This item is only available as the following downloads:
EXECUTIVE CONTROL AND LANGUAGE MODE IN MONOLINGUAL AND BILINGUAL YOUNG ADULTS BY EMMA WARD A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfilment of the requirements for the degree Bachelor of Arts Under the sponsorship of Dr. Heidi Harley Sarasota, Florida M ay, 2011
Running head: EXECUTIVE CONTROL AND LANGUAGE MODE ii Acknowledgments I would like to thank Dr. Harley and Dr. Barton for their guidance throughout my New College career and their abilities to nurture and challenge my curiosity. They taught me how to critique research, how to ask the right questions, and most importantly how to admit that I do not know the answers. I would also like to thank Dr. Cooper for his tireless support and Dr. Labrador Rodriguez for the escape from psychology t hat her classes provided for me, and for the Spanish skills without which I would not even have attempted this thesis. My mum and boyfriend have supported me in ways I could not have hoped for and I appreciate them all the more for listening to my problems and half formed ideas along the way.
EXECUTIVE CONTROL AND LANGUAGE MODE iii Table of Contents Acknowledgements ......................................................................................................... ........... ii Table of Contents ............................................... .............................................................. ......... iii List of Tables and Figures............................................................................................... ............ v Abstract ...................................... .................................................................................. ............. vi Introduction .................................................................................................................. ............. 1 The Attentional System .................................................................................... ............ 2 Conflict Control ................................................................................... ............. 4 Neurological Processes in Conflic t Control ................... ............ ..........7 Cognitive Processes in Non verbal Tasks ................................. ........... 9 Cognitive Processes in Verbal Tasks ....................................... ........... 20 Language Activation ..... ..................................................................................... .......... 26 Age as a Participant Variable ............................................................................. ......... 30 The Current Study ...................... .............................. ......................................................... ....... 31 Method ............................................................................................................................. ....... 33 Participants ..... ............................................................................................................33 Materials and Instruments .........................................................................................33 Procedure ................. ..................................................................................................35 Results ........................................................................................................ ..............................36 Language P roficiency .................................................................... ..............................36 Simon and Flanker Equivalence..................................................................................36 Congruence ................... ................................................................ ..............................38 Cue Type in the Flanker Task ........................................................ ..............................40 Word Type in the LDT .................... ............................................... ..............................41 Discussion .................................................................................................. ..............................43 References ......................... ........................................................................ ..............................49
EXECUTIVE CONTROL AND LANGUAGE MODE iv Appendix A ................................................................................................. ..............................5 3 Appendix B ................................................................................................................... ........... 60 Appendix C ................................................................................................. .................... ..........61 Appendix D................................................................................................ ..............................62
EXECUTIVE CONTROL AND LANGUAGE MODE v List of Tables and Figures Figure 1: The att entional system and its subcomponents ...................... ...................................3 Figure 2: Examples of Simon stimuli and the position response congruence ...........................9 Figure 3: Examples of flanker stimuli with fixation point ........................................................14 Fi gure 4: Cue types presented in the ANT ...............................................................................16 Figure 5: Schematic representation of an incongruent flanker trial with a misleading cue ....34 Table 1: Relationship between reaction ti mes on Simon and flanker tasks ............................37 Table 2: Reaction times of all participants by task and congruence .......... ..............................38 Table 3: Difference between incongruent and congruent trials ................ .......... ....................39 Figure 6: Difference between incongruent and congruent trials ............................... ..............39 Table 4: Reaction times of all participants by cue type on the flanker task ............................ 40 Table 5: Reaction times of all participants by word type on the LDT ...................................... 41 Figure 7: Reaction times of all participants by word type on the LDT ..................................... 42
EXECUTIVE CONTROL AND LANGUAGE MODE vi EXECUTIVE CONTROL AND LANGUAGE MODE IN MONOLINGUAL AND B ILINGUAL YOUNG ADULTS Emma Ward New College of Florida 2011 ABSTRACT Bilingual and monolingual children and adults differ in their performance on tasks that require conflict control, the ability to attend to pertinent information in the face o f distractors but inconsistency in the literature suggests that an additional element is at play some of these tasks containing conflict before and after a manipulation of language mode, designed to increase the salience of the linguistic context and possibly activate or enhance language group differences. The groups did not differ in their responses to cueing, but their responses to stimulus response congruence and word typ es in a lexical decision task (LDT) showed interesting patterns. The bilingual participants showed a decreased congruence effect after the mode manipulation, suggesting that conflict hindered their progress less when their language background was more sali ent while the monolingual participants maintained a large congruence effect. On the LDT, t hose more proficient in their second language responded more slowly than those less proficient before the manipulation but this difference disappeared in the second session possibly a sign of bilingual participants adapting to the task demands after the manipulation
EXECUTIVE CONTROL AND LANGUAGE MODE vii The previous research on bilingual executive control is fraught with inconsistency and the current study does not resolve those problems, but it does p rovide support for the relatively new perspective of language mode. ________________________________ Dr. Heidi Harley Psychology
EXECUTIVE CONTROL AND LANGUAGE MODE 1 Executive Control and Language Mode in Monolingual and Bilingual Young Adults Bilingual people differ fr om monolingual people in many different ways, on tasks diverse enough to suggest far reaching cognitive and social consequences of acquiring two languages at an early age. Bilingual children more often line, appearance reality and alternate image tasks than monolingual children of the same age (Martin Rhee & Bialystok, 2008), perform better on the Dimensional Change Card Sort (DCCS; Bialystok, 1999; Carlson & Meltzoff, 2008) develop theory of mind earlier (Goetz, 2003, as cited in Emmorey, Luk, Pyers, & Bialystok, 2008) and show less performance decline when tasks involve irrelevant but salient cues (Martin Rhee & Bialystok, 2008 ) However, bilinguals also differentiate between phonemes later than monolinguals (Fennel, B yers Heinlein & Werker, 2007) and tend to have smaller vocabularies than monolinguals when measured by standa rdised tests in one language (Martin Rhee & Bialystok, 2008 ; Moreno, Bialystok, Wodniecka & Alain, 2010 ) While not all of the group differences fo und in childhood persist into adulthood, there remain a few domains in which adult bilingual participants differ from their monolingual peers. Bilingual adults differ from monolingual adults in ability to combat proactive interference (Bialystok & Feng, 2009) ; the ir later average age of onset of senile dementia (Bialystok, Craik & Freedman, 2007 as cited in Bialystok & Craik, 2010 ) ; enhanced metacognitive skills, working memory, planning, abstract thinking, rule acquisition and selecting relevant sensory information (Baddeley, 1996 as cited in Adesope et al., 2010 ); and neurological activation patterns for example, the left inferior prefrontal gyrus has been associated with completing a go/no go task (Rodriguez Fornells, et al. 2005 as cited in Wang, Ku hl, Chen, & Dong, 2009 ) and task switching (Garbin et al., 2010) for bilingual people but not for monolingual people
EXECUTIVE CONTROL AND LANGUAGE MODE 2 Although language group differences occur in metacognitive skills (Duncan, 2005; Kemp, 2007 as cited in Adesope, et al., 2010 ), a meta a nalysis suggested that only the youngest participants in a pool of preschool to post secondary students showed a metacognitive advantage (Adesope, et al., 2010) Regarding the working memory differences, the direction of the effect is highly disputed (see Adesope, et al., 2010 for a discussion) and depends on the attentional demands of the task (Adesope et al., 2010) R esearch on bilingual executive control has produced mixed results (see Costa Hernndez, Costa Faidella and Sebastin Galls, 2009 for an overview) and there are many methodological challenges within the field including dispute s over relevant definition s and non systematic differences across studies which impede direct comparison s E ach study lacks a small number of elements and no one study fills the gaps, leaving questions such as, functioning in young adulthood conclusive answers because the studies designed to answer the questi on vary in task type, sociolinguistic environment and definitio n of bilingualism. The current study will address these issues by comparing some of the tasks used in previous literature to assess their functional equivalence, and by measuring bilingualism w ith two different self report measures The Attentional System An area of bilingual cognition that has garnered a lot of interest is the attentional system which consists of three elements: the alerting network, which leads to the heightened state requir ed for rapid stimulus detection; the orienting network, which allocates cognitive resources; and executive control, which is usually defined as involvement (Her nndez, et al., 2010, p. 316; although see Daniels, Toth & Jacoby (2006 as cited in Bialystok, 2007 ) for an account which takes attention to be a component of executive
EXECUTIVE CONTROL AND LANGUAGE MODE 3 control rather than the other way around ) Many authors also further divide executive control into two components: conflict monitoring and conflict resolution (Costa et al., 2008 ). Figure 1: The attentional system and its subcomponents Bilingual adults have been shown, however inconsistently, to have reduced switching costs a nd reduced mixing costs in comparison to monolingual adults. That is, s ome authors claim that in problem solving tasks that sometimes contain distracting information, bilinguals have an advantage over monolinguals in their processing speed of the trials th at do contain information that they must ignore in the form of a smaller difference between reaction times to the two types of trial (Bialystok & Feng, 2009; Colzato, et al. 2008; Costa, Hernndez, & Sebastin Galls, 2008; Emmorey, et al. 2008; Prior & MacWhinney, 2010; for a meta analysis see Adesope, Lavin, Thompson, & Ungerleider, 2010). Another bilingual advantage that has been debated is a reduced overall processing speed relative to monolinguals on tasks which sometimes contain distracting elements to be ignored in the form of faster responses to all trials regardless of type (Hernndez, Costa, Fuentes, Vivas, Sebastin Galls, 2010; for a meta analysis see Adesope, Lavin, Thompson, & Ungerleider, 2010). These results suggest that bilinguals have m ore attuned processes for monitoring and resolving conflict in tasks requiring rapid decision making ; when a conflict arises their
