1. Introduction
Daily, bilingual people confront the need to control their languages and also the challenge of having to perform many tasks in their second language. A large body of research suggests that bilinguals access information from their two languages even in situations where only one language is required (Blumenfeld & Marian, Reference Blumenfeld and Marian2007; Dijkstra & Kroll, Reference Dijkstra and Kroll2005; Hoshino & Thierry, Reference Hoshino and Thierry2011; Kroll, Dussias, Bice & Perrotti, Reference Kroll, Dussias, Bice and Perrotti2015; Kroll & Stewart, Reference Kroll and Stewart1994; Macizo, Bajo & Martín, Reference Macizo, Bajo and Martín2010). As a result, bilinguals need to negotiate their languages to avoid competition and must select the more appropriate language for a given context (Morales, Gómez-Ariza & Bajo, Reference Morales, Gómez-Ariza and Bajo2013; Morales, Yudes, Gómez-Ariza & Bajo, Reference Morales, Yudes, Gómez-Ariza and Bajo2015; Green & Abutalebi, Reference Green and Abutalebi2013). This, in turn, influences language production and comprehension (Ma, Li & Guo, Reference Ma, Li and Guo2016; Pérez, Hansen & Bajo, Reference Pérez, Hansen and Bajo2019; Roessel, Schoel, Zimmermann & Stahlberg, Reference Roessel, Schoel, Zimmermann and Stahlberg2019). Thus, there is a vast literature on second language (L2) reading comprehension indicating that bilinguals are less efficient and/or poorer comprehenders in their second than in their first language (L1; for a revision see Melby-Lervåg & Lervåg, Reference Melby-Lervåg and Lervåg2014). Most of this literature has focused on exploring the underlying abilities related to reading comprehension such as word reading, vocabulary, and working memory (Droop & Verhoeven, Reference Droop and Verhoeven2003; Geva & Siegel, Reference Geva and Siegel2000; Lesaux, Koda, Siegel & Shanahan, Reference Lesaux, Koda, Siegel and Shanahan2006), but also high cognitive processes such as prediction or updating that may also determine the success and/or the cost of comprehending in L1 and L2 (Foucart, Romero-Rivas, Gort & Costa, Reference Foucart, Romero-Rivas, Gort and Costa2016; Pérez et al., Reference Pérez, Hansen and Bajo2019). The results of the later studies suggest that the cognitive patterns shown in L2 discourse comprehension are similar to those shown in L1 comprehension, but that there are qualitative differences in the ERP components elicited by incongruent information, indicating extra processing in updating information in L2 comprehension (Foucart et al., Reference Foucart, Romero-Rivas, Gort and Costa2016). In addition, Pérez et al. (Reference Pérez, Hansen and Bajo2019) also found that differences in executive control differentially affected L1 and L2 text comprehension, with higher proactive control being predictive of L2 comprehension compared to the L1, which required a more balanced proactive/reactive control. Overall, these studies suggest that the ability to generate predictions, detect incongruences (i.e., monitoring comprehension), and update information to accommodate new information, is costlier in the L2 than in the L1.
Interestingly, research has also shown that encoding information in the L2 has effects that go beyond language processing in the purely linguistic sense (Bialystok, Dey, Sullivan & Sommers, Reference Bialystok, Dey, Sullivan and Sommers2020; Morales et al., Reference Morales, Yudes, Gómez-Ariza and Bajo2015; Rosselli, Loewenstein, Curiel, Penate, Torres, Lang, Greig, Barker & Duara, Reference Rosselli, Loewenstein, Curiel, Penate, Torres, Lang, Greig, Barker and Duara2019; Schroeder & Marian, Reference Schroeder and Marian2014). For instance, it has been shown that decision making is modulated by the language in which people are reasoning (Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner & Keysar, Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014; Costa, Corey, Hayakawa, Aparici, Vives & Keysar, Reference Costa, Corey, Hayakawa, Aparici, Vives and Keysar2019; Hayakawa, Costa, Foucart & Keysar, Reference Hayakawa, Costa, Foucart and Keysar2016; Hayakawa, Tannenbaum, Costa, Corey & Keysar, Reference Hayakawa, Tannenbaum, Costa, Corey and Keysar2017). Explanations for these results include reductions in a) emotional responses (Costa et al., Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014), b) mental imagery or c) the access to episodic information when bilingual people work in a foreign language (Hayakawa & Keysar, Reference Hayakawa and Keysar2018). Moreover, the impact of being bilingual has been reflected across other domains such as visual attention (Chabal & Marian, Reference Chabal and Marian2015), perception of multisensory emotions (Chen, Chung-Fat-Yim & Marian, Reference Chen, Chung-Fat-Yim and Marian2022), and long-term memory (Marian, Bartolotti, van den Berg & Hayakawa, Reference Marian, Bartolotti, van den Berg and Hayakawa2021). Consequently, given the implications of being bilingual on a high range of cognitive domains, we would expect that working in a L2 has an influence in a wide set of real-world phenomena. This idea is especially relevant if we think that people who speak more than one language constantly face different linguistic contexts that force them to use one language or the other (or even both at the same time) while they perform different tasks in day to day life. Therefore, it is relevant to explore how using a non-native language may modulate performance in different cognitive tasks that are also used in daily life.
