Elsevier

NeuroImage

Volume 174, 1 July 2018, Pages 44-56
NeuroImage

The bilingual language network: Differential involvement of anterior cingulate, basal ganglia and prefrontal cortex in preparation, monitoring, and execution

https://doi.org/10.1016/j.neuroimage.2018.02.010Get rights and content

Highlights

  • Distinct networks were observed during language preparation and execution.

  • The ACC was only active when processing cues about which language to use.

  • The BG were consistently active across all phases of bilingual language control.

  • The DLPFC activity gradually increased from preparation to execution.

Abstract

Research on the neural bases of bilingual language control has largely overlooked the role of preparatory processes, which are central to cognitive control. Additionally, little is known about how the processes involved in global language selection may differ from those involved in the selection of words and morpho-syntactic rules for manipulating them. These processes were examined separately in an fMRI experiment, with an emphasis on understanding how and when general cognitive control regions become activated. Results of region-of-interest analyses on 23 early Spanish-English bilinguals showed that the anterior cingulate cortex (ACC) was primarily engaged during the language preparation phase of the task, whereas the left prefrontal (DLPFC) and pre-supplementary motor areas showed increasing activation from preparation to execution. Activation in the basal ganglia (BG), left middle temporal lobe, and right precentral cortical regions did not significantly differ throughout the task. These results suggest that three core cognitive control regions, the ACC, DLPFC, and BG, which have been previously implicated in bilingual language control, engage in distinct neurocognitive processes. Specifically, the results are consistent with the view that the BG “keep track” of the target language in use throughout various levels of language selection, that the ACC is particularly important for top-down target language preparation, and that the left prefrontal cortex is increasingly involved in selection processes from preparation through task execution.

Introduction

Bilingual language control refers to the set of mechanisms used for the selection and maintenance of a target language in the face of competing symbolic word representations and morpho-syntactic rules for manipulating them (Costa et al., 1999; Hatzidaki et al., 2011). Such control is likely underpinned by multiple processes, including the selection of the language to use at a given situation, the generation of linguistic goals (e.g., pluralizing a word based on the target language) and the selection of word forms and rules for manipulating words to achieve the goal (e.g., Guo et al., 2011; Branzi et al., 2015; Hoversten et al., 2015). As a result, bilingual language control likely involves multiple sub-component neurocomputations deployed across various situations (e.g., speaking a foreign language continuously while abroad vs. translating between individuals) and applied to different levels of selection (e.g., the need to speak in the Spanish language vs. the need to conjugate the verb "hablar" in Spanish). Many of these intricacies are yet to be addressed in the bilingual language control literature. The current study aims to advance understanding of the neurocognitive mechanisms of bilingual language control.

The existing body of literature investigating the neural underpinnings of bilingual language use has widely implicated three regions known to be more broadly involved in cognitive control: the dorsolateral prefrontal cortex (DLPFC), the basal ganglia (BG), and the anterior cingulate cortex (ACC). In the first fMRI investigation of bilingual language switching, Hernandez et al. (2000) used a picture-naming paradigm in which the target language either switched between Spanish and English or remained stable in either language within a block. The results showed that activation in the left DLPFC increased in the switching condition where competition for selection between two available languages became maximized. In a series of follow-up studies, Hernandez and colleagues replicated and extended their original findings, showing repeatedly that the DLPFC is specifically engaged when bilinguals are asked to switch between target languages as opposed to maintaining a particular language (Hernandez et al., 2001; Hernandez, 2009). This is consistent with the broad body of literature implicating the DLPFC in cognitive control. Specifically, when tasks involve response conflict of some kind (e.g., Mansouri et al., 2009) the DLPFC is involved in goal maintenance or storing a set of rules for behaving given specific conditions (Miller and Cohen, 2001; Wallis et al., 2001; Cole et al., 2010; Becker et al., 2016).

In parallel, evidence from neuropsychological (e.g., Abutalebi et al., 2000; Fabbro, 2001), neurosurgical (Robles et al., 2005), and neuroimaging (e.g., Crinion et al., 2006; Lehtonen et al., 2005) studies has implicated the BG, and particularly the caudate nucleus, in bilingual language control. The BG are a set of subcortical nuclei composed of the subthalamic nucleus, the substantia nigra, the external and internal segments of the globus pallidus, and the striatum. The striatum consists of the caudate and putamen, and serves as the input station of the circuit. The BG receive inputs from the entire cortex and modulate signals to prefrontal regions (including both DLPFC and ACC) in a manner well-suited for dynamically reprioritizing responses (e.g., Stocco et al., 2014; Stocco et al., 2010).