EXECUTIVE CONTROL AND LANGUAGE MODE 4 performance is not hindered as much as it is for monolinguals, nor is their overall performance hindered as much by the gener al expectation of a sudden conflict as it is for monolinguals. Conflict C ontrol Researchers have attempted to pinpoint the specific cognitive processes in which bilingual people excel, and it has been suggested that linguistic processes such as language s election are related to domain general executive control processes as opposed to those specific to a certain type of task (Hernndez, et al., 2010) although this is by no means a universally accepted proposal F or example ) study showed that performance on a Simon task, which they called a test of visual controlled attention, was associated with visual but not verbal working memory. Bialystok (1999) does not use the psychological framework of the attentional system and its networks but she makes a similar distinction in the skills needed for processing tasks involving conflict; i.e., she specifies analysis the ability to represent abstract structures, and control the ability to selectively attend, particularly in mislead ing situations. These structures appear to be fairly arbitrarily defined, and Bialystok fails to place them in a context compatible with the accepted cognitive network s Rather, Bialystok juxtaposes the model containing analysis and control with another, t he cognitive com plexity and control (CCC) model. This model also contains two elements consciousness or complex rule systems, and executive functioning or reflec tive awareness of these rules implying that the analysis and control framework and CCC mo del are equivalent structures. She does note, however, that in the CCC model the two elements are interdependent and cannot be separated or analysed in isolation. To test her theory, Bialystok asked monolingual and bilingual children, one group of 4 year olds and one group of 5 year olds to complete two tasks, the Dimensional Change Card Sort and the Moving Word task. In the Dimensional Change Card Sort paradigm,
EXECUTIVE CONTROL AND LANGUAGE MODE 5 participants are asked to sort cards into two piles based on a characteristic of the objects pictured, for example red pictures in one pile and blue pictures in the other. The participants are then asked to learn a new rule and sort the cards again accordingly, for example rabbits the participants viewed two pictures and the experimenter named the objects in each, then showed the child a label naming one of the pictures and read the label to the child. The experimenter then put the label next to the picture it named and asked the c hild what the label said. The child was next introduced to two puppets that interacted for a short time before starting to fight. In their scuffle, the puppets knocked the label towards the other picture, which it did not name. The experimenter asked the c hild what the label said again, Bialystok calls this the inconsistent question answered this question, the experimenter pointed out that the puppets had made a mess and moved the label back to its appropriate place, asking again what it said, which Bialystok calls the consistent question. There were three blocks in the Moving Word task, each requirin g the participants to answer the introductory, inconsistent and consistent que stions about the content of the label. Age and bilingualism were both positively associated with higher scores on the inconsistent question, when the label was next to the wrong picture. Age and bilingualism were also positively associated with performance on the second phase of the Dimensional Change Card Sort, when the participants had to sort according to the new rule (rabbits and boats, in the example) and ignore the old rule (red and blue, in the example). There was a positive correlation between score on the inconsistent question on the Moving Word task and the card sort after the rule change, even when vocabulary test score and visually cued memory were considered which suggests that the tasks tap related concepts These results show that bilingual c hildren have an advantage in both perseveration of response despite a change (the position of the label in the Moving Word) and adaptation of response due to a
EXECUTIVE CONTROL AND LANGUAGE MODE 6 change (the new rule in the card sort) These two skills involve intentional selection of inform ation in the face of salient misleading information, whic h is the same concept as that behind the tasks used to study executive control in adults. Bialystok (1999) discusses her results in terms of the analysis control and CCC models, and cites the lack o representation as support for the analysis control model. She propose s testing the two models empirically through problem solving tasks that involve abstract representations called analysis and consciousness in the two paradigms respectively Studying these abstract representations further could reveal bilingual advantages, providing evidence for both models, or could show that bilingual children are no different from monolingual children in thei r consciousness/analysis skills providing evidence for the analysis control model only, since the CCC takes executive control and consciousness to be non separable and interdependent. Further testing could lead not only to substantiation of one of the par adigms put forward by Bialystok but also to providing more evidence for the debate regarding the bilingual advantage as lying either in only the conflict monitoring subcomponent or in both subcomponents of executive function. Since executive functioning is one of the subcomponents of the CCC it seems that the paradigms tap higher level processes than the more specific areas of conflict monitoring and conflict resolution, so the experimental task must narrow down the locus of difference (if a difference is even found) with a more focused design T he current study will zoom in on the control/executive functioning element, which incidentally can be placed in the framework of the attentional system and descriptions and its involvement in task s with distracting information, seem s to be equivalent or very similar to executive control.
EXECUTIVE CONTROL AND LANGUAGE MODE 7 Neurological P rocesses in C onflict C ontrol The neurological processes during task switching, the practice of moving mentally between rules like the switch from one dimension to another in the Dimensional Change Card Sort also reveal differences between monolingual and bilingual participants Garbin et al. (2010) administered a Dimensional Change task similar to that used by Bialystok (1999) to young adults in w hich the participants were asked to decide based on a label presented along with the stimulus, either its colour (blue or red) or shape (square or circle). The focus ther it was to examine their ability to hold the two rules in mind and switch between them rapidly. Two mixed dimension blocks, each seven and a half minutes long, were administered in an MRI machine through MRI adapted goggles. The stimulus was presented for 1000 ms with a 2000 ms interstimulus interval, much longer periods than in comparable studies with adult participants. This is note worthy because of th e mediating effect of delayed response on this sort of task when comparing bilinguals and monolingua ls (see discussion of Martin Rhee and below). switch, those that were classified on the opposite dimension (colour or shape) to the previous trial, or non switch, those that were classified on the same dimension (colour or shape) as the previous trial Monolingual participants made more errors on switch trials than non switch trials, and also responded more slowly to switch trials, while bilingual participants did not re spond differently to the two types of trial. There were also physiological differences between the groups. Monolingual cingulate cortex and the left inferior par ietal lobe for switch trials than for non switch trials, frontal gyrus during switch trials, an area that was not affected in monolinguals. These
EXECUTIVE CONTROL AND LANGUAGE MODE 8 results are consi stent with other findings of a monolingual disadvantage in task switching, of the right inferior frontal gyrus as involved in reactive inhibition and volunt ary co ntrol to resolve conflict; in this case, monolingual participants were consciously trying to suppress the rule from the previous trial The involvement of the inferior parietal lobe and the anterior cingulate cortex is also consistent with previous researc h on task switching, as they have been implicated in the maintenance of cognitive representations and cognitive control there is evidence that the left inferior frontal gyrus is involved in the suppression of inappropriate responses (Robbins, 2007, as cited in Garbin et al., 2010 ; Rodriguez Fornells et al., 2005, as cited in Wang et al., 2009 ) a process related to the reactive inhibition seen in monolinguals Bi lingual and monolingual participants clearly differ in how they respond to changing task demands, at least in these specific tasks. A notable element of the two studies discussed above is that they mixed verbal and non verbal elements. The Moving Word task (Bialystok, 1999) requires the use of metalinguistic knowledge along with the response perseveration when the label moves, and Garbin and colleagues (2010) claimed that the computerised Dimensional Change task was non verbal, but the labels appearing with the stimulus to instruct the participant which rule to use were words Separating verbal from non verbal tasks could be useful in the comparative study of bilingualism due to the fact that it is presumed to a large extent that bilinguals and monolinguals differ in their linguistic processing, and any difference o n non verbal tasks c ould suggest differences between the two groups deeper than those at the linguistic level differences at a more domain general and fundamental level. These differences would al low researchers to investigate which very basic processes are at work in linguistic tasks, or which resources are shared by linguistic and
EXECUTIVE CONTROL AND LANGUAGE MODE 9 non linguistic tasks, to make generalisations about how skills build upon one another and how different cognitive mec hanisms interact. Cognitive P rocesses in N on verbal T asks Many types of tasks have been used to non verbally tap the executive control network of the attentional system in the past, but the two most prevalent are the Simon task and the f lanker task both of which rely on the presentation of some stimuli that activate conflicting responses In the Simon task, participants typically see circles of one of two colours appear on a computer screen. They are asked to press a button on the left for one colour re d in this case and a button on the right for the other blue in this case The position of the circle on the screen is the crux of the distraction; in half of the trials the position congruent trials the circles appear on the side of the visual field that corresponds to the side of the correct response, in the other half the position incongruent trials the circles appear on the opposite side of the visual field to the correct response. This task is intended to target the executive control network, with the participant constantly assessing the presence or absence of conflict and, when it is present, attempting to resolve said conflict. Figure 2: Examples of Simon stimuli and the position response congruence In previous research, reac tion times have increase d position is incongruent with the position of the response, although this effect is reduced for Red i ncongruent Blue i ncongruent Red c ongruent Blue c ongruent
EXECUTIVE CONTROL AND LANGUAGE MODE 10 bilingual participants ( Bialystok, 2006; Bialystok, Craik, Klein & Viswanathan, 2004 ; Emmorey et al., 2008 ) resolve conflict, although Martin Rhee and Bialystok emphasise that this advantage is only present for the specific type of conflict present in the Simon task. They supporte d this claim by administering a Simon task in study 1 and a Stroop and a Simon task in study 2 to mono lingual and bilingual children. In study one 34 5 year olds, half of whom were bilingual, completed three Simon tasks that varied only in the time betwee n presentation of stimulus and response. In one version, the children could respond as soon as they saw the stimulus, in the other two they had to wait for a visual cue (a picture of a hand with a finger extended) before they could respond. These cues for delayed response appeared 500ms or 1000ms after the stimulus In the immediate response version of the task, bilingual participants responded faster than monolingual participants and all participants responded faster to congruent than incongruent trials. There was no difference in the size of the congruence effect by language group, i.e., the difference between the responses to congr uent and incongruent trials was the same for the bilinguals and monolinguals. In the delayed versions, however, the participa nts did not differ in their response times overall and did not respond any faster to congruent than incongruent trials. In study two, 41 4 year olds, 20 monolingual and 21 bilingual, completed the Simon task and a Stroop naming task. The Stroop naming task consisted of the presentation of a picture of either the sky or an animal. The sky could represent daytime, with a bright sun, or night time, with a moon, and the animal could either be a dog or a cat. For the first block of 24 trials, the same name trial s, the children were asked to name the pictures as fast as they could. For the second block, however, the opposite name trials, the children were asked to reverse the names and say sun, and so on.