In this regard, recalling future intentions plays an essential role in everyday experiences. Prospective memory (PM) allows us to create intentions and to execute them in the future. Many critical actions – such as taking medications at the proper time, getting to an appointment on time, doing the shopping when needed, or taking the cake out of the oven before it gets burned – depend on efficient PM functioning. In a PM task, participants are asked to carry out an ongoing activity (e.g., object naming) while maintaining the intention to perform a certain action (prospective intention) when they encounter a specific contextual cue (e.g., pressing a specific key on the keyboard when the object is of a specific colour). Thus, participants might receive instructions to name objects as they are presented on the screen (i.e., ongoing task), and to remember to stop naming when the presented object is in a particular colour (e.g., the colour “red”) in which case they have to press a specific key (i.e., prospective action). Successfully remembering a prospective action involves monitoring the time or the context to perform the prospective task and switching from the ongoing task to the prospective task (Bisiacchi, Schiff, Ciccola & Kliegel, Reference Bisiacchi, Schiff, Ciccola and Kliegel2009; Scullin, Mullet, Einstein & McDaniel, Reference Scullin, Mullet, Einstein and McDaniel2015). Similarly, prediction, monitoring, and switching abilities are also engaged during language processing especially in bilingual situations (Beatty-Martínez, Navarro-Torres, Dussias, Bajo, Guzzardo Tamargo & Kroll, Reference Beatty-Martínez, Navarro-Torres, Dussias, Bajo, Guzzardo Tamargo and Kroll2020; Martin, Thierry, Kuipers, Boutonnet, Foucart & Costa, Reference Martin, Thierry, Kuipers, Boutonnet, Foucart and Costa2013; Moreno, Bialystok, Wodniecka & Alain, Reference Moreno, Bialystok, Wodniecka and Alain2010). Bilingual people must engage these abilities to choose the correct language in each situation (Declerck, Grainger, Koch & Philipp, Reference Declerck, Grainger, Koch and Philipp2017; Adamou & Shen, Reference Adamou and Shen2019). Thus, one might expect that the bilingual experience in monitoring and switching would modulate the cognitive processes that emerge during prospective remembering. In this line, López-Rojas, Rossi, Marful and Bajo (Reference López-Rojas, Rossi, Marful and Bajo2022) found differences in prospective memory between bilinguals and monolinguals. Specifically, they explored how different bilingual experiences (i.e., age of acquisition and the linguistic context in which the bilingual was immersed) modulated the performance in a PM task and the neural correlates associated to prospective processing. Hence, participants with different linguistic history were asked to complete a PM task with prospective cues varying in distinctiveness (focal vs non-focal cues). Additionally, brain activity during the task was recorded to explore ERP components related to prospective recall (N300 and P3b). Differences in the wave amplitudes between ongoing and prospective trials in the N300 and P3b components have been associated to efficient monitoring and updating strategies (West, Reference West2011). Results by López-Rojas et al. (Reference López-Rojas, Rossi, Marful and Bajo2022) showed larger differences between the ongoing activity and the prospective intention in the N300 and P3b components for early bilinguals compared to late bilinguals or monolinguals, suggesting enhanced monitoring for early bilinguals. The fact that these differences were found in the more difficult PM conditions also suggest that early bilinguals adapted their monitoring processes to the requirements of the task. Similar ERP patterns – that is, ERP differences between ongoing activity and prospective intention depending on the monitoring capacities of the group – have also been found when comparing children, or older people with younger adults (Cejudo, López-Rojas, Gómez-Ariza & Bajo, Reference Cejudo, López-Rojas, Gómez-Ariza and Bajo2022; Hering, Wild-Wall, Falkenstein, Gajewski, Zinke, Altgassen & Kliegel, Reference Hering, Wild-Wall, Falkenstein, Gajewski, Zinke, Altgassen and Kliegel2020).