Importantly for cognitive control, the BG are rich in dopamine, and thus have been associated with cognitive flexibility more so than the DLPFC (Pasupathy and Miller, 2005). Based on modeling work demonstrating “Conditional Routing” of signals to the prefrontal cortex through the BG (Stocco et al., 2010), Stocco et al. (2010) proposed a shared role for the BG, and the striatal nuclei in particular, in bilingual language control. According to the model, the BG actively mediate signaling to the prefrontal cortex according to the dynamically changing target language being used by a bilingual at any given time.

In their theoretical review paper, Abutalebi and Green (2007) discuss research on bilingual language production under the lens of general cognitive control mechanisms. This review and subsequent refinements from the group (Green and Abutalebi, 2013; Abutalebi and Green, 2016) included an important role for the ACC, which is generally characterized as a region that detects or monitors conflict (e.g., Botvinick et al., 1999; Kerns et al., 2004), as well as for the DLPFC and BG. Specifically, they proposed that controlled language production in bilingual individuals involves the dynamic interplay between conflict monitoring in the ACC, executive functioning (including response selection and inhibition) in the DLPFC, language planning, selection, and switching executed by the BG, and maintenance of representations in working memory in the parietal lobe. In a subsequent neuroimaging investigation, Abutalebi and colleagues confirmed that the ACC and BG were involved in monitoring target language during a bilingual picture-naming task (Abutalebi et al., 2007). They continued to demonstrate that consistent conflict monitoring in bilingual individuals shapes the ACC both structurally and functionally in a way that gives rise to more efficient processing of conflict in non-linguistic tasks as well (Abutalebi et al., 2011).

In summary, research on the neural basis of bilingualism has repeatedly implicated the DLPFC, BG, and ACC in bilingual language control. The goal of the current study was to understand the role of bilingual language control regions in different phases of bilingual language control processing.

Cognitive control research has identified two classes of control mechanisms: Proactive control, which is deployed early, and typically makes use of predictive cues to guide information processing in a top-down and goal-oriented manner; and reactive control, which is largely driven by bottom-up processes that trigger a corrective function following unanticipated detection of conflict (e.g., Braver et al., 2007; Braver, 2012). The role of proactive control in bilingual language use has been largely ignored. For example, in a recent meta-analysis of the neural networks supporting bilingual language control (Luk et al., 2012), none of the ten experiments employed paradigms in which language preparation could be investigated separately from language use. Interestingly, this meta-analysis did not find significant activation in the ACC across experiments. In the real world, however, bilinguals likely use predictive cues about which language they should speak, whether it be broad contextual cues such as the location (e.g., at home versus at work), previous experience with the individual they are speaking to, or more subtle (and certainly less predictable) cues such as the ethnicity of a person they are about to interact with.

One recent experiment by Woumans et al. (2015) investigated the cognitive effect of preparatory processes on bilingual language control by training participants with faces that were reliably associated with particular language profiles. Each face was presented 2,000 milliseconds before a speech event. Certain faces were reliably followed by speech in one language, while other “bilingual” faces were followed by speech in two languages. When given a noun in either language, participants were able to more rapidly produce associated verbs when a familiar face, regularly associated with speech in one particular language, served as the preparatory cue for a trial. In contrast, participants experienced more difficulty with the task when either an unfamiliar face, or a familiar bilingual face preceded the trial. These results are consistent with research on general cognitive control which has shown that predictive cues enable proactive adjustment for the desired subsequent task (Braver, 2012; Sohn and Carlson, 2000; Ruge et al., 2013; Zhang et al., 2013).

To the best of our knowledge, only one neuroimaging investigation to date has measured the role of preparatory cuing in bilingual language control. Reverberi et al. (2015) presented an abstract cue indicating “target language” in advance of a to-be-named picture. They found that language switching during preparation resulted in activation in the left middle temporal gyrus, right parietal lobe, and bilateral precuneus. In contrast, during task execution, the medial prefrontal cortex was more highly activated when target language switched than when it was repeated. Thus, when task preparation and execution were separately examined in a naming task, different sets of regions were implicated in different phases. The current paper aims to extend the existing research by investigating the neural mechanisms associated with preparatory cuing during a novel bilingual language task.

