gestures orchestrate brain
networks for language understanding. Current Biol-
ogy,19, 661667.
Smith, Y., Surmeier, D. J., Redgrave, P., & Kimura, M.
(2011). Thalamic contributions to basal ganglia-
related behavioral switching and reinforcement.
Journal of Neuroscience,31, 1610216106.
Soveri, A., Rodriguez-Fornells, A., & Laine, M. (2011).
Is there a relationship between language switching
and executive functions in bilingualism? Introducing
a within group analysis approach. Frontiers in
Psychology,2(183), 18.
Stout, D., & Chaminade, T. (2012). Stone tools, lan-
guage and the brain in human evolution. Philoso-
phical Transactions of the Royal Society,367B,7587.
Wei, L. (2013). Code-switching. In R. Bayley, R.
Cameron, & C. Lucas (Eds.), The Oxford handbook
of sociolinguistics (pp. 362377). Oxford: Oxford
University Press.
530 GREEN AND ABUTALEBI
... In order to successfully mitigate parallel activation and to ultimately select the appropriate target language, multilinguals must employ a language control mechanism on the non-target language (Abutalebi & Green, 2007;Christoffels, Firk, & Schiller, 2007;Costa & Santesteban, 2004;Declerck, Koch, Duñabeitia, Grainger, & Stephan, 2019;Green, 1998). Here, language control is conceptualised as a collection of control mechanisms applied to multilingual speech production and comprehension (Abutalebi, 2008;Green & Abutalebi, 2013). From a theoretical point of view, this notion is featured in Green's (1998) Inhibitory Control (IC) model of language control, which postulates that the non-target language needs to be suppressed prior to the linguistic output. ...
... Inhibitory control is an executive function used to regulate and inhibit irrelevant information with respect to thoughts or behaviour, as well as switching attention (Diamond, 2013;Miyake, Friedman, Emerson, Witzki, Howerter, & Wager, 2000). Some studies indicate that language control impacts executive functionsfor example, inhibitory control (Bialystok, 2010;Bialystok, Craik, Klein, & Viswanathan, 2004a;Green & Abutalebi, 2013;Kroll & Bialystok, 2013;Miyake et al., 2000;Wiseheart, Viswanathan, & Bialystok, 2016). Critically, evidence further suggests that language control may share some underlying processing mechanisms with inhibitory control (Declerck et al., 2021;Green, 1998;Linck, Hoshino, & Kroll, 2005;Weissberger, Gollan, Bondi, Clark, & Wierenga, 2015), although this notion is still debated (Branzi, Della Rosa, Canini, Costa, & Abutalebi, 2016;Calabria, Hernandez, Branzi, & Costa, 2012). ...
... A second interpretation of our findings could be that managing cross-language interference between two typologically similar languages does not directly transfer to strengthening the networks underlying inhibitory control. While we know that speaking multiple languages has a direct impact on language control (Coderre & Van Heuven, 2014;Coderre et al., 2013;Green, 1998;Green & Abutalebi, 2013;Mosca, 2019), this may not generalise to broader executive functions such as inhibitory control. Contrary to the predictions by the CRM (Stocco et al., 2014), it may be the case that speaking typologically similar languages does not result in a quantitative difference in the amount of training of executive functions over time compared to typologically dissimilar languages. ...
Article
Both inhibitory control and typological similarity between two languages feature frequently in current research on multilingual cognitive processing mechanisms. Yet, the modulatory effect of speaking two typologically highly similar languages on inhibitory control performance remains largely unexplored. However, this is a critical issue because it speaks directly to the organisation of the multilingual's cognitive architecture. In this study, we examined the influence of typological similarity on inhibitory control performance via a spatial Stroop paradigm in native Italian and native Dutch late learners of Spanish. Contrary to our hypothesis, we did not find evidence for a differential Stroop effect size for the typologically similar group (Italian–Spanish) compared to the typologically dissimilar group (Dutch–Spanish). Our results therefore suggest a limited influence of typological similarity on inhibitory control performance. The study has critical implications for characterising inhibitory control processes in multilinguals.