Hence, López-Rojas et al. (Reference López-Rojas, Rossi, Marful and Bajo2022) showed how the bilingual experience influences prospective memory processes when bilinguals performed the PM task in their L1. However, they did not manipulate the language in which they completed the PM task, and therefore, they could not assess whether the differences were also modulated by whether the task was executed in the L1 or the L2. In the present study, we aimed to investigate possible differences between monolinguals and bilinguals when performing a PM task in different linguistic contexts. Our critical manipulation specifically assesses whether prospective remembering varies if it is performed in a L1 or L2 continuous task context. We argue that, given that L2 processing is costlier and more resource–consuming (Morishima, Reference Morishima2013; Pérez et al., Reference Pérez, Hansen and Bajo2019), when the PM task is carried out in the context of a L2 ongoing task, the bilingual capacity to dedicate executive control to monitor the environment for prospective cues and to switch from the ongoing task to the prospective intention might be compromised. At the moment, to the best of our knowledge, there are no studies exploring the influence of L2 processing during PM activities.
Therefore, the purpose of the current study is to explore how PM processes such as monitoring and switching are modulated when the ongoing task involves L1 or L2 processing. With this aim, we introduced an adapted version of the text comprehension task developed by Pérez, Cain, Castellanos, and Bajo (Reference Pérez, Cain, Castellanos and Bajo2015) as the ongoing task in the PM procedure. The aim of introducing this task was to manipulate the linguistic requirements of the ongoing activity. This task requires participants to read short narrative texts in which information can be congruent or incongruent with a previous generated inference. When incongruences are encountered, participants need to be able to monitor their comprehension by detecting the mismatch, and subsequently update the initial (but no longer plausible) interpretation, which can occur either early or late in the text, followed by comprehension questions about the texts. Previous experiments have shown that a late updating is more demanding than an early updating (Pérez et al., Reference Pérez, Cain, Castellanos and Bajo2015, 2019), and therefore, this manipulation allows us to explore whether PM is affected by more difficult language conditions. Moreover, the purpose of using this comprehension task it is to resemble the rich and complex linguistic context in which bilinguals are immersed in their daily activities.
In addition, we manipulated the nature of the prospective memory task to vary its cognitive demands. Recent research suggests that the monitoring demands of the PM activity depend on the focality of the cue signaling the prospective task. Focal and non-focal cues differ in the extent to which processing of the cue engages the main features of the ongoing activity (Kliegel, Jäger & Phillips, Reference Kliegel, Jäger and Phillips2008). For example, a focal condition may consist of participants receiving instruction to name the colour of objects as presented on the screen (ongoing task), and remember to stop naming when the presented object is in a particular colour (e.g., the colour “red”) and instead, press a key. In this example, the item “heart” (that is red) is considered a focal cue because identifying the colour is involved in both the ongoing activity and processing of the prospective cue (colour red). In contrast, non-focal PM tasks refer to tasks where processing of the PM cues differ from the processing needed for the ongoing activity. In the previous example, if participants are asked to stop naming when the item on the screen belongs to a given category (e.g., parts of the human body), the category represents a non-focal cue, since the identification of a category differs from the ongoing activity (the colour naming task). This manipulation is theoretically important since it has been proposed that focal cues have higher probability of eliciting “spontaneous retrieval of the intention” without engagement of costly monitoring or retrieval processes (Einstein & McDaniel, Reference Einstein and McDaniel2005; McDaniel & Einstein, Reference McDaniel and Einstein2000; Scullin et al., Reference Scullin, Mullet, Einstein and McDaniel2015), whereas non-focal cues induce monitoring and costlier retrieval. In consequence, non-focal cues (compared to focal) require more attentional prospective resources resulting in more difficult and less accurate performance (Cona, Bisiacchi & Moscovitch, Reference Cona, Bisiacchi and Moscovitch2013; McDaniel, Umanath, Einstein & Waldum, Reference McDaniel, Umanath, Einstein and Waldum2015).
Similar to studies where the L2 modulated performance in linguistic and non-linguistic tasks (Costa et al., Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014; Foucart et al., Reference Foucart, Romero-Rivas, Gort and Costa2016; Hayakawa & Keysar, Reference Hayakawa and Keysar2018; Pérez et al., Reference Pérez, Hansen and Bajo2019), we expected that the language in which the prospective task is performed interacted with the focality of the PM task to modulate performance. Thus, we assumed that monitoring and switching would be more demanding in the L2 than in the L1.