A second limitation of the existing bilingual control literature is that the majority of it has been limited to lexico-semantic selection processes, most commonly operationalized through picture-naming or picture-word-matching tasks. While lexical selection in the face of competing representations is clearly one of the demands placed on a bilingual language control system, such selection also occurs in morpho-syntactic processing. To the best of our knowledge, none of the switching paradigms typically used to study bilingual language control has included morpho-syntactic manipulations. This is important to consider, however, as research has shown that co-activation of linguistic information in the bilingual brain is not limited to the lexico-semantic level (Pickering and Ferreira, 2008; Hatzidaki et al., 2011). Intersentential codeswitching and cross-linguistic structural priming provide additional evidence that the need to manage interference between languages extends to morpho-syntactic levels (Pickering and Ferreira, 2008).

A third, but less pervasive, limitation of the existing bilingual control literature is that it is difficult to separate top-down linguistic control processes from any bottom-up influences that are driven by stimuli occurring in a particular language. It is likely that hearing or reading a word in a particular language will prime other related words in the same language and the rules for manipulating words in that particular language. Often in bilingual language control experiments, the target language itself is used to indicate which language a particular task should be executed in. For example, in their seminal bilingual language switching investigation, Hernandez et al. (2001) used the words “say” and “diga” presented simultaneously with a picture to instruct participants which language to name an object in. Similarly, in a Chinese-English switching task, Wang et al. (2009) used either a Chinese character “读” or the English counterpart ‘read’ to indicate which language a trial should be executed in. In the real world, bilingual language control, like any other cognitive control, involves dynamic interactions between top-down planning and attention allocation and bottom-up, stimulus driven biasing of information. However, including both types of information in a paradigm makes it complicated to understand what is driving behavior and brain activation. Therefore, the current paradigm uses non-linguistic, symbolic cues to attempt to isolate stimulus-driven effects from preparatory processes.

The current paper addresses these three limitations through the development of a novel paradigm that allows: (1) the separation of preparatory processes from task execution, (2) the investigation of bilinguals' morpho-syntactic rule selection and application, and (3) the separation of top-down control structures from bottom-up linguistic influences. Specifically, we employed a variant of the Rapid Instructed Task Learning (RITL) paradigm, which is gaining popularity as a tool for understanding how the human brain executes rule-based behaviors (Stocco et al. 2012; Cole et al. 2013; Stocco and Prat, 2014). One critical feature of the RITL paradigm is that the rules for completing a subsequent task are presented before the stimuli on which the rules need to be applied. This allows one to estimate the neural processes involved in dynamically constructing a mental program for controlled behavior separately from the execution of that behavior. In the current experiment, this design feature provides the ability to separate preparation from execution in bilingual language control, and to separate the top-down processes associated with generating a control structure from the bottom-up influences of linguistic stimuli. This separation allows one to investigate the control mechanism(s) that is established to perform a task without influence from the stimuli.

Our predictions focus on three questions central to bilingual language control research that can be extracted using this novel research design: (1) How and when do general cognitive control regions participate in bilingual language control? (2) How do the neural networks involved in proactive control differ from those involved in bilingual language production? and (3) How do the neural networks involved in global language selection differ from those involved in morpho-syntactic rule selection?

To address these questions, the analyses conducted herein will be centered primarily upon the regions identified in a quantitative meta-analysis on the neurobiology of bilingual language control (Luk et al., 2012). Specifically, Luk et al. (2012) identified a network of brain regions that have been consistently reported in investigations of bilingual language control. This network includes regions discussed herein that are typically associated more generally with cognitive control including: DLPFC, BG (specifically bilateral caudate nuclei), and pre-supplementary motor areas (pre-SMA), also discussed by Abutalebi and Green as part of the ACC conflict monitoring network (Abutalebi and Green, 2016). Not surprisingly, the results also included regions more broadly associated with language processes such as the left inferior frontal gyrus (BA 44, 47) and the left middle temporal gyrus (BA 37). However, as our task does vary significantly from those reported in the meta-analysis, we also conducted exploratory, voxelwise, whole-brain analyses.

To address the first question, “How and when do general cognitive control regions participate in bilingual language control?” our primary analyses will investigate how activation changes in these regions of interest, defined a priori based on the meta-analysis, across the three task phases. Based on the Conditional Signal Routing Theory (Stocco et al., 2010; Stocco et al., 2014) and the results of Crinion et al. (2006), we predict that the BG will be consistently involved in tracking the target language throughout the different phases of a single trial. In contrast, we predict that brain activity in the DLPFC will increase as the trial progresses from language preparation to rule execution, due to increases in working memory demands. Finally, we predict that the ACC will be most highly activated during task execution, as it is believed to work in concert with the Pre-SMA and DLPFC to address response conflict (Abutalebi and Green, 2007, 2016; Becker et al., 2016).