... Other studies have argued that it is the heterogeneity of the bilinguals' linguistic backgrounds that have contributed to the controversies surrounding bilinguals' advantage in EF, citing several factors: the varying definitions of bilingualism (simultaneous vs. sequential), language proficiency (less vs. more proficient), the context in which languages are used (single vs. dual language context, e.g., workplace or home, with same or different speakers), age of L2 acquisition (early vs. late), differences in participants' backgrounds (i.e. education, socioeconomic status, immigrant and cultural background), to task characteristics (verbal vs. nonverbal) as well as sample size effects and variability in power (e.g., Chamorro & Janke, 2020;Green & Abutalebi, 2013;Grundy, 2020;Leivada et al., 2021;Valian, 2015). ...
... One variable emphasized by de Bruin (2019) for which we did not control is the interactional context of conversational exchanges. It has been argued that language-control abilities, and therefore cognitive-control abilities, are better in bilinguals who use both their languages in a duallanguage context (i.e., both at work and at home, where they have to constantly monitor which of their languages is more appropriate to use; Green & Abutalebi, 2013). For bilinguals whose language interaction is restricted to a single-language context (i.e., one language is used at work and the other one at home), the demands of monitoring and inhibiting the appropriate language may not be so high. ...
Article
Full-text available
Among the factors argued to contribute to a bilingual advantage in executive function (EF), the combination of languages spoken by the bilingual is often overlooked. In this study, we explored the role of language similarity on memory and EF task by comparing performance of three groups of young adults—Hungarian–Serbian and Slovak–Serbian early balanced bilinguals, and Serbian-speaking monolinguals. Slovak is typologically related to Serbian, which are both Slavic, in contrast to Hungarian, which is Finno–Ugric. On the computerized tasks from the CANTAB battery (CANTAB Cognition, 2016, www.cantab.com), differences between monolinguals and bilinguals emerged on the EF tasks: Stockings of Cambridge (SOC) and Attentional Set Shifting (AST), but not the memory tasks: Delayed Matching to Sample (DMS), Paired Associate Learning (PAL), Spatial Working Memory (SWM). Both Hungarian–Serbian and Slovak–Serbian bilinguals outperformed the monolinguals on the more difficult SOC tasks, solved using more than a minimally required number of moves. This is in line with reports that bilinguals perform better under more complex conditions that require more monitoring and switching. However, bilinguals speaking Hungarian and Serbian spent less time preparing to execute the simpler SOC tasks, which can be solved in a minimum of two or three moves; they also exhibited reduced local switching cost and were faster overall on AST than both the Slovak–Serbian bilinguals and Serbian monolinguals. The advantageous performance of speakers of the typologically unrelated languages in our study suggests that these bilinguals may have more efficient attention switching and inhibition systems than bilinguals who speak typologically similar languages.
... Researchers have predominantly used raw scores (percentage exposure, rating on the Likert scale) or normalized scores as an indicator of language usage to predict cognitive performance. Estimating the frequency with which each language is used daily is difficult, but it is even more complicated when bilinguals vary the use of a particular language depending on the context of language usage (Grosjean, 1998;Green and Abutalebi, 2013). To obtain a more comprehensive language usage scores, questionnaires often assess exposure and usage in diverse situations, such as with different interlocutors (e.g., family, friends), at different stages of life (e.g., primary school, high school), and topics (e.g., emotions, leisure activities, media). ...
... To obtain a more comprehensive language usage scores, questionnaires often assess exposure and usage in diverse situations, such as with different interlocutors (e.g., family, friends), at different stages of life (e.g., primary school, high school), and topics (e.g., emotions, leisure activities, media). In addition, studies have categorized bilingual participants under three interaction contexts: Singlelanguage contexts, dual-language contexts, and dense codeswitching contexts based on the interaction of language usage frequencies in various contexts (Green and Abutalebi, 2013;Lai and O'Brien, 2020). Rodriguez-Fornells et al. (2012) developed the Bilingual Switching Questionnaire (BSWQ) to assess better language context, which examines multiple aspects of codeswitching. ...