EXECUTIVE CONTROL AND LANGUAGE MODE 11 Again, on the Simon task, par ticipants responded faster to congruent trials than to incongruent trials, and bilingual participants responded faster overall than did monolingual participants. The difference between reaction times to congruent and incong ruent trials was smaller for bilingual participants (M = 122ms, SD = 131) than for monolingual participants (M = 185ms, SD = 140), and although the difference was not significant this is an example of the evidence for the bilingual advantage in executive control. On the Stroop task, the bilingual and monolingual participants did not differ from each other, and the same name and opposite name trials did not differ from each other. Martin Rhee and Bialystok (2008) explain the absence of a congruence effect a nd bilingual advantage in processing speed on the Stroop naming task, in the presence of those effects on the Simon task, in terms of the types of executive control involved in the tasks. They call the Simon task a bivalent task, because two elements, colo ur and position, are present in the stimulus, one of which must be ignored (position). This bivalent display requires what Martin Rhee and Bialystok call interference suppression since of the two elements one must be ignored. The Stroop task, on the othe r hand, is univalent because there is only one element in the stimulus: it only ever contains a sun, a moon, a dog or a cat This univalent display requires response inhibition since t he distractor in this task is a habitual response that must be ignored in opposite name trials, something that is a part of Although the literature may suggest that bilinguals should have an advantage in response inhibition this study illuminates the difference between the types of executive control involved and clarifies the locus of the bilingual advantage as purely within the interference suppression sphere which is an important distinction when designing tasks to further explore group differences Namazi an d Thordardottir (2010) challenge the interpretation of the language group differences on the Simon task as a bilingual advantage in interference suppression, and instead claim that working memory is the true predictor of performance, presumably with
EXECUTIVE CONTROL AND LANGUAGE MODE 12 biling ualism as a confound that co occurs with increased working memory capacity. They tested French speaking, English speaking and French English bilingual children approximately 5 years old, the same age group as the participants in Martin Rhee and Bialystok ( 2008), on verbal working memory, verbal short term memory, visual working memory and what the experimenters call visual controlled attention. Although working memory and short term memory are largely considered synonymous terms, these are the names the aut hors gave to the task, and thus the two will be differentiated here. V erbal working memory was measured with a sentence verification and recall task: the children were read sentences of increasing length and asked if the sentences were true and then to rep eat the final word of the sent ence. Visu al short term memory was measured with non word repetition and digit span tasks, d igit span involving the participants repeating strings of numbers that increase d in length. V isua l working memory was measured with a task on a computer in which children saw a 3x3 grid with first 2, then 3, 4 and 5 cartoon characters, each in a square. After the characters had been present for 2 seconds, they disappeared and the child was asked to touch the squares in which characters h ad appeared, and the child was only allowed as many guess es as there had been characters (2 5 depending on the stage of the task). Finally, the controlled attention task was the Simon task as described above with a fixation point present between trials, wh ich each consisted of two stimuli, one of each colour. The bilingual children were tested in two sessions that were each conducted entirely run by a different experimenter in order to observe in comparative studies bilingual participants should not know that they are being studied as bilinguals, as this could change their language usage, their mental state and their perception of the task, the experime nter and themselves (Grosjean, 1998, as cited in Namazi & Thord ardottir, 2010;
EXECUTIVE CONTROL AND LANGUAGE MODE 13 Grosjean, 2001; s ee below for a more in depth description of the mode hypothesis and its implications ) Namazi and Thordardottir (2010) found that the bilingual and monolingual participants did not differ i n reaction time or accuracy on the Simon task although visual memory (the task with the grid containing cartoon characters) was negatively correlated with reaction time and positively correlated with accuracy, so children with poorer visual memory responded more slowly than those with better visual memory, and children with better visual memory responded more accurately than those with poorer visual memory. None of the verbal memory measures was associated with performance on t he Simon task, leading the researchers to conclude that the Simon task does not tap a domain general mechanism the visual task was only related to visual memory. This conclusion precludes a bilingual advantage in interference suppression, which is, at le ast in Martin Rhee and a domain general skill that could not be claimed to be only visual. It is possible that Namazi and Thordardottir (2010) uncovered working memory as the true predictor of Simon performance, indepen dent of bilingualism, but it is difficult to reconcile their results with those of the previous study since Martin participants did not differ on digit span, one of the memory measures used by Namazi and Thordardottir. Another study th at found no difference between monolingual and bilingual participants on the Simon task was conducted by Morton and Harper (2007), in which 34 6 and 7 year olds, half of whom were bilingual, completed the Simon task. All participants responded faster and more accurately to congruent trials than to incongruent trials but the language groups did not differ. H owever, higher socioeconomic status, regardless of language group, was associated with error rates that were less affected by the congruence of the tria l. Morton and Harper therefore propose that socioeconomic status mediates the effects of bilingualism on executive control. This claim is difficult to evaluate in terms of
EXECUTIVE CONTROL AND LANGUAGE MODE 14 previous findings because of the scarcity of studies of bilingualism that measure so cioeconomic status, although Bialystok aims to control for it by recruiting participants from the same schools and neighbourhoods (Bialystok, 1999; Martin Rhee & Bialystok, 2008). The other popular non verbal executive control task is the f lanker task. In this task participants see a cue followed by a stimulus made up of five horizontal arrows in a row (see Figure 3 below for f lanker stimuli), and are asked to indicate which way the central arrow in the stimulus, the target, is pointing. In half of the tr ials, all four of the flanker arrows point in the same direction as the target arrow, their direction is congruent with that of the target and in the other half the all four of the flanker arrows point in the opposite direction to the target arrow, their direction is incongruent with that of the target Figure 3: Examples of flanker stimuli with fixation point Incongruent Congruent Incongruent Congruent
EXECUTIVE CONTROL AND LANGUAGE MODE 15 All version s of the flanker task are useful for testing the executive control network in that they require the monitoring and reso lution of conflict (Costa et al., 2008), but the task has been used in many forms, often including an additional variable. There is a version with an additional go/no go element (Costa et al., 2008), a version with multiple possible visual cues (Fan et al. 2002 as cited in Hernndez, et al., 2010 ; Costa et al., 2008) and a version with visual and auditory cues (Callejas et al, 2005). The cue congruence variable is intended to target the alerting and orienting networks of attention, in that it requires par ticipants to be ready to detect stimuli and to allocate cognitive resources appropriately (Hernndez, et al., 2010). Using cue information to tap the alerting and orienting networks of the attentional system in addition to the arrow congruence element allo ws the task the power to predict and address questions about all three subcomponents of the attentional system. The Attentional Network Task (ANT; Fan et al., 2002; Costa et al., 2008) is one such flanker task with multiple possible visual cues presented b efore the arrow stimuli. The cue manipulation was aimed at targeting the alerting and orienting networks, as discussed above The cue could either indicate the position of the upcoming stimulus (a valid cue); appear centrally, in the same place as the fixa tion point; appear at both possible stimulus positions at once (a double cue); or not appear at all. The valid and central cues were intended to test the orienting network, since they both should have activated the alerting network by signalling the onset of the stimulus, but only the valid cue should have activated the orienting network by giving location information. The double and absent cues were intended to test the alerting network, since neither should have activated the orienting network because nei ther contained location information, but the double cue did signal the onset of the stimulus, so it should have activated the alerting network.
EXECUTIVE CONTROL AND LANGUAGE MODE 16 Network activated Cue Alerting Orienting Valid Yes Yes (correct information ) Central Yes Yes (incorrect in formation ) Double Yes Yes (incorrect information ) None No No Figure 4: Cue types presented in the ANT Since the valid cue is the only cue that gives correct information about the position of the upcoming stimulus, participants were expected to respond fastest to trials preceded by valid cues. They were expected to respond slowest to trials not preceded by any cues since their attentional system was neither in a heightened alert state nor had it allocated resources to a particular domain, in this case t he location of the upcoming stimulus. The problem with these predictions is that the central and double cues both provide alerting information and both provide orienting information, albeit incorrect orienting, but Fan et al. (2002) and Costa et al. (2008) claim that they are measuring different variables. This leads to the criticism that the ANT does not adequately separate measures of the orienting and alerting networks and the suggestion that attempting to activate the different networks should be done u sing different modalities (Callejas, et al., 2005). Costa et al. (2008) used the ANT despite these criticisms and found that young adult Catalan Spanish bilingual participants did respond faster overall than young adult Spanish speaking monolingual partici pants In addition, although all participants showed a congruence effect such that they were faster to respond to congruent trials than incongruent trials, bilingual participants were faster still than monolinguals. The difference in response times between congruent and incongruent trials was smaller for bilingual
EXECUTIVE CONTROL AND LANGUAGE MODE 17 participants than for monolingual participants, i.e the conflicting information hindered the bilingual participants less, but the difference diminished with practice. By the end of the experimen t, the difference between the two types of trial was the same for both groups. All participants responded faster when there was a double cue than when there was no cue, and this effect was greater for bilinguals than for monolinguals, showing that the bili nguals benefitted more from the alerting cue than monolinguals did All participants also responded faster when there was a valid cue than when there was a central cue, but this effect was not greater for bilinguals than for monolinguals, showing that the bilinguals did not benefit more from the orienting cue than monolinguals did. The congruence effect was greater when there was no cue (lack of alerting information) and smaller when the cue was valid (presence of accurate orienting information) but these d ifferences were not related to language group. The results point to a diminished congruence effect and advantages in executive control and alerting for bilinguals, with no difference betwe en language groups in ori enting. According to Costa et al. (2009), there are two possible mechanisms underlying this diminished congruence effect, which can be thought of as less of a performance decline in the presence of conflict. First, bilinguals may have more highly developed general executive function, which can be divided into two components: conflict monitoring and conflict resolution. Advanced conflict monitoring should lead to greater ease switching between tasks that do and tasks that do not contain conflict, and advance d conflict resolution should aid in responding quickly and correctly to those trials that do include distracting information. Second, bilinguals may only differ from monolinguals in terms of conflict monitoring, but an advantage in monitoring impacts eleme nts of a task differently. If the argument can be made that a conflict monitoring advantage produces the same effect as a conflict resolution advantage would produce, the question of utility of the distinction is raised. Evidence for whether the two constr ucts can
