Language Control in Bilinguals

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

Introduction

At least 50% of the world's population is bilingual (Baker, 2011), and having proper language control to easily switch between two languages is imperative. It is understood that bilinguals have two different stores of knowledge for words, one in each language, otherwise known as a mental lexicon. Lexical concepts are necessary to express ideas in a specific language, and at some point, there must be cognitive competition between the two languages in order to pick the correct language. A meta-analysis of previous studies have identified an isolated control network responsible for selecting one language over another, which consists of the pre-supplementary motor area/anterior cingulate gyrus cortex, prefrontal cortex, and left caudate nucleus (Abutalebi, Della Rosa, Ding, Weekes, Costa, & Green, 2013). Speech production requires two mental process before one can articulate in the correct language. First, in the preverbal stage, one must conceptualize, during which you are thinking of something to say. Next, one must express the intended idea by using linguistic tools, and hence, formulate in the verbal stage. However, it is important to note a major difference in cognitive control is between the intention to speak, and actually speaking. Previous studies have not been able to separate the brain regions involved in processing the intent to use the language during the preverbal stage and retrieving the proper lexical information for speech production in the verbal stage. An explicit separation of both processes would show during which stage the competition for choosing the 'correct' language for production occurs, and if cognitive processes are different at these stages.

This fMRI study investigates how the bilingual brain uses cognitive control processes to encodes and act upon the intention to speak one of the two languages. The study uses a task-cueing paradigm to present a separation between neural activity associated with forming the intention to speak from the neural activity of retrieving the specific lexical items/concepts in the selected language. The study predicts having the intention to speak a specific language may suppress the unintended language, and thus show systematic neural activation differences in brain areas earlier during cognitive control, during the intent to speak stage. If competition between languages continues until a late time of cognitive control, during articulation/ actually speaking, there should be systematic neural activation differences in brain areas during this time, not during the intention phase.

Methods

For this study, twenty-one subjects were recruited, twelve female and nine male. Subjects were bilinguals that reported German as their first language and English as their second language, with no learning or language impairment. Language proficiency was assessed by a German version of the Language Experience and Proficiency Questionnaire and a picture-naming task, which tested how extensive their vocabulary was.

The stimuli for this picture-naming task were 120 black and white line drawings of objects retrieved from the International Picture Naming Project, and were available in German and English. The experiment was divided into six runs with forty trials in each run, for a total of 200 trials. Each trial began with a fixation cross for 0.5s, followed by an abstract visual cue telling the subject which language to use for naming the upcoming picture, which lasted 1s. Then another fixation cross was presented for 8s as a delay phase, during which is assumed to be the language intention phase, such that subjects can prepare to respond in the target language which was cued to them. Finally, a line drawing was presented for 3s, at which time subjects were to name the drawing, and the verbal response was recorded by an MRI compatible microphone, alongside an fMRI (Fig. 1). This would allow the experimenters to retrieve the reaction time of each word and word agreement (the proportion of subjects using a specific word to name an object).

Figure 1: Timeline of on trial of the experimental task   

Between each trial, there was a 1s-7s delay. Furthermore, the sequence of languages in the trails was pseudo-randomized so that the same language cue did not appear more than three times in a row. Trials with the same cued language consecutively were called "stay trials" and a trial where the cued language was different from the previous trial was called a "switch trial". It is important to note subjects had behavioral training prior to beginning the actual experiment to ensure proper response to the cued language, and following the specific time frames assigned to each portion of the stimuli. Each subject trained until they were able to correct identify the language associated with each cue at least eight times in a row.

Discussion

This fMRI study investigated how the bilingual brain uses cognitive control processes to encodes and act upon the intention to speak one of the two languages, specifically looking at two different parts of speech production: the intention to speak and the active language execution (actually speaking). The results showed differences in brain area activation during those phases, thus indicating that two different language systems enable language selection during bilingual language production. This further tells us that the activated system during language intention prepares the individual to speak a particular language, but does not rely on cognitive language control. However, the activated system during language execution does rely on cognitive language control, and it is this system that is likely in charge of resolving the competition conflict between the two languages. During the language intention phase, the bilateral precuneus, right superior lateral parietal lobe, and left middle temporal gyrus were more active in switch trials, regardless of German or English being spoken. This is because during these trials, the subject has to reselect the language that has to be produced (going to English in trial #1 to German in trial #2). As a result of having to switch between languages, it is assumed that this would require more cognitive control than using the same language consecutively (but not more than 3 times consecutively in this study), and hence, the brain needs to be set up in order to have this cognitive switch occur. This contributes to our understanding of bilingualism, and how bilinguals, as myself, are able to switch from one language to another. When looking at reaction times, however, due to the 8s delay period between trials, the effect of switching languages on reaction times probably did not have an effect, considering the 8s delay allowed the subject to prepare prior to speaking. The results from the switch and stay trials also did not show any brain region which showed a significant interaction between switch and language, meaning that the effect of switching from language to another is similar across the two languages.

The switch and stay trials also support the idea that specific cognitive control processes regulate the use of one language over another. During the switch trials, there was an effect during language intention and language execution, however not as pronounced as the language intention phase. These results support previous studies that show brain activation been associated with cognitive control during bilingual language production, specifically in the anterior cingulate and the caudate (Abutalebi et al., 2013), which were more active when the subjects spoke in English rather than German. This possibly suggests that speaking in a second language, the one that is not the primary learned language requires more cognitive control. An interesting finding along with this study is that since the German speakers L1 is German, there is a specific patter of brain activity that is not linked to cognitive language control, but instead, a "default mode network" that is negatively correlated with brain area activity for executive function and attention. This means that this default network may not have as much neural activation because it is the first language, and thus does not require as much cognitive processing as an L2 would. More studies need to be done to further investigate this phenomenon.

In conclusion, the findings of this study show that cognitive control is more prominent during language execution compared to language intention, and thus the control during actual speech production and articulation is specific to the language spoken. The earlier prediction of suppression of the unintended language during the intent to speak is compatible with the results, as the brain maintains the intention to use one language without pre-activating other regions. This study has made a helpful contribution to the study of general bilingual processing, cognitive control and attention. As stated previously, at least 50% of the world's population is bilingual, and thus this study focused on bilinguals. However, it may be interesting to further study those who can speak more than two languages, and see how fMRI results compare to bilinguals.

References    

Abutalebi, J., Della Rosa, P. A., Ding, G., Weekes, B., Costa, A., & Green, D. W. (2013).

Language proficiency modulates the engagement of cognitive control areas in multilinguals. Cortex, 49(3), 905-911.

Reverberi, C., Kuhlen, A., Abutalebi, J., Greulich, S.R., Costa, A., Shima Seyed-Allaei,

S., Haynes, J.D.2015). Language control in bilinguals: Intention to speak vs. execution of speech. Brain and Language, 144, 1-9