Article
Full-text available
A better understanding and more reliable classification of bilinguals has been progressively achieved through the fine-tuning methodology and simultaneously optimizing the measurement tools. However, the current understanding is far from generalization to a larger population varying in different measures of bilingualism—L2 Age of acquisition (L2 AOA), L2 usage and exposure, and L2 proficiency. More recent studies have highlighted the importance of modeling bilingualism as a continuous variable. An in-depth look at the role of bilingualism, comparing groups, may be considered a reductionist approach, i.e., grouping based on one measure of bilingualism (e.g., L2 AOA) may not account for variability in other measures of bilingualism (L2 exposure, L2 use or L2 proficiency, amongst others) within and between groups. Similarly, a multifactorial dimension is associated with cognitive performance, where not all domains of cognition and subcomponents are equally influenced by bilingualism. In addition, socio-cultural and demographical factors may add another dimension to the impact of bilingualism on cognitive performance, especially in older adults. Nevertheless, not many studies have controlled or used the multiple socio-cultural and demographical factors as a covariate to understand the role of different aspects of bilingualism that may influence cognitive performance differently. Such an approach would fail to generalize the research findings to a larger group of bilinguals. In the present review paper, we illustrate that considering a multifactorial approach to different dimensions of bilingual study may lead to a better understanding of the role of bilingualism on cognitive performance. With the evolution of various fine-tuned methodological approaches, there is a greater need to study variability in bilingual profiles that can help generalize the result universally.
... Even though the suppression of this competition is mostly an unconscious process, this monitoring of two languages calls on processes that are related to cognitive control, in particular executive functioning processes. This would in turn enhance bilinguals' executive functioning (Green, 1998;Green & Abutalebi, 2013). ...
Chapter
In this study, it was investigated whether child Dutch-English bilinguals, Dutch children learning English as a second language in schools from kindergarten onwards (early-English education), and Dutch (functionally) mon-olinguals differ in their Dutch or English language development and / or their executive functioning development (switching and working memory). In addition , the relation between language balance and executive functioning performance in bilingual children was examined. Participants were children of three age groups: 4-5-year-olds, 8-9-year-olds, and 11-12-year-olds. Executive functions (switching and working memory) and receptive Dutch and English vocabulary were measured. Bilingual children had larger English vocabularies and more balanced Dutch-English lexicons than early-English pupils. Early-English pupils did not have greater English vocabulary (except for those in the 11-12-year-old group) or more balanced lexicons (except for 4-5-year-olds) than their monolingual peers. Bilingual and monolingual children of all three age groups outperformed their age-matched peers from early-English on verbal working memory. For switching, no differences between the groups were found. For the bilingual group, lexical balance, which was computed by dividing the English vocabulary scores by the Dutch vocabulary scores, was related to verbal working memory. The results of this research provide evidence that not only group differences in bilingual status may play a role in executive functioning development, but that individual differences in language development do, too.
... Overall, shorter response latencies in dual-language contexts would reflect enhanced conflict monitoring for congruent and incongruent trials. This is also in line with the adaptive changes in language control processes (Adaptive Control hypothesis, Green & Abutalebi, 2013). Furthermore, we do not predict enhanced inhibition (i.e., flanker effect indexed by a smaller difference between incongruent and congruent trials) in a dual-language context as compared to a single-language context. ...
Article
The relationship between bilingual language control and executive control is debated. The present study investigated the effect of short‐term language switching in a comprehension task on executive control performance in unbalanced bilinguals. Participants were required to perform a context task and an executive control task (i.e., flanker task) in sequence. A picture‐word matching task created different language contexts in Experiment 1 (i.e., L1, L2, and dual‐language contexts). By modifying the color‐shape switching task, we created different contexts that do not involve language processing in Experiment 2 (i.e., color, shape, and dual context). Experiment 1 showed overall faster responses (in both congruent and incongruent trials) in the flanker task after a language switching context than after single (L1 or L2) contexts. This suggests that the language switching in a comprehension task affected general monitoring performance. By contrast, the nonlinguistic contexts in Experiment 2 did not affect flanker performance. This provides further evidence for the crucial role of language processing during switching to elicit short‐term adaptions on domain‐general conflict monitoring. Overall, our findings add to the previous studies by showing cross‐talk between bilingual language control and domain‐general conflict monitoring when language switching occurs in a comprehension task.