EXECUTIVE CONTROL AND LANGUAGE MODE 18 be distinguished from one another would be essential for assessing the structure of attention in general, and more specifically where group differences lie. This knowledge would be applicable to other areas of research besides that on bilingualism such as modelling the development of attention and decision making, cross cultural comparisons and psychopathological research. The concerns about the orienting variable in the ANT above (Callejas et al., 2005; also see Figure 4) may explain the lack of differentiation between the groups in the activation of their orienting networks or it could be that there truly is no bi lingual advantage in orienting. The best way to disambiguate the two possibilities and clarify t he extent of the language group differences seems to be a slight change in the methodology Callejas et al. (2005) did exactly that and modified the ANT to inve stigate the interactions between the three attentional networks. Instead of using different for ms of the visual cue to activate the alerting and orienting networks, to address their concerns about the separation of the networks t hey used an auditory signal to activate the alerting network and a visual cue to activate the orienting network. Since the auditory signal could not provide any information about the location of the upcoming stimulus, there was no chance that the alerting cues could simultaneously activate the orienting network. There was still activation of the alerting network in the presen tation of orienting cues, but in order to study the networks separately, in this study Callejas and colleagues added a no cue level to their orienting network variable, so there were three possibilities for the visual cue: a valid cue indicating the positi on of the upcoming stimulus, an invalid cue presented misleadingly in the position in which the stimulus would not appear, or no visual cue at all. The study maintained the two level alerting network variable with an alerting auditory signal in half of the trials and no auditory signal in the other half. This design allowed the presentation of trials with both misleading orienting information and helpful alerting information and trials
EXECUTIVE CONTROL AND LANGUAGE MODE 19 with no orienting information a nd helpful alerting information. I n cont rast, in the ANT misleading orienting information was deemed to be equivalent to no orienting information the central cue was supposed to represent the presence of an alerting cue and the absence of an orienting cue when in fact it did include orienting information. Any spatial information should activate the orienting network, and the fact that the stimulus never appeared in the central position does not counteract this activation. Callejas et al. (2005), with this modified flanker task, found that orie nting activation facilitated executive control, that alertness facilitated orienting and that alertness inhibited executive control. That is, when the visual cue was valid, the effect of the flanker congruence was diminished, and so there was less of a dif ference between congruent and incongruent trials. When the auditory signal was present, the visual cueing effect was greater, that is, there was more of a difference between uncued and cued trials. When the auditory signal was present, the effect of flanke r congruence was increased (Callejas et al., 2005). This version of the flanker task has not been used in comparative studies with bilingual and monolingual participants, but if the interactions found by Callejas et al. (2005) are generalisable, there coul d very well be language group differences in the strengths of the associations due to the bilingual executive control advantage. At short stimulus onset asynchronie s there could be a bilingual advantage in orienting, since orienting facilitates executive c ontrol, or a bilingual disadvantage in alerting, since alerting inhibits executive control. These findings could serve to help explain the lack of significant differences between language groups in Colzato and coll t asynchronies O n an inhibitio n of return task, which required participants to decide whether a line was vertical or horizontal when the line was presented randomly at one of two possible locations monolingual and bilingual participants only differed at short stimulus onset
EXECUTIVE CONTROL AND LANGUAGE MODE 20 asynchronies The line stim ulus was preceded by a rectangular cue that delineated one of the possible locations, which was present for 50ms, and this cue was followed by a period of 50, 350, 650 or 950ms before the stimulus was present ed (the stimulus onset asynchrony). When the stimulus appeared in the same location that had been indicated by the cue, the response was facilitated and participants responded more quickly. At the shortest stimulus onset asynchrony 100ms between the cue a ppearing and the stimulus appearing, only the monolingual participants showed response facilitation, whereas at the longer onset asynchronies, the groups both experienced the facilitation. That is, at longer onset asynchronies the groups did not differ in their performance. This can be addressed by ensuring that any cues designed to allow participants to orient are presented within 100ms of the presentation of the stimulus they are cueing. Cognitive P rocesses in V erbal T asks Many tasks that investigate bi stimuli include the use of interlingual or interlexical homographs. Interlingual homographs are words that belong to both languages but are pronounced differently and usually have different meanings in the two languages, for example kind red although some interlingual homographs are cognates in the two languages, in which case the words have the same or very similar meanings for example plural and aroma in Spanish and English Most researchers distinguish between these two types of words because cognates seem to be a special kind of homograph they share their written form and their meaning in the two languages. The methodological para digm using lexical decision tasks (LDT s ) in which participants indicate whether a string of letters is a real word or not, or some variant there of, to analyse the difference between control words and interlingual homographs was used to investigate the org anisation of the lexica, or mental dictionaries, of bilingual people. The main question addressed whether the lexica were integrated, as a single store of
EXECUTIVE CONTROL AND LANGUAGE MODE 21 linguistic information, or separate, as two independent stores, but some of the results led to more c omplex conclusions about the lexica and the way they are organised. Gerard and Scarborough (1989) found that in an LDT in only one language, participants responded to homographic noncognates, for example red, Spanish and dime, which m words purely in the target language of the LDT. The only factor that was significantly related to reaction time target la nguage words, homographic noncognates like red cognates like social with the same meaning in English and Spanish, and pseudowords) had no effect in either group except for bilinguals in the second session of the experiment. In this session, participants completed a second LDT with some stimuli repe ated from the first session and the bilinguals completed the task in their other language The homographic noncognates ( red ) and cognates ( social ) were still presented, but the purely target language words were translated into the other language, for example if participants had seen dog family red and social in the first session they would see perro familia red and social in the second. T he monolingual participants responded faster to repeated stimuli tha n they had to those same stimuli the first time which was an expected instantiation of repetition priming. The bilingual participants also responded faster in the second session to the homographic noncognates (like red ) and cognates (like social ) that had been presented in the first session, however they did not show this facilitation for translations of words purely in the first target language for example if they saw dog in the first session and perro in the second, they did not respond any faster to p erro even though the concept of that specific animal may have been active in their minds The se results point to a complicated answer to the question of how bilingual lexica are organised. The repetition effect for homographic cognates (like red ) and cogna tes (like social ) suggests integrated lexica, since the bilingual participants were asked to concentrate on the target language of the LDT. If the lexica were
EXECUTIVE CONTROL AND LANGUAGE MODE 22 separate, switching tasks from an LDT in one language to an LDT in the other should have reduced the activation of the homographs and cognates to a baseline since they would be represented completely independently from their readings in the new target language regardless of shared orthography or semantic content The lack of repetition effect for the translations of the purely target language words suggests that the lexica are separate, since if they were integrated the concepts associated with the purely target language words in the first session should have been primed, leading to facilitation of th eir translations in the target language of the second session, which was not the case. Gerard and Scarborough (1989) separate the processes involved in making lexical decisions into three categories: encoding, memory search (also called lexical look up) an d the decision/response. They use their results to argue that encoding is facilitated by repetition, but that lexical look up, or memory search, is not. This is interpreted to mean that bilingual lexica are integrated in some aspects but not others. Based on the data from studies like those discussed above, o ne could arg ue that the lexica are fully integrated but that they are organised in such a way that orthography, and not semantic content, is the defining feature of a node, or entry. In this way the bi lingual lexica may resemble a dictionary, containing some clusters of identical word forms that correspond to concepts unrelated to one another such as the different meanings of bank in English, as opposed to a thesaurus, with clusters of similar concepts that correspond to word forms unrelated to one another such as a group of synonyms for lucky T he lack of differentiation between the effect of repetition on cognates and homographic noncognates is consistent with orthography or visual coding in fact bei ng the deciding factor in lexical access. Purely target language words do not share their form with other words crosslinguistically, homographic noncognates share form but not meaning, and cognates share both their form and meaning. Had semantic content be en the most important element of access to words, the types would have clustered differently: responses to purely target
EXECUTIVE CONTROL AND LANGUAGE MODE 23 language words and homographic noncognates would have been similar and responses to cognates would have been faster. The only problem w ith this explanation is that the bilingual participants completing the Spanish target LDT responded faster to cognates than purely Spanish words and homographic noncognates with comparable frequency. Although this only happened in the Spanish target task, it suggests that the organisation of the bilingual lexica is more complicated than either a dictionary or thesaurus analogy could convey. Faste r reactions to cognates suggest the intuitive; that orthographic and semantic encoding and retrieval cues allow f or faster decision making than either one type of cue or the other alone. This phenomenon ha d been documented previously ( Caramazza & Brones, 1979; Cristoffanini, Kirsner, & Milech, 1986, as cited in Gerard & Scarborough, 1989) but it i s interesting that it was only observed in Spanish and was much more pronounced when Spanish was the second language presented i.e., when some of the purely Spanish words were translations of words that had been previously presented. I n the first session, when all of the st imuli were novel, the bilingual participants who completed the Spanish target LDT responded faster to cognates and purely Spanish words than to homographic noncognates but the difference was not statistically significant There is evidence that supports this view that organisation of lexica is not solely orthographic, and in fact suggests that orthography is not a major factor at all. Fox (1996) asked bilinguals to make a decision about whether a number to which they were attending was even or odd while i gnoring words that were presented at the same time. Later, in a LDT the participants were slower to respond to the translations of those words that had appeared earlier than they were to new words, even though participants were trying to ignore the words in the first task. This suggests that there is a strong semantic element to the organisation of the lexica. Elston Gttler, Gunter & Kotz (2005) also suggest semantic l anguage sufficiently that in a reading comprehension task including interlingual
EXECUTIVE CONTROL AND LANGUAGE MODE 24 homographs, barely any (only two out of twenty eight) participants reported having noticed first l anguage readings could not have been activated The phonological element of words could also be important in lexica l organisation, as Schwartz and Aras ( 2008 ) suggest. They found that the phonological characteristics of primes can cross linguistically act ivate targets f or example bark activated the Spanish barco in a relatedness LDT such that the pair bark boat took participants longer to judge than when bark was paired with an unrelated control wor d Bark activating barco is likely to be due to phonology rather than orthography since the /k/ sound is represented differently in the orthography of the two words but sounds the same. The mixed results regarding the organisation of the lexica in the bil ingual mind suggest that the linguistic systems people use are either extremely complex and multi faceted or are fluid and dynamic. It seems reasonable to assume that lexica are organised based not on one element of their contents but on multiple elements Van Heuven, Schriefers, Dijsktra and Hagoort (2008) conducted an imaging study using interlingual homographs to examine what they call language conflict. They characterise this conflict as triggered by three properties of interlingual homographs : belongin g to both languages having two different pronunciations and being semantically ambiguous. They speculate d that this conflict could activate representations in both languages, so they administered an English only LDT in which participants asked themselve and a General LDT in which participants asked in any of the languages I The LDT s both contained interlingual homographs, but in the English LDT participants were not aware of their prese nce them even though they were in fact English words, due to the fact that they were also Dutch words, which was a criterion for rejection. This would have been an incorrect