To address the second question, “How do the neural networks involved in proactive control differ from those involved in bilingual language production?” we will compare patterns of whole-brain activation during language preparation to those obtained during task execution. Based on the research reported by Reverberi et al. (2015), we predict that posterior regions including the middle temporal gyrus, parietal lobes, and precuneus will be more active during preparation, whereas the medial frontal gyrus and all of the canonical left hemisphere language processing regions (e.g., Broca's area in inferior frontal gyrus and Wernicke's area in superior-posterior temporal gyrus) will be more active during execution, which involves the processing of linguistic stimuli. As previously discussed, we also predict that ACC and DLPFC will be more active during task execution.

Finally, to address the third question, “How do the neural networks involved in global language selection differ from those involved in morpho-syntactic rule selection?” we will compare patterns of whole-brain activation during target language preparation to those obtained during morpho-syntactic rule selection. In their review, Buchweitz and Prat (2013) suggested that linguistic rules are organized hierarchically in the bilingual brain, according to target language. Based on research showing that more abstract rules engage more rostral regions of the PFC (Badre, 2008; Koechlin et al., 2003), Buchweitz and Prat (2013) proposed that target language may be represented more rostrally than morphosyntactic rules. Thus, we predict that both selecting target language and selecting morpho-syntactic rules will recruit the left lateral prefrontal cortex (and in particular the inferior frontal gyrus) but that global target language will recruit more rostral PFC regions.

Section snippets

Participants

Twenty-four, right handed Spanish-English bilinguals (20 females, aged 18–31 years) were paid for participation in the current study. Participants were required to be highly proficient in both languages (as assessed through grammatical proficiency tests) and to have learned both languages before the age of seven. Bilingual language experience information is summarized in Table 1.

All participants were healthy, with no history of developmental or neurological disorders. All participants provided

Behavioral results

Accuracy rates to the response verification probes were high across participants (M = 88.5%, SEM = 1.76%). Response times for correct trials were analyzed using a one-way repeated measures analysis of variance (ANOVA) with three within-participants task phases: Prepare Target Language (mean = 1547.57 ms, sem = 34.93), Select Rule (mean = 2726.83 ms sem = 47.69, and Execute (mean = 2953.33 ms, sem = 49.60). A significant main effect of task phase was revealed (F (2,66) = 57.570, p < 0.0001).

Discussion

The results from this experiment highlight the complexity of the information processing demands associated with bilingual language control. Specifically, they show that when top-down preparatory processes are investigated prior to linguistic task execution, and in the absence of bottom-up linguistic cues, distinct patterns of involvement emerge for the three core cognitive control regions (ACC, BG, and DLPFC), as well as from the broader network of regions previously implicated in bilingual

Conclusions and caveats

Up until this point, we have focused on the response profiles of individual regions including the ACC, BG, and DLPFC across the various phases of our bilingual language control task. A mechanistic explanation of how these regions accomplish bilingual language control, however, requires a more complete understanding of how and when these regions interact and influence one another. In a recent dynamic causal modeling analysis, Becker and colleagues investigated such interactions by comparing

Funding

This research was supported by a grant from the Office of Naval Research (ONRBAA13-003) and by a bridge-funding grant awarded to Dr. Chantel Prat from the Department of Psychology at the University of Washington.

Supplementary material

The experimental stimuli and data (including individual de-identified datasets, individual statistical analysis, group-level results, and the MATLAB and shell scripts to reproduce the analyses) are freely available on the Harvard Dataverse: https://doi.org/10.7910/DVN/W6WJVN.

Acknowledgement

The authors would like to thank Justin A. Abernethy, Brianna Yamasaki, and Jose M. Ceballos for help with design, and data collection and Ione Fine for providing feedback on earlier versions of this manuscript.

References (89)

  • J. Duncan

    The multiple-demand (MD) system of the primate brain: mental programs for intelligent behaviour

    Trends Cognit. Sci.

    (2010)
  • F. Fabbro

    The bilingual brain: cerebral representation of languages

    Brain Lang.

    (2001)
  • T. Guo et al.

    Local and global inhibition in bilingual word production: fMRI evidence from Chinese–English bilinguals

    Neuroimage

    (2011)
  • A. Hatzidaki et al.

    Co-activation of syntax in bilingual language production

    Cognit. Psychol.