... This is explained by spreading activation across the language-independent conceptual store which activates words in both languages (Collins & Loftus, 1975). Bilinguals have more options to consider and language control mechanisms are needed to select the word in the language chosen for communication (Green & Abutalebi, 2013). ...
Article
Lexical selection is a key process in any language-based communicative event, but in translation it occurs in the semantic network activated by two languages. The question asked in this article is how the direction in which translation proceeds affects the process and outcome of lexical selection by experienced bidirectional translators. The prediction from the available empirical evidence that lexical selection when translating into the translator’s L2 (learned language) is more cognitively demanding than when working into L1 (native language) is tested in an experimental study with translators who regularly translate into their L1 (Polish) and L2 (English). The participants performed verbal fluency tasks and translated two texts (a product description text and a film review) into their L1 and L2 (four texts in total). The entire process was recorded by key-logging, eye-tracking and screen capture programs. The results confirm that lexical selection is more demanding and less successful in L1>L2 translation, thus confirming the L2 cognitive disadvantage. Equipping translation students with effective error-preventing strategies and encouraging collaboration between translators and proofreaders could optimise lexical selection in L1>L2 translation.
... The bottom-up approach is closer to reality than the top-down approach, as code-switching is performed voluntarily and not because bilinguals are cued to do so. The dense codeswitching poses a low demand on cognition as the speakers retrieve the most available language without needing to inhibit the unwanted one (Green and Abutalebi 2013). In short, voluntary code-switching in production does not require much cognitive effort. ...
Article
Full-text available
Previous studies on the comprehension of code-switched sentences often neglected the code-switching habit of the specific community, so that the processing difficulty might not have resulted from the change in language but from unnatural switching. This study explores the processing cost of habitual and nonhabitual code-switching. Thirty-one young adults participated in the sentence-reading task with their eye movement tracked. A two-by-two factorial design was used, with Habit (habitual/nonhabitual) and Language (unilingual/code-switched) as the factors. The main effect of Language was observed only in First Fixation Duration, suggesting that the language membership was already identified in an early processing stage. However, for habitual switches, no switching cost in overall processing effort was found, as reflected by Total Fixation Duration and Visit Counts. Our results indicate that the cognitive load was only larger when the switch occurred nonhabitually, regardless of the language membership. In light of this finding, we propose that habitual code-switching might promote the formation of bilingual collocations, or prefabs, which are then integrated into the mental lexicon of the dominant language. Despite a conscious language tag of a foreign origin, these bilingual prefabs are not processed as a language switch in the lexicon. View Full-Text
... However, differences in GMV in early bilingual adults also exist in brain regions outside of those associated with language, such as right parietal areas (Mechelli et al., 2004;Olulade et al., 2016) and bilateral frontal areas (Olulade et al., 2016) known to subserve executive control (EC). These observations are fitting with the notion that in addition to conferring the benefits of speaking two languages, bilingualism heightens some aspects of EC (Bialystok, 1999;Peal & Lambert, 1962) as a consequence of having to constantly suppress and select between the two languages (Green & Abutalebi, 2013). ...
Article
Full-text available
Gray matter has been shown to be greater in early bilingual adults relative to monolingual adults in regions associated with language (Mechelli et al., 2004), and executive control (EC; Olulade et al., 2016). It is not known, however, if language experience-dependent differences in gray matter volume (GMV) exist in children. Further, any such differences are likely not to be the same as those observed in early bilingual adults, as children have had relatively shorter duration of dual-language exposure and/or less development of brain regions serving EC. We tested these predictions by comparing GMV in Spanish-English early bilingual and English monolingual children, and Spanish-English early bilingual and English monolingual adults (n = 122). Comparing only children revealed relatively more GMV in the bilinguals in bilateral frontal, right inferior frontal, and right superior parietal cortices (regions associated with EC). Bilinguals, however, had less GMV in left inferior parietal cortex (region associated with language). An ANOVA including these children with bilingual and monolingual adults revealed interactions of Language Background by Age Group. There were no regions of more GMV in bilinguals relative to monolinguals that were less pronounced in children than adults, despite the children's shorter dual-language experience. There were relative differences between bilingual and monolingual children that were more pronounced than those in adults in left precentral gyrus and right superior parietal lobule (close to, but not directly in areas associated with EC). Together, early bilingual children manifest relative differences in GMV, and, surprisingly, these do not diverge much from those observed in studies of bilingual adults.