EXECUTIVE CONTROL AND LANGUAGE MODE 25 response, of course, due to the fact that they were also English words. The experimenters compared responses from the two tasks because interlingual homographs response in both tasks, but only in the General LDT does the homograph require the same response for both languages. Th e activation in the l eft i nferior p refrontal c ortex was greater for bilingual participants, particularly to words rather than pseudowords and in the English task more so than the General task. All of the participants showed stronger activation for words th an for pseudowords. In the English LDT there wa T he bilingual participants showed greater activation in the left inferior prefron tal cortex, the pre supplementary motor area and the anterior cingulate cortex for interlingual homographs than for purely English words. On the General LDT however, activation to interlingual homographs was only greater in the left inferior prefrontal co rtex suggesting that the pre supplementary motor area and the anterior cingulate cortex are related to executive control in language conflict. The anterior cingulate cortex was also involved in task switching in ly for monolingual participants. This could suggest parallels between monolingual processing of rule switching and bilingual processing of language inhibition (see Wang et al., 2009 for a more detailed neurological theory) Due to the nature of these tas ks, Dutch could have been completely inactive throughout the experiment since there were never any purely Dutch words and if it had been, there would have been no observed conflict. The results suggest otherwise, however: in the English LDT response t ime s to interlingual homographs were much longer than response times to control English words and there was greater activation of the left inferior prefrontal cortex pre supplementary motor area and anterior cingulate cortex In the General LDT none of t he participants respond ed differently to interlingual homographs and
EXECUTIVE CONTROL AND LANGUAGE MODE 26 controls, but for bilingual participants there were still differences in activation patterns, with greater activation of the left inferior prefrontal cortex for interlingual homographs th an for controls Dutch must have been active at least in some way, causing responses to certain English words (those that share a written form with Dutch words for example kind which ) to be much harder than others (those that do not have a Dutch reading ) when participants were trying to inhibit the influence of Dutch. Even though none of the words were purely Dutch, participants were not slower to respond to interlingual homographs (they did not differ from control words ) when the bi lingual participants were not trying to inhibit their first language. The similarities in the activated ar eas of the brain, however, show that the underlying mechanisms of the tasks were the same. This difference in responses to interlingual homographs an d control words based on the task instructions suggests that the language participants are using as well as the constraints of the task have a significant effect on performance. The state of p languages as in use or inhibited is referred to as language activation, and the literature provides multiple theoretical perspectives from which to address this activation. Language Activation Another perspective on the controversial bilingual advantage is that the advantage does exist but is only manifes ted in certain contexts. These contexts have been described by some as varying levels of language activation. Language activation models are diverse in their approaches but they all hinge upon the assumption that or words within a language, can be activated by cert ain events or circumstances. Because of the different focuses of the models, they are difficult to compare directly, but in and of itself this difference is interesting to examine. The Inhibitory Control model (Green, 19 98) claims that the activation levels of individual lexical representations are affected by language task schemas that compete and co operate with one another. The Bilingual Interactive Activation and Bilingual Interactive Activation + model s ( BIA and BIA+
EXECUTIVE CONTROL AND LANGUAGE MODE 27 respectively ; Dijkstra & van Heuven, 1998, 2002, as cited in Dijkstra & van Hell, 2003) claim that words have a resting activation level which reflects the state of language activation and proficiency. The BIA (Dijkstra & van Heuven, 1998 as cited in Dijk stra & van Hell, 2003) however the BIA+, the more recent formulation, claims that participant expectations can affect the task/decision system (note that the se models refer to activation states of words or lexical representations as opposed to the activation level of languages as a whole discussed below ). Grosjean (2001) puts forth a theory of language activation in which the languages of bilinguals are continuously changing i n their relative activation levels, and although he does not mention the role of individual words or lexical representations, according to Grosjean activation is extremely sensitive and has any number of contributing factors. He predicts language group dif ferences depending on the linguistic and non linguistic context, for example the location, the social dynamics, the languages being spoken and the languages understood in any given interaction. On this account, bilinguals function on a continuum of languag e mode from one pole, a fully monolingual mode, to the other, a fully bilingual mode. Bilinguals are constantly unconsciously evaluating the environment and activating one language more than the other, or both equally in a fully bilingual mode, which can affect which languages are in use but also other cognitive variables related to bilingualism like the processing speed advantage in tasks involving distracti ng information ( some examples). The more aware bilingual people are of their bilingualism and its potential utility in a given context the more bilingual is their mode, and theoretically the more effects of bilingualism they demonstrate, both behaviourally and cognitively. Very few psychologists have applied in an empirical sett ing however Grosjean (2001) interpreted the results of Dijkstra 1998) study as support for his language mode hypothesis. In the English language LDT, Dutch English
EXECUTIVE CONTROL AND LANGUAGE MODE 28 bilinguals responded as fast to interlingual homographs as to purely Eng lish words, but their responses to cognates were faster. In a second English LDT that also included some purely Dutch stimuli (words which participants would have recognised but to which they must partic ipants responded more slowly to interlingual homographs than to purely English words, and even more slowly to interlingual homographs with highly frequent Dutch readings. Grosjean explains that in the first case, the participants were towards the monoling ual end of the mode continuum, with enough of a bilingual influence to exhibit a cognate effect but not enough to produce a homograph effect. mode is reasonable since the experiment was framed as an English language task, and participants saw English Dutch homographs but never saw any purely Dutch words One problem with this interpretation is the exist ence of evidence for the activation of word representations from both languages even when RTs do not show a difference between interlingual homographs and words that are purely in one language For example, Van Heste (1999, as cited in Dijkstra & van Hell, 2003) showed that in a LDT in language A a n interlingual homograph from languages A and B primed the language A translation of the language B reading of the homograph leading to a reliable yet small (35ms) facilitation of the translation when it followed the homograph. For example, if participants were asked to complete an English LDT they would be asking themselves the question, this string Should the word kind appear, which means child in Dutch, Dutch English bilingual participants would be primed to respond more quickly to child as the next stimulus. These results lend authority to an activation model that takes into account the activation of single lexical representations rather than the activation of whole languages. While this does not necessarily invalidate fundamental claim, it does seem tha t the mode theory needs some adjustment. The continuity and sensitivity of
EXECUTIVE CONTROL AND LANGUAGE MODE 29 A lthough it is appealing to attempt to apply a theory to the available data it is of course preferable to test a theory with a bespoke design. Dijkstra and van Hell (2003) carried out such a study on mode by testing a group of trilinguals on word association and LDT s. The participants in experiments one and two were Dutch students who spoke English fluently and French less well, those in experiment three spoke French just as well as they spoke English. The participants were recruited in Dutch by Dutch speakers and were never emphasis on bilingualism. They received their instructi ons in Dutch and only responded to Dutch words, however some of the stimuli were interlingual homographs, that is they were words from more than one language that were spelt the same with different pronunciation and meaning. T he RTs to these homographs wo uld be the same as RTs to purely Dutch words if the participants were in a reasonably monolingual mode, with only Dutch highly active. The participants actually responded f aster to Dutch English homograph s than to purely Dutch words, suggesting that their English lexicon was active even in a purely Dutch setting. Dijkstra and van Hell explain this result by applying the BIA+ (2003) which addresses slightly differently than do the previous models, stating that participant expectations can affect the task/decision system, a change that is consistent with Grosjean described how to test participants in a monolingual mode, stressing that participants should not be separated by language group in any way and that demographics including questions about language experience should be collected at the very end of their participation. He also suggested how to put participants in a bilingual mode : testing should occur in an informal setting and the r esearchers who should be well known members of the should ensure that participants are comfortable hearing and using mixed language Dijkstra and van Hell (2003) do not go into extensive detail about their collection of demograp hic data, but they do note that their French speaking participants were students of French at a university recruited by a fellow student of French
EXECUTIVE CONTROL AND LANGUAGE MODE 30 which c ould feasibly move participants towards the bilingual pole of the mode continuum before the experimen t even started. participants in a bilingual mode, the elements concerning a community member and comfort with mixed language were both potentially violated. Age as a P articipant V ariable Some authors have explained the difficulty in consistently finding bilingual advantages as being due group often targeted by researchers due to the availability of students, the attentional processes are at the ir peak efficiency which may lead to a ceiling effect on the sorts of tasks used in attention studies ( Bialystok, 2007 attentional processes peak, they are on par with those of their bilingual peers. This would be a satisfactory explanation if there were very little evidence for a bilingual advantage in young adulthood and a plethora of evidence for the advantage at other time points, however, a s it turns out, during this peak period bilingual participants can still outperform monolingual participants, although sometimes only on the hardest tasks administered ( Bialystok, 2007). Costa et al. (2008) found bilingual advantages in participants aged 17 32 (M=22 years) on the ANT, and Bialystok (2006) found a bilingual ad vantage on the harder task she administered, a Simon task, for participants in their early 20s. In a meta analysis of the cognitive correlates of bilingualism, Adesope et al. (2010) found larger positive effect sizes for aggregated results regarding biling ual young people in post secondary education than for bilingual children in grades K 12, suggesting that either the proposed peak occurs later than college a ge or that it was present in the studies they analysed and did not affect the results. Colzato et al. (2008) found differences between language groups on their inhibition of response, in which the mean age of the monolinguals was 22 (SD=2.7) and the mean age of the bilinguals was 22 (SD=3.6), and the attentional blink task, in which the mean age of