    (2011)
  • A.E. Hernandez

    Language switching in the bilingual brain: what's next?

    Brain Lang.

    (2009)
  • A.E. Hernandez et al.

    In search of the language switch: an fMRI study of picture naming in Spanish–English bilinguals

    Brain Lang.

    (2000)
  • A.E. Hernandez et al.

    Language switching and language representation in Spanish–English bilinguals: an fMRI study

    Neuroimage

    (2001)
  • A.E. Hernandez et al.

    Beyond the bilingual advantage: the potential role of genes and environment on the development of cognitive control

    J. Neurolinguistics

    (2015)
  • W.J. Van Heuven et al.

    Orthographic neighborhood effects in bilingual word recognition

    J. Mem. Lang.

    (1998)
  • K. Irlbacher et al.

    Mechanisms and neuronal networks involved in reactive and proactive cognitive control of interference in working memory

    Neurosci. Biobehav. Rev.

    (2014)
  • T.L. Luks et al.

    Evidence for anterior Cingulate cortex involvement in monitoring preparatory attentional set

    Neuroimage

    (2002)
  • D.E. Nee et al.

    Neural evidence for a 3-state model of visual short-term memory

    Neuroimage

    (2013)
  • S.D. Newman et al.

    Differential effects of syntactic and semantic processing on the subregions of Broca's area

    Cognit. Brain Res.

    (2003)
  • G. Northoff et al.

    Cortical midline structures and the self

    Trends Cognit. Sci.

    (2004)
  • R.C. Oldfield

    The assessment and analysis of handedness: the Edinburgh inventory

    Neuropsychologia

    (1971)
  • C. Reverberi et al.

    Language control in bilinguals: intention to speak vs. execution of speech

    Brain Lang.

    (2015)
  • A. Stocco et al.

    Bilingualism trains specific brain circuits involved in flexible rule selection and application

    Brain Lang.

    (2014)
  • K.A. Vaughn et al.

    Individual differences in the bilingual brain: the role of language background and DRD2 genotype in verbal and non-verbal cognitive control

    J. Neurolinguistics

    (2016)
  • Y. Wang et al.

    Sustained and transient language control in the bilingual brain

    Neuroimage

    (2009)
  • J. Abutalebi et al.

    Neuroimaging of language control in bilinguals: neural adaptation and reserve

    Biling. Lang. Cognit.

    (2016)
  • J. Abutalebi et al.

    Do subcortical structures control ‘language selection’ in polyglots? Evidence from pathological language mixing

    Neurocase

    (2000)
  • J. Abutalebi et al.

    Language control and lexical competition in bilinguals: an event- related fMRI study

    Cerebr. Cortex

    (2007)
  • J. Abutalebi et al.

    Bilingualism tunes the anterior cingulate cortex for conflict monitoring

    Cerebr. Cortex

    (2011)
  • A.D. Barber et al.

    Cognitive control involved in overcoming prepotent response tendencies and switching between tasks

    Cerebr. Cortex

    (2004)
  • V. Benet-Martínez et al.

    Negotiating biculturalism: cultural frame switching in biculturals with oppositional versus compatible cultural identities

    J. Cross Cult. Psychol.

    (2002)
  • E. Blanco-Elorrieta et al.

    Bilingual language switching in the laboratory versus in the wild: the spatiotemporal dynamics of adaptive language control

    J. Neurosci.

    (2017)
  • M. Botvinick et al.

    Conflict monitoring versus selection-for-action in anterior cingulate cortex

    Nature

    (1999)
  • F.M. Branzi et al.

    Language control in bilinguals: monitoring and response selection

    Cerebr. Cortex

    (2015)
  • T.S. Braver et al.

    Explaining the many varieties of working memory variation: dual mechanisms of cognitive control

    Var. Work. Mem.

    (2007)
  • M. Brett et al.

    Using the Talairach atlas with the MNI template

    Neuroimage

    (2001)
  • J.I. Brown et al.

    Nelson-denny Reading Test: Manual for Scoring and Interpretation Forms G & H

    (1993)
  • C.S. Carter et al.

    Anterior cingulate cortex, error detection, and the online monitoring of performance

    Science

    (1998)
  • C.S. Carter et al.

    The contribution of the anterior cingulate cortex to executive processes in cognition

    Rev. Neurosci.

    (1999)
  • A.E. Cavanna et al.

    The precuneus: a review of its functional anatomy and behavioural correlates

    Brain

    (2006)
  • Cited by (0)

    View full text