In summary, the purpose of this experiment was to study possible changes in PM processes when the prospective task was performed in the context of an L1 or L2 ongoing task. To this end, monolinguals and bilinguals performed an event-based task in which the nature of the PM cue (focal vs. non-focal) and the linguistic requirements of the ongoing task (early updating vs. late updating) were manipulated. As mentioned, the manipulation of the cue focality (focal vs. non-focal) referred to the PM task, whereas manipulations of the language (L1 vs. L2) and updating conditions (early vs. late) referred to the ongoing task. We introduced additional baseline ongoing conditions in which the ongoing task was performed by itself (varying language and updating conditions) to be able to assess the cost associated with monitoring when the PM task has to be additionally performed. Thus, analyses of the task involve time and accuracy in the comprehension ongoing task as well as time and accuracy on the PM task. These analyses permit assessment of cue monitoring during the ongoing task, and cue detection and execution of the intention in PM trials. In addition, direct comparison between the ongoing trials (ON trial) and PM trials would allow us to assess switching processes. Note also that, by comparing monolinguals and bilinguals, and by having the bilingual participants perform the task in their L1 and L2, we were permitted: 1) to compare possible differences between monolinguals and bilinguals, and 2) to assess bilingual PM performance in L1 and L2 contexts. Also, since PM involves performance in the ongoing task (engaging context monitoring for cue detection), performance in the PM tasks (engaging cue detection, retrieval and implementation of the intention) and performance differences between ongoing and PM (engaging switching processes), we had specific predictions for each of these processes.
Regarding the ongoing comprehension task, we expected to observe better and faster performance when the ongoing task was performed by itself (baseline) compared to when participants performed the PM intention during the ongoing activity (focal and non-focal conditions). This effect would reflect the cost associated with cue monitoring, and we expected it to be larger for non-focal than focal condition. Additionally, we predicted this cost to also vary depending on the language and updating conditions. Specifically, we expected a greater comprehension cost when the updating requirements were introduced late in the text (Pérez et al., Reference Pérez, Cain, Castellanos and Bajo2015, 2019), and more so when the task was performed in the L2. Overall, we anticipated that introducing cognitive demanding conditions, either in the ongoing comprehension tasks (late vs. early updating; and L2 vs. L1) or in the PM (focal and non-focal PM vs. baseline), would affect comprehension – with longer and less accurate performance in the more difficult conditions. Moreover, as long as bilinguals benefit by their context monitoring experience, we would expect better performance than their monolingual counterparts when they were both working in their L1 task.
With regard to the PM tasks, we expected better performance in the focal than in the non-focal trials due to the more demanding monitoring requirements of the non-focal cues (McDaniel & Einstein, Reference McDaniel and Einstein2000). We also predicted an interaction between cue focality and updating and language conditions, which should be reflected in slower and/or less accurate performance in the most difficult L2 non-focal-late updating condition. Regarding monolingual and bilingual comparisons in PM performance, we expected bilinguals to better adjust to the task demands and to reduce differences between the focal and non-focal conditions relative to the monolinguals (see Morales et al., Reference Morales, Gómez-Ariza and Bajo2013, 2015 for a similar conclusion in a different tasks).
Finally, regarding switching – that is, the comparison between the ongoing (ON) and PM trials – we expected an effect of cue focality, indicating that focal cues are more easily detected than non-focal cues (McDaniel et al., Reference McDaniel, Umanath, Einstein and Waldum2015), and an effect of language and updating, where the more demanding L2 language and late updating conditions would result in a costlier switching performance. Again, we expected differences between bilinguals and monolinguals if their language experience influences PM performance.
2. Methods
Participants
This study was approved by the Research Ethics Committee of the University of Granada (registration number, 2262/CEIH/2021). Sample size analysis (power = 90%, α = .05) in G*power (Faul, Erdfelder, Lang & Buchner, Reference Faul, Erdfelder, Lang and Buchner2007) revealed that a sample of 34 participants per group was enough to detect a large effect (Cohen's f effect = .40; Cohen, Reference Cohen1969) in an ANOVA with repeated measures and between/within interactions. In addition, similar studies on the field with a sample size of thirty (or fewer) participants per group (e.g., López-Rojas et al., Reference López-Rojas, Rossi, Marful and Bajo2022) have found medium to large effect sizes, which provides additional evidence for the selected sample size.