Article
In individuals who know more than one language, the languages are always active to some degree. This has consequences for language processing, but bilinguals rarely make mistakes in language selection. A prevailing explanation is that bilingualism is supported by strong cognitive control abilities, developed through long-term practice with managing multiple languages and spilling over into more general executive functions. However, not all bilinguals are the same, and not all contexts for bilingualism provide the same support for control and regulation abilities. This paper reviews research on hearing sign–speech bimodal bilinguals who have a unique ability to use and comprehend their two languages at the same time. We discuss the role of this research in re-examining the role of cognitive control in bilingual language regulation, focusing on how results from bimodal bilingualism research relate to recent findings emphasizing the correlation of control abilities with a bilingual’s contexts of language use. Most bimodal bilingualism research has involved individuals in highly English-dominant language contexts. We offer a critical examination of how existing bimodal bilingualism findings have been interpreted, discuss the value of broadening the scope of this research and identify long-standing questions about bilingualism and L2 learning which might benefit from this perspective.
Article
Full-text available
Bilingualism imposes additional requirements on the cognitive system. As such, it can be a driving force of neuroplastic changes in the brain of a person who speaks more than one language. The need to store and use two systems of representations corresponding to the two languages as well as to develop an efficient control system which allows to use the intendent, contextual-ly appropriate language, may result in both functional and structural changes. Neuroimaging studies show that the neural organization of language representations in a bilingual brain depends to a large degree on the type of representation. Conceptual representations seem to share common neural underpinnings between the different languages. Lexical representations, related to the vocabulary and words, are processed by the same brain regions regardless of the language, however , they have been shown to be coded by distinct neu-ronal populations. Finally, neuroplastic reorganization of syntactic representations is highly dependent on factors related to individual experiences of bilingualism, such as age of acquisition and proficiency in the second language. Neuroplastic changes in the bilingual brain have also been linked to the increased demands that using two languages imposes on the cognitive control mechanisms. Both structural and functional changes in the brain of bilinguals were observed withing a wide network referred to as language control network. Summing up, neuroimaging studies have demonstrated that bilingualism is related to neuroplastic changes in both structure and functioning of the brain. However, the newest reports suggest the extent and intensity of the neuroplastic changes are most likely dependent on individual experiences of each bilingual speaker.
Article
Full-text available
This study investigated the role of domain-general inhibitory control in trilingual speech production. Taking an individual differences approach, we examined the relationship between performance on a non-linguistic measure of inhibitory control (the Simon task) and a multilingual language switching task for a group of fifty-six native English (L1) speakers learning French (L2) and Spanish (L3). Better inhibitory control was related to reduced switch costs, but only when switching into or out of the more dominant L1, where inhibitory control has been theorized to be most important (Green, 1998). The results provide evidence of a direct link between inhibitory control abilities and language switching capabilities, and suggest constraints on the conditions under which a domain-general inhibitory control mechanism supports language switching.
Article
Full-text available
A key question in bilingual language production research is how bilingual individuals control the use of their two languages. The psycholinguistic literature concerning language control is unresolved. It is a matter of controversy whether (a) issues to do with control are central to understanding bilingual language processing; and (b) if they are, what is the site or sites of control; and (c) whether language control in bilinguals relies upon inhibitory mechanisms.One way to deepen our understanding of language control is to consider the implications from research on functional neuroimaging. In the present paper, we illustrate that neuroimaging research shows that bilinguals engage cognitive control networks for achieving tasks such as language switching. The neural evidence points to multiple neural regions of control that may rely upon an inhibitory mechanism. These ‘brain data’ may, in turn, stimulate the development of neurocognitive accounts of bilingual language processing.