EXECUTIVE CONTROL AND LANGUAGE MODE 31 the monolinguals was 22 (SD=0.8) and the mean age of the bilinguals was 23 (SD=3.6), but these tasks were interesting because bilinguals showed stronger inhibition which was manifested as a processing dis advantage. Colzato et al. (2008) also used a stop signa l task, in which participants, bilinguals with a mean age of 22 (SD=2.8) and monolinguals with a mean age of 22 (SD=2.3), were asked to indicate quickly and accurately whether a green arrow was pointing left or right using computer keys, but to inhibit the ir response s if the arrow changed colour to red, which happened without warning. On this task, the language groups did not differ, which could easily be explained by the fact that the task is not very difficult Similar tasks that have been used successful ly to elicit bilingual advantages have included more information, for example the flanker task, which involves a cue which is sometimes misleading, and each stimulus consists of a row of five arrows, of which the direction of only the central one is releva in the stop signal task, the bilinguals may have outperformed the monolinguals. The C urrent S tudy of the bilingual occupying a point on a mode continuum was written as a theoretical universal model, and whilst it is a blow to its credibility that Dijkstra and van Hell (2003) found evidence contrary to its claims, the model is not necessarily invalid. The experiments which challenge the theory were performed in the Ne therlands Because the Netherlands is a highly bilingual society and borders Belgium, an officially bilingual nation, to claim that this sample is representative of all bilinguals is to ignore salient social information namely that their citizens may spen d most of their lives in a bilingual mode Conducting a similar study in the USA, a highly monolingual society at least in public spaces might increase the likelihood of recruiting bilingual participants at the monolingual end of the continuum, and pos sibly demonstrate a differentiation in the
EXECUTIVE CONTROL AND LANGUAGE MODE 32 Grosjean called for in regard to the nature of the experiment; participants should not be aware of the researcher to study bilingu alism and the researcher should not be bilingual. In the current study, participants will be recruited in a monolingual mode or as close to a monolingual mode as is possible which means that they will participate in a n English language only setting wi th an English speaking experimenter They will complete the first half of the experiment three executive control tasks (the flanker, Simon and lexical decision tasks), with no knowledge that their bilingual status is relevant to the study. There will be n o mention or use of their other language in the instructions or materials, and they will be told that the task is measuring their attentional skills in various ways. The experimenter will then attempt to manipulate mode halfway through the experimental ses sion by directing participants to think about their second language while answering detailed and specific questions about their proficiency in that language. They will then complete each of the three executive control tasks again. This will allow the compa rison of performance on the tasks in different modes within participants, which has not been studied heretofore Previous research has used multiple tasks, such as the Simon and flanker tasks, to measure what has been assumed to be the same mechanism, and has produced very mixed results as regards a bilingual advantage on these tasks. The current study will seek to answer three questions: first, can the Simon and flanker tasks be reasonably equated as measurements of the same cognitive constru ct, executive control? Second, c ould language mode be mediating the bilingual advantage that has been differentially documented, i.e., control tasks involving non verbal stimuli? And third, does language mode have any effect
EXECUTIVE CONTROL AND LANGUAGE MODE 33 Method Participants Participants were recruited in person in the student union building of a small college in Flor ida, USA and via a student listserve e mail They were asked to participate in a study on attention and rapid decision making and given a short description of the Simon, flanker and lexical decision tasks There were 72 participants in total ( 49 females an d 23 males ; 27 bilinguals and 45 monolinguals ) and the age range was 18 24 years (M= 20.01 SD= 1.32 ). Twenty three of the bilingual participants spoke Spanish as their other language, 3 spoke German and 1 spoke French. Of the monolingual participants, 28 sp oke Spanish as their second language, 1 spoke Portuguese, 1 spoke Hebrew, 6 spoke German and 9 spoke French. Seventy of the seventy two participants were right handed Materials and I nstruments Participants completed two Simon task s two f lanker task s (Co sta et al., 2008, 2009 ; Callejas et al., 2005 ) and two LDT s They also completed the I nteragency L anguage R oundtable self assessment scales of language proficiency (govtilr.org) for their second language to indicate their self assessed degree of bilinguali sm and a short demographic questionnaire S ee Appendix A for the proficiency tests and Appendix B for the demographic questionnaire The Simon, f lanker and lexical decision tasks were administered on an HP Elitebook 6930P laptop with a Cedrus RB 830 respo nse pad. Superlab 4.0 (www.cedrus.com) was used to present stimuli and record responses and reaction time s The language proficiency tests and the demographic questionnaire were administered on paper and were not timed In the Simon task, a circle, either red or blue, appeared on the screen for 1700ms or but it was always clearly to t he left or right of the centre. S ee Figure 2 for examples of Simon stimuli. The participant was asked to respond with a left button press when the circle was
EXECUTIVE CONTROL AND LANGUAGE MODE 34 red and a right button press when the circle was blue, with the position of the stimulus as the source of potentially distracting information. In the flanker task, a plus sign ( +) marking the fixation point was present throughout the task, and after 400ms, a cue either above or below the fixation point appeared for 100ms. As soon as the cue disappeared, stimulus was the same distance from the fixation point as the centre of the cue, but it could trial. This r esulted in four equally represented trial types in the f lanker task: trials in which the target arrow pointed in the same direction as the flanker arrows and trials in which it did not, and trials in which the cue accurately indicated the position of the s timulus and those in which it did not. The participant was asked to keep his/her eyes on the fixation point and was not warned about the cues. The participant was asked to respond with a left button press when the target arrow pointed left and a right but ton press when the target arrow pointed right. The cue and the flankers both served as distractions from the relevant information, which was the direction of the central arrow.
EXECUTIVE CONTROL AND LANGUAGE MODE 35 Figure 5: Schematic representation of an incongruent flanker trial with a misleading cue In the LDT, strings of letters appeared in the centre of the screen in 20 point Tahoma font for 1700ms or until the participant responded. The stimuli were purely English words English Spanish homographic cognates, English Spanish homographic non cognates and pseudowords. The word stimuli were taken from Macizo, Bajo and Martn (2010) and Gerard and Scarborough (1989), and the pseudoword stimuli were generated using Wuggy (Keuleers & Brysbaert, 2010) based on the phonologic al patterns of Spanish and English. See Appendices C and D for a full list of the stimuli. Procedure Participants were asked to participate in a study on attention and decision making and completed the experiment in a quiet room in the student union. No m ention of language was made, and participants were told that the experimenter needed as many people as possible with the single stipulation that their eyesight be normal or corrected to normal. The order of presentation of the Simon and f lanker tasks was counter balanced, since one of the objectives of the study was to compare participants 400ms 100m s 1700ms
EXECUTIVE CONTROL AND LANGUAGE MODE 36 Simon and f lanker tasks. The LDT was always presented after the Simon and f lanker tasks since it was not being compared and was the most likely of th e tasks to lead to fatigue. After these three tasks, the participants were told that they were being given a break. In the break, participants provided demographic information and completed the language proficiency tests for listening, reading and speakin g. After the break the participants returned to the computer based tasks and were told that their instructions were the same as before. The Simon and f lanker tasks were counter balanced again, which resulted in four groups for the purpose of comparing the tasks. Results Language P roficiency The range of possible scores on the language proficiency l istening and speaking tests was 0 5 in increments of 0.5 and on the reading test the possible range was 0.5 4, also in increments of 0.5. The mean scores for all 72 participants were 2.24 (SD = 1.58) on the listening scale, 1.98 (SD = 1.59) on the speaking test, and 2.33 (SD = 1.11) on the reading test. Scores were representative of the full range, with four participants scoring the minimum on all three scales and eleven participants scoring the maximum. classification as bilingual or monolingual was positively correlated with their second language proficiency as measured by the I nteragency L anguage R oundtable tests The listening subscale sco re was positively correlated wit h bilingualism, r = .684 p <.0001 as was the s peaking subscale score, r = .663, p <.0001 and the reading subscale score, r = .709 p <.0001 Simon and F lanker E quivalence Reaction times to the Simon and flanker tasks were h ighly correlated with each other before the break, all p values less than .01. Reaction times to the congruent trials of the flanker task after the break were correlated with reaction times to both types of trial in the Simon task before the break, however there was no relationship between the
EXECUTIVE CONTROL AND LANGUAGE MODE 37 incongruent trials of the flanker task after the break and any portion of the Simon task, nor was there a relationship between the congruent trials of the flanker task after the break and the Simon task after the bre ak. Table 1 : Relationship between reaction times on Simon and flanker tasks Note: p <.05 ** p <.01 *** p <.001 Simon before break Simon after break Congruent Incongruent Congruent Incongruent Flanker before break Congruent r=.478*** r=.432*** r=.384*** r=.381*** Incongruent r=.410*** r=.381*** r=.305** r=.302** Flanker after break Congruent r=.333** r=.269* r=.199 r=.217 Incongruent r=.197 r=.128 r=.111 r=.133
EXECUTIVE CONTROL AND LANGUAGE MODE 38 Congruence All participants responded more quickly to congruent trials than to incongruent trials in bo th the Simon and f lanker tasks, before and after the break S ee Table 2 for reaction times to each trial type in the Simon and flanker tasks Overall, t he extent of the difference between congruent and incongruent trials, th e congruence effect was the same for bilinguals and monolinguals, and there was no correlation between language proficiency scores and the extent of the effect all p value s greater than .4 50 Task Trial type M (ms) SD t value (df) p value Before the br eak Flanker Congruent 1008.42 69.78 12.71 (71) <.0001 Incongruent 1058.20 76.69 Simon Congruent 535.93 78.18 6.61 (71) <.0001 Incongruent 563.41 78.74 After the break Flanker Congruent 949.44 59.34 10.14 (71) <.0001 Incongruent 984. 15 65.90 Simon Congruent 508.08 87.56 6.07 (71) <.0001 Incongruent 530.98 92.26 Table 2 : Reaction times of all participants by task and congruence Participants responded more quickly to both trial types after the language mode focussed break t han they had before it, all p values <.003, but this difference was affected by language group. Difference scores were calculated to examine the change in congruence effect from before to after the break. The congruence effects in the two sessions were not significantly different for monolingual participants, M = 6.35, SD = 29.15, t (44) = 1.46, p =
EXECUTIVE CONTROL AND LANGUAGE MODE 39 .15, but they were significantly different for bilingual participants, M = 15.60, SD = 35.51, t (26) = 2.28, p = .03, with the difference between reaction times t o congruent and incongruent trials before the break larger than the difference between them after the break. See Table 3 and Figure 6 for differences between incongruent and congruent trials before and after the break, by language group. Table 3: Difference between incongruent and co ngruent trials Figure 6: Difference between incongruent and congruent trials M (ms) SD t val ue (df) p value Monolingual Before the break 35.39 25.39 After the break 29.03 20.90 1.46 (44) 0.151 Bilingual Before the break 44.03 23.91 After the break 28.43 23.62 2.28 (26) 0.031
EXECUTIVE CONTROL AND LANGUAGE MODE 40 The difference scores of the two language groups were not significantly different from each other before the break, where the mean monolingual difference score was 3 4.51ms ( SD = 23.56) and the mean bilingual difference score was 43.82 ( SD = 21.68), t (51) = 1.45, p=.153, or after the break, where the mean monolingual difference score was 34.31ms ( SD = 17.42) and the mean bilingual difference score was 31.77ms ( SD = 18 .67), t (51) = 0.51, p=.615. Cue T ype in the F lanker T ask All participants responded more quickly to trials with a valid cue or with no cue than to trials with a misleading cue both before the break, F(2 69 ) = 46.16, p<.0001 and after it F(2 69 ) = 63. 22 p<.0001 See Table 2 for reaction times to each cue type in the flanker task Participants responded more quickly to each cue type after the break than they had before ( all p value s <.0001) but this difference was not affected by language group. Cue type M (ms) SD F value p value Before the break Valid 955.60 80.43 46.16 <.0001 Misleading 994.43 74.52 None 949.90 71.47 After the break Valid 887.96 64.09 63.22 <.0001 Misleading 925.06 65.18 None 887.37 62.19 Table 4 : Reaction ti mes of all participants by cue type on the flanker task
EXECUTIVE CONTROL AND LANGUAGE MODE 41 Word T ype Before the break, all participants responded faster to purely English words ( M = 660.88 SD = 80.48 ) than to homographs, regardless of the whether the homograph was a cognate ( M = 697.15 SD = 94.98 to homographic cognates; M = 720.83, SD = 93.63 to homographic non cognates) and faster to homographs than to pseudowords ( M = 763.96 SD = 81.97 ) F = 36.02, p <.0001 After the break, the type of homograph became influential and participant s responded most quickly to purely English words ( M = 629.90 SD = 71.45 ) less quickly to homographic cognates ( M = 682.86 SD = 80.77 ) less quickly to homographic non cognates ( M = 709.29, SD = 85.58 ) and the most slowly to pseudowords ( M = 769.33 SD = 99.40 ) F = 66.31, p <.0001. See Table 3 and Figure 7 for reaction times to each word type before and after the break Ta ble 5: Reaction times of all participants by word type on the LDT M (ms) SD F value p value Before the break Purely English words 660.88 80.48 Homographic cognates 697.15 94.98 Homogra phic non cognates 720.83 93.63 Pseudowords 763.96 81.97 36.02 <.0001 After the break Purely English words 629.90 71.45 Homographic cognates 682.86 80.77 Homographic non cognates 709.29 85.58 Pseudowords 769.33 99.40 66.31 <.0001
EXECUTIVE CONTROL AND LANGUAGE MODE 42 Figure 7 : Reaction times of all participants by word type on the LDT Regardless of context manipulation, reaction time to pseudowords was positively correlated with score on the listening subscale of the language proficiency test, r = .29, p =.037. T he correlation between the speaking subscale and reaction time to pseudowords was not as strong r = .26, p =.061, nor was the correlation between speaking subscale and reaction time to homographic cognates, r = .26, p =.065. Before the break, reaction time to homographic cognates was positively correlated with score on all three of the subscales of the language proficiency test : listening, r = .31, p =.009; speaking, r = .36, p =.002; a nd reading, r = .26, p =.026 There was no such effect after the break, all p value s greater than .163. Score on the listening subscale was positively correlated with reaction times to purely English words, though again this was only significant before the break r = .31, p =.025. Only the purely English words we re subject to a practice effect; the mean reaction time to purely English words was 660.88ms ( SD =80.48) before the break and 629.90ms ( SD =71.45) after the break, t (71) = 3.94, p =.0002.
EXECUTIVE CONTROL AND LANGUAGE MODE 43 Discussion Bilingual and monolingual young adults seem to respond differently to tasks involving conflict between relevant and distracting information but this difference has been inconsistently demonstrated in the literature (Costa et al., 2008; Martin Rhee & Bialy stok, 2008; Morton & Harper, 2007; Namazi & Thordardottir, 2010) One of the aspects mediating the inconsistency could be language mode, the theory that bilingual people move along a continuum of language activation. This is supported by Namazi and Thordar dottir (2010) who showed that when language mode was controlled in a Simon task, working memory and not language group predicted performance. At the monolingual end of this mode continuum bilinguals have only one language active but at the bilingual end, both of the languages are active, and bilinguals are more likely to switch betw een languages (Grosjean, 2001). The three executive control tasks before and afte r a langua ge mode manipulation. The research questions it attempted to answer were whether the Simon and flanker tasks both measured executive control and whether a language mode manipulation would produce or enhance differences between monolingual and bilingual par ticipants. There were two elements to this question: the difference between groups on linguistic tasks and the difference on non linguistic tasks. The non linguistic tasks were intended to tap elements of the attentional system that have been linked to la nguage group differences in the past, in order to provide a comparison between the language groups when they should be most similar, with the bilinguals in a monolingual mode, and when they should differ more, with the bilinguals in a bilingual mode. The f indings of the current study are, in part, consistent with the previous research, particularly regarding the congruence effect. The findings regarding cues in the flanker task and word stimuli in the LDT are not as consistent with the literature, and may p ose problems for the language mode hypothesis and its application to non verbal stimuli.
EXECUTIVE CONTROL AND LANGUAGE MODE 44 The Simon and flanker tasks did appear to be equivalent with one important caveat: practice seemed to affect the tasks differently. Reaction times to the two tasks we re highly correlated before the break, but reaction times to each task after the break were not as closely associated. This is not a problem for post hoc comparisons between studies that used the different tasks unless participants were exposed to either task for an extended period of time. In the current study, participants completed 96 trials of each task before, and again after, the break (the maximum time spent was around 3 minutes), so the tasks may be equivalent u p until around 96 trials or a 3 minut e period. In the non verbal tasks, a ll participants responded faster to congruent trials, those in which there was no conflict present, than to incongruent trials. Congruent trials were those in the Simon task in which the stimulus was presented on the sa me side of the visual field as the correct response and those in the flanker task in which the target arrow and the four flanking arrows were pointing in the same direction. The reaction times of the two language groups did not differ from each other, but when difference scores for before and after the break were compared (the mean reaction time to incongruent trials minus the mean reaction time to congruent trials), the difference for bilingual participants decreased while the difference for monolingual pa rticipants did not. T his result is consistent with the ide a that a mode manipulation can enhance any language group differences In the more bilingual mode, the congruence effect demonstrated by the reaction times of the bilingual participants diminished, in line with the expectation that bilingual participants would experience less of a congruence effect than the monolingual participants ( as found in Bialystok, 2006; Bialystok, Craik, Klein & Viswanathan, 2004; Emmorey et al., 2008) It is logical that th e results of the post manipulation blocks of the current study would be consistent with the previous literature that did not account for language mode, since those studies recruited bilinguals as bilinguals and their participants were likely to be at least partly in a bilingual mode.
EXECUTIVE CONTROL AND LANGUAGE MODE 45 There was no such confirmation of the language mode element of the hypothesis in regard to the type of cue presented in the flanker task. The language groups did not differ from each other before or after the break. The cluster of the valid cue and no cue conditions with faster reaction times than the misleading cue condition suggests that the visual cue may have activated the orienting network but not the alerting network ( see the diagram of the attentional system in Figure 1 ) If the cues had activated the alerting network, the valid and misleading cues would both have facilitated faster reaction times than no cue at all, and if the cues had activated both the alerting and the orienting networks, the valid cue would have facili tated faster reaction times than both the misleading cue and no cue with the difference between the misleading and no cue conditions depending on the relative strengths of the networks. One could claim that t he verbal task does not provide evidence for t he mode hypothesis either; the relationship between second language proficiency and reaction times to the LDT diminished greatly after the break. Before the break, reaction times to purely English words, pseudowords and homographic cognates were all positi vely correlated with at least one of the subscales of the Interagency Language Roundtable proficiency test, but after the break only pseudowords maintained this relationship. This weaker difference between the participants with higher second language profi ciency (which was highly correlated with self identified bilingualism) and those with lower proficiency could be interpreted as a problem for the mode hypothesis. If participants perform more similarly when the bilinguals are supposedly in a bilingual mode does that not suggest that either the groups do not differ at all, or that the mode hypothesis is invalid? Not necessarily; t here is another interpretation. The participants with higher second language proficiency demonstrated a processing speed disadva ntage in the LDT before the break i.e. the greater the proficiency the longer the reaction time latency. This processing speed disadvantage makes sense in the context of the past literature, for example Michael
EXECUTIVE CONTROL AND LANGUAGE MODE 46 and Gollan (2005, as cited in Bialystok & Craik, 2010) found that bilingual participants experienced a processing speed disadvantage for picture naming, and made more errors when doing so, than monolingual participants. In the context of language mode, however, the phenomenon of interest is not th e processing speed disadvantage but the disappearance of this disadvantage after the break. Although the change could be interpreted as above, as support for the lack of difference between groups, it can also be seen as a relative bilingual advantage in a bilingual mode. The bilingual participants can be seen, rather than as slower than monolingual participants, as overcoming a disadvantage a disadvantage that has been documented previously to catch up with monolingual participants only when they were p laced in a more bilingual sociolinguistic context. This interpretation would be consistent with the claims of Bialystok (1999) and other researchers who have found that bilingual participants adapt to a new rule faster than monolingual participants do, suc h as in the dimensional change card sort task. In this study, the bilinguals initially were at a disadvantage, but with practice and a more bilingual mode, they adapted and this resulted in an improvement in performance. The processing disadvantage was la rgely found in the first half of the experiment, and while the original language mode hypothesis might predict that the two language not less, after the mode manipulation, i t is also possible that the mode manipulat ion activated some mechanism that produced a boost to processing speed in the participants with higher levels of second language proficiency. The success of the language mode manipulation is integral to this study, and future research should address mode from this experimental standpoint, and investigate not only manipulated. If the current study did manage to manipulate the language mode of the participants, its findings suppo rt and expand upon Grosjean (2001) original predictions made. Not only did bilingual participants overcome their processing speed disadvantage in
EXECUTIVE CONTROL AND LANGUAGE MODE 47 the LDT after the manipulation, they also showed a reduced congruence effect on the Simon and flanker tasks after the break. This would allow an extension of the idea that sociolinguistic context affects language activation to include the claim that mode affects not only linguistic processing but also performance on non linguistic tasks The fact that the monoli ngual and bilingual participants did not differ in their responses to the different types of cues in the flanker task, either before or after the break, is problematic for this explanation, but (2005) task, the removal of the auditory cue and the immediate presentation of the stimulus after the cue rather than a 400ms interval between the two. This could easily be remedied in a follow up study and should not overshadow the other find ings. Of course, i t is possible that the language proficiency tests during the break did not Grosjean (2001) descri bes mode as extremely sensitive. In this case, pr actice would be the likely cause of the homogenisation of the reaction times in the LDT possibly a regression towards the mean after 90 trials but could not explain the differentiation between the monolingual and bilingual participants after the break on the non verbal tasks. Whether the current manipulation was successful or not, in future research care must be taken to closely examine which elements trigger mode changes. For example, a monolingual researcher conducting the first half of the study and a bilingual researcher administering the language proficiency tests, speaking with the participants in mixed language, and conducting the second half of the study, would be a strong mode manipulation, but would require more resources than were available for this study. Manipulations that are less resource intensive but just as strong would be extremely valuable. With this information, much more can be learnt about the mode hypothesis and whether it is an accurate model of the bilingual mind Future research should also investigate further those tasks in which bilingual participants demonstrate an advantage to establish the domain general processes involved.
EXECUTIVE CONTROL AND LANGUAGE MODE 48 More evidence from the combination of linguistic and non linguistic tasks could lead to generalisable conclusions about the fundamental mechanisms involved in language use by monolinguals and bilinguals alike. This information could be used to adapt education, mass media and interpersonal communication to best suit each group s needs, which could prove ext remely useful in an increasingly globalised world with growing numbers of multilingual people.
EXECUTIVE CONTROL AND LANGUAGE MODE 49 References Adesope, O. O., Lavin, T., Thompson, T., & Ungerleider, C. (2010). A systematic review and meta analysis of the cognitive correlates of bilingualism. Review of Educational Research, 80 (2), 207 245. Bialystok, E. (1999). Cognitive complexity and attentional control in the bilingual mind. Child Development, 70I (3), 636 644. Bialystok, E. (2006). Effect of bilingualism and computer video game experience o n the Simon task. Canadian Journal of Experimental Psychology 60 68 79. Bialystok, E. (2007). Cognitive effe cts of bilingualism: How linguistic experience leads to cognitive change. International Journal of Bilingual Education and Bilingualism, 10 (3), 21 0 223. Bialystok, E. & Craik, F. I. M. (2010). Cognitive and linguistic processing in the bilingual mind. Current Directions in Psychological Science, 19 19 23. Bialystok, E., Craik, F. I. M., Klein, R., & Viswanathan, M. (2004). Bilingualism, aging, and cognitive control: Evidence from the Simon task. Psychology and Aging, 19 290 303. Bialystok, E. & Feng, X. (2009). Language proficiency and executive control in proactive interference: Evidence from monolingual and bilingual children and adults. Brain & Language, 109 93 100. Carlson, S. M. & Meltzoff, A. N. (2008). Bilingual experience and executive functioning in young children. Developmental Science, 11 (2), 282 298. Callejas, A., Lupiez, J., Jess Funes, M., & Tudela, P. (2005). Modulations among the alerting, orienting and executive control networks. Experimental Brain Research, 167 27 37. Cedrus Corporation. Superlab (Version 4) [computer software]. www.cedrus.com Colzato, L. S., Bajo, M. T., van den Wildenberg, W., Paolieri, D., Nieuwenhuis, S., L a Heij, W., & Hommel, B. (2008). How does bilingualism improve executive control? A
EXECUTIVE CONTROL AND LANGUAGE MODE 50 comparison of active and reactive inhibition mechanisms. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34 (2), 302 312. Costa, A., Hernndez, M., Cost a Faidella, & Sebastin Galls, N. (2009). On the bilingual Cognition, 113 135 149. Costa, A., Hernndez, M., & Sebastin Galls, N. (2008). Bilingualism aids conflict resolution: Evidence f rom the ANT task. Cognition, 106 59 86. Dijkstra, T. & van Hell, J. G. (2003). Testing the language mode hypothesis using trilinguals. International Journal of Bilingual Education and Bilingualism, 6 (1), 2 16. Elston Gttler, K. E., Gunter, T. C., & Kotz, S. A. (2005). Zooming into L2: Global language context and adjustment affect processing of interlingual homographs in sentences. Cognitive Brain Research, 25 (1), 57 70. Emmorey, K., Luk, G., Pyers, J. E., & Bialystok, E. (2008). The source of enhanced cog nitive control in bilinguals: Evidence from bimodal bilinguals. Psychological Science 19 1201 1206. Fennell, C. T., Byers Heinlein, K., & Werker, J. (2007). Using speech sounds to guide word learning: The case of bilingual infants. Child Development, 78 1510 1525. Fox, E. (1996). Cross language priming from ignored words: Evidence for a common representational system in bilinguals. Journal of Memory and Language, 35 (3), 353 370. Garbin, G., Sanjuan, A., Forn, C., Bustamante, J. C., Rodriguez Pujadas, A., Belloch, V., ... vila, C. (2010). Bridging language and attention: Brain basis of the impact of bilingualism on cognitive control. NeuroImage, 53 1272 1278. Gerard, L. D. & Scarborough, D. L. (1989). Language specific lexical access of homographs by bil inguals. Journal of Experimental Psychology: Learning, Memory and Cognition, 15 (2), 305 315.
EXECUTIVE CONTROL AND LANGUAGE MODE 51 Hernndez, M., Costa, A., Fuentes, L. J., Vivas, A. B., & Sebastin Galls, N. (2010). The impact of bilingualism on the executive control and orienting networks o f attention. Bilingualism: Language and Cognition, 13 (3), 315 325. Keuleers, E. & Brysbaert, M. (2010). Wuggy: A multilingual pseudoword generator. Behavior Research Methods, 42 (3), 627 633. Keuleers, E., & Brysbaert, M. Wuggy [computer software]. Availabl e from http://crr.ugent.be/programs data/wuggy/downloading and installing Macizo, P., Bajo, T., & Martn, M. C. (2010). Inhibitory proccesses in bilingual language comprehension: Evidence from Spanish English interlexical homographs. Journal of Memory and Language, 63 232 244. Martin Rhee, M. M. & Bialystok, E. (2008). The development of two types of inhibitory control in monolingual and bilingual children. Bilingualism: Language and Cognition, 11 (1), 81 93. Moreno, S., Bialystok, E., Wodniecka, Z., & Alai n, C. (2010). Conflict resolution in sentence processing by bilinguals. Journal of Neurolinguistics, 23 564 579. Morton, J. B. & Harper, S. N. (2007). What did Simon say? Revisiting the bilingual advantage. Developmental Science, 10 (6), 719 726. Namazi, M & Thordardottir, E. (2010). A working memory, not bilingual advantage, in controlled attention. International Journal of Bilingual Education and Bilingualism, 13 (5), 597 616. Prior, A. & MacWhinney, B. (2010). A bilingual advantage in task switching. Bil ingualism: Language and Cognition, 13 (2), 253 262. Schwartz, A. I. & Aras da Luz Fontes, A. B. (2008). Cross language mediated priming: Effects of context and lexical relationship. Bilingualism: Language and Cognition, 11 (1), 95 110.
EXECUTIVE CONTROL AND LANGUAGE MODE 52 Van Heuven, W. J. B. Schriefers, H., Dijkstra, T., & Hagoort, P. (2008). Language conflict in the bilingual brain. Cerebral Cortex, 18 2706 2716. Wang, Y., Kuhl, P. K., Chunhui, C., & Dong, Q. (2009). Sustained and transient language control in the bilingual brain. NeuroIma ge, 47 414 422.
EXECUTIVE CONTROL AND LANGUAGE MODE 53 Appendix A Interagency Language Roundtable Language Proficiency Self Assessment Tests
EXECUTIVE CONTROL AND LANGUAGE MODE 54
EXECUTIVE CONTROL AND LANGUAGE MODE 55
EXECUTIVE CONTROL AND LANGUAGE MODE 56
EXECUTIVE CONTROL AND LANGUAGE MODE 57
EXECUTIVE CONTROL AND LANGUAGE MODE 58
EXECUTIVE CONTROL AND LANGUAGE MODE 59
EXECUTIVE CONTROL AND LANGUAGE MODE 60 Appendix B Demographic questionnaire Age: Biological sex: Do you consider yourself to be bilingual? If so, which language (s) do you speak other than English, and at what age did you start learning this language? In which environments do you speak or hear this language? For example, at home with parents, during religious services, when visiting friends. What percent age of your everyday communication with others would you say is in this language? Try to think about how often you speak the language during any given week, and how long you spend speaking the language in each instance. If you do not consider yourse lf bilingual, which languages do you speak other than English? If Please indicate which language you believe you are most proficient in.
EXECUTIVE CONTROL AND LANGUAGE MODE 61 Appendix C LDT Stimuli from before the break Purely English words F armer Music Noose Throw Thie f Paper Receive Cards Light Shoe Ball Tennis Replace Get Book Rob Home Wound Shove Weeping Knot Chew Goodbye Out Bracelet Castle Lift Street Charity Share Homograph cognates Central Motor Media Facial Civil Jade Favor Canal Pedestal Dental Final Solo Col lar General Sol Homograph non cognates Lean Case Tire Den Valor Par Actual Fin Familiar Red Come Gusto Tender Van Ten Pseudowords Feban Abe Sile Sen Daror Sar Anctua Fien Gasaliar Hed Orre Chunto Sonder Gan Dien Cran Nabel Perpame Acore Plutul Ferbo ur Atbar Encea Spudal Shillo Dran Cocar Lolal Lapar Soveal
EXECUTIVE CONTROL AND LANGUAGE MODE 62 Appendix D LDT Stimuli from after the break Purely English words Shine Now Erase Poker Lollipop Bow Ditch Empty Fire Money Hear Write Present Together Recent Finish Gate Such Rope Divorce Shov el Tongue Beach Field Fat Begin Pull Tree Speak Live Homograph cognates Clan Nasal Perfume Aroma Plural Fervor Altar Idea Frugal Villa Plan Color Total Labor Social Homograph non cognates Sale Son Meter Mate Era Dime Sane Primer Mayor Quite Soy Pan H ay Arena Sin Pseudowords Samps Mon Peler Magged Elo Dites Sab Chiper Mactor Quike Sote Paps Hagged Acete Rin Cerjal Cetor Mesio Vanial Cirrel Brane Facher Aal Pevastas Vintal Vilal Roro Ellar Neceras Dol