A total of 67 young adults from the University of Granada participated in this study and received course credits per participation (mean age = 22.87, SD = 3.59; mean years of education = 17.64, SD = 3.73). Of those, 35 were Spanish monolinguals and 32 Spanish-English bilinguals. Data from another 6 participants were also removed after data trimming (see analysis section). The study was disseminated by means of an institutional emailing list and the institutional online platform for experiments. All participants fulfilled the following criteria: 1) they were between 18-35 years old; 2) they had Spanish as a native language; 3) they reported normal or corrected-to-normal vision; 4) and they had no language disorders. Furthermore, monolinguals were explicitly required to have a very basic, almost null, level in any possible L2. Although they reported having enrolled in the mandatory English courses at school, they all reported being functionally monolinguals (Beatty-Martínez, Bruni, Bajo & Dussias, Reference Beatty-Martínez, Bruni, Bajo and Dussias2021; Perrotti, Reference Perrotti2012), since they had not used English after high school (Granada is a very monolingual community where most people only speak/understand Spanish). In contrast, bilinguals were required to have at least a C1 level in English (corresponding to a proficient use of this language), and they reported to use English frequently in their daily life. Hence, both groups of participants differed extensively in their use of English (see Table 1). Whereas bilingual participants used English daily in different contexts, monolingual participants had a minimum exposure to it.
Mean score and standard deviations in questions about L1 and L2 from the LEAP-Q for the monolingual and bilingual group.
*Indicated significant differences (p < .05) between monolinguals and bilinguals group comparisons.
In addition, we verified participants’ English self- informed proficiency by means of the Michigan English Language Institute College Entrance Test (MELICET). This test consisted of two exercises to assess English grammar through 50 cloze questions with three answer options. Higher scores in this test revealed an advanced knowledge of English grammar. For this reason, participants who obtained a direct score of 35< (out of 50) in this questionnaire were included in the bilingual group (32 participants). Those who scored 25 or less were classified as monolinguals (35 participants). Notice that this questionnaire was applied as a screening test – therefore, those potential participants with an intermediate level of English (scores between 26 to 34) did not qualify to participate in the study and were not invited to participate in the experiment. We also collected data from the LEAP-Q (Marian, Blumenfeld & Kaushanskaya, Reference Marian, Blumenfeld and Kaushanskaya2007) to obtain the history of language use of the bilinguals and monolinguals. The questionnaire consisted of a first section with questions related to the participant's linguistic history, such as listing the languages they know (even in a basic way), percentage of exposure to them, and preference for reading/speaking in each language. In a second section, questions regarding the use and exposure to their native language were presented. Both sections were completed by monolingual and bilingual participants in Spanish. Additionally, those participants who reported knowing a second language at an advanced level, and qualified as bilinguals in the MELICET test, were asked to complete the section about the use and exposure to their L2. That section was presented in English. Table 1 reports a summary of the average scores provided by bilinguals and monolinguals to relevant items from the questionnaire. Finally, a standard digit span task was used to ensure baseline working memory scores were comparable between groups (monolinguals: M = 9.42, SD = 2.98; bilinguals: M = 10.82; SD = 2.54; t (59) = −1.95; p > .05; d = −0.51). All participants gave written informed consent and filled out a sociodemographic questionnaire (e.g., age, illnesses, years of education, etc.). The two groups matched in their sociodemographic characteristics (all ps > .05).
Design
We followed a factorial mixed design using Group (monolinguals vs. bilinguals) as a between subject factor and level of Prospective load (baseline vs. focal vs. non-focal), Updating (early vs. late), Focality (focal vs. non-focal) and Language (L1 vs. L2) as within subject factors. Since language was not completely crossed with all other variables (monolinguals could not perform the task in an unknown language), we performed analyses for monolinguals and bilinguals without considering language, and for L1 and L2 languages considering only the bilingual group.
Procedure and materials
The experimental procedure consisted of a first session of approximately 60 minutes where participants carried out first the MELICET and LEAP-Q (Marian et al., Reference Marian, Blumenfeld and Kaushanskaya2007). Later, they performed the PM task during text comprehension in the L1 and, finally, the digit span working memory task as a control measure. Additionally, in a second session, only for bilinguals, participants completed the PM task in the L2 text comprehension context. For these participants, the order of the two sessions was counterbalanced, to avoid undesirable order effects between the two language contexts. The study was programmed using Gorilla Experiment Builder (Anwyl-Irvine, Massonnié, Flitton, Kirkham & Evershed, Reference Anwyl-Irvine, Massonnié, Flitton, Kirkham and Evershed2020) and conducted online.
PM task during text comprehension
The text comprehension task was adapted from the situation model revision task used by Pérez et al. (2019; see Table 2). In each text, the first two sentences (Introduction) primed a specific inference (for example, the concept of “guitar”). Later, this inference was replaced with new information that required revising their initial interpretation and encoding an alternative inference (i.e., piano). This updating process might occur either in Sentence 3 (early updating) or in Sentence 4 (late updating). Results by Pérez et al. (Reference Pérez, Cain, Castellanos and Bajo2015, Reference Pérez, Hansen and Bajo2019) indicated that late updating produces slower and less accurate performance than early updating texts, and therefore, we used this manipulation to vary the linguistic difficulty of the ongoing task. For this task, we measured, first, the reading times for the complete text. Then, as each text was followed by a comprehension cloze question with three response options (participants were asked to respond by pressing the key corresponding to the correct option), we measured accuracy and response times to the question (see Figure 1). Hence, we assessed the following dependent variables for the ongoing text comprehension task: 1) text reading times; 2) accuracy and response times in the cloze question; 3) reading times in responding to the cloze question.
Example of a cloze comprehension question (comprehension question and three response options) for each block: baseline, focal, and non-focal blocks. The baseline condition served as an ON trial. Whereas, the focal and the non-focal conditions were the PM trials in which a focal PM and a non-focal cue respectively, appeared between the response options (marked in red).
Example of a text trial (late updating vs early updating). Each trial was composed of two introduction sentences, the third sentence with two types of sentences (congruent/incongruent) and the fourth sentence with two conditions of inference updating (non-updated/updated). Finally, a comprehension question with three answer options appeared.
Across the short-texts, we manipulated the baseline and focal conditions. On the one hand, there was a baseline condition where the ongoing comprehension task was performed by itself, with no mention of PM instructions. Notice that this baseline condition permits us to assess performance in the ongoing task without the possible cost of PM instructions. On the other hand, there were two blocks where the PM task was introduced to the participants after explaining the ongoing comprehension task. The PM cue appeared exclusively at the end of six texts from the block (composed of a total of 30 texts) as part of the response options to the cloze comprehension questions (care was taken that across trials the PM appeared unpredictably in different points in the block). In the PM focal condition, participants were instructed to press a specific key whenever the words “necklace” or “bicycle” appeared among the three response options of the comprehension cloze question. These cues were considered focal because they were part of the features of the ongoing activity (identifying the correct word to answer the question) and thus, they were within the focus of attention of the participant. In the PM non-focal condition, participants were asked to press a specific key when a word belonging to the semantic category “profession” or “city” appeared among the response options. These cues were considered non-focal because the detection of the PM cues involved additional processing (semantic classification) that was not required for PM detection. It is important to remark that, for both focal and non-focal cues, participants were asked to interrupt the ongoing activity and execute the prospective intention by pressing a specific key which was different from the keys used to respond to the comprehension question. Performance in the PM task was assessed by accuracy scores and response times in the comprehension cloze questions were the prospective cues (i.e., the words necklace/bicycle or words belonging to “profession” or “city” categories) appeared among the response options; here they are termed “PM” because they correspond to the PM task. Trials in the comprehension cloze question that did not contain the prospective cues were termed ON trials (ongoing trials) and they were used for comparison with the PM trials to assess the cost of the disengaging from the ongoing linguistic task and switching to the PM task. Baseline, focal and non-focal block were counterbalanced across participants. Instructions were provided at the beginning of each block defining the PM conditions (baseline, focal and non-focal).
As stimulus materials, we used 150 experimental texts counterbalanced across the focality conditions and languages. Also, the texts rotated between updating conditions and type of trial (ongoing vs PM). Additionally, 3 practice trials were performed at the beginning of each condition.
Each trial started with a fixation point (‘+’) that remained on the screen until the participant pressed the space bar to see the first sentence. Sentences 1–4 were presented one sentence at a time, and participants were instructed to read each sentence at their own pace, pressing the space bar to display the next sentence. The positions of the correct answer or the prospective cue in the questions were randomized. Texts were printed in black and appeared centred on the screen in a white background. Both groups of participants performed the task in Spanish. In addition, the bilingual group performed the task in English. The order in which the bilingual participants performed the task was counterbalanced across sessions. The order in which bilingual participants carried out the task (first Spanish or first English) did not have an effect either in RTs or accuracy (all ps > .05), so this variable was not considered in the following analyses. Finally, in order to control