Article
Full-text available
This paper builds on an earlier one (Gardner-Chloros & Edwards, 2004), where we pointed to various problems underlying the search for grammatical constraints on codeswitching (CS), these problems being largely derived from the assumption that CS is the sum of two finite and discrete systems which together make up the competence of the bilingual speaker. Here we look more closely at the type of grammatical knowledge involved in intrasentential CS, by examining a widely attested structure in CS, bilingual compound verbs (BCVs). We consider the properties of compound verbs in general, including in monolingual contexts, their significance in the CS literature, notably for the notion of Matrix Language, and their implications for an understanding of bilingual grammars, broadly defined. We conclude that this widespread feature of bilingual speech is more than a convenient frame for inserting foreign words, and should be seen as part of a more general verb-formation process. The types of linguistic knowledge which allow such formations to emerge are based on what Sebba (1998) has called conceptual work, rather than on the simple interaction of two linguistic systems as such. We list a number of factors to be taken into account in trying to identify a CS grammar, within which idiolect, metalinguistic knowledge and community norms are particularly significant.
Article
A series of three experiments is reported that have examined how the specialist components of the working memory model (Baddeley, 1986; Baddeley & Hitch, 1974) are involved during the performance of mental synthesis. The results indicate that the spatial manipulation of material during synthesis draws on the resources of spatial working memory, whereas the maintenance of verbal representations can play an important role in making effective use of cognitive resources during the task as a whole. These results are interpreted within the framework of a revised model of visuo-spatial working memory, in which the specialist components act as temporary storage buffers for visuo-spatial and verbal material, whereas the generation and maintenance of conscious visual images is primarily seen as being a function of the central executive component.
This paper aims to accomplish two things: first, to develop precise criteria to establish profiles for bilingual speech, following the typology of insertion, alternation and congruent lexicalisation developed in Muysken (2000); and second, to test these criteria on specific data sets. A first set involves Welsh–English bilingual data analysed by Deuchar, a second set comprises Tsou–Mandarin Chinese data collected and analysed by Sung-Lan Wang, and a third set involves Taiwanese–Mandarin Chinese data, also collected and analysed by Sung-Lan Wang. We conclude that it is indeed possible to establish more precise quantitative profiles which capture the intuition that different data sets show different codeswitching properties, but that there are a number of conceptual and methodological issues that require further investigation.
Article
This paper aims to accomplish two things: first, to develop precise criteria to establish profiles for bilingual speech, following the typology of insertion, alternation and congruent lexicalisation developed in Muysken (2000); and second, to test these criteria on specific data sets. A first set involves WelshEnglish bilingual data analysed by Deuchar, a second set comprises TsouMandarin Chinese data collected and analysed by Sung-Lan Wang, and a third set involves Taiwanese Mandarin Chinese data, also collected and analysed by Sung-Lan Wang. We conclude that it is indeed possible to establish more precise quantitative profiles which capture the intuition that different data sets show different codeswitching properties, but that there are a number of conceptual and methodological issues that require further investigation.
Article
Growing evidence shows that executive functioning benefits from bilingual experience. However, the nature of the mechanisms underlying this advantage remains to be clarified. While some have put forward single process accounts to explain the superior performance of bilinguals relative to monolinguals in executive control tasks, recent findings have been interpreted by considering the dynamic combination of monitoring and inhibitory processes to overcome interference from distracter information. In the present study we explored this idea by comparing monolinguals and highly proficient bilinguals in the AX-CPT. This task requires individuals to adjust proactive (monitoring) and reactive (inhibition) control to achieve efficient performance. We also examined the extent to which a well-known index of inhibitory capacity, the stop-signal reaction time, predicts accuracy in the AX-CPT. Results showed that bilinguals outperformed monolinguals in the experimental condition where higher requirement of proactive-reactive control adjustment was required. Interestingly, the inhibition index predicted errors in this condition only in the sample of bilinguals. These findings suggest that a better understanding of the cognitive benefits of bilingualism may require consideration of how bilinguals adjust different executive control mechanisms to cope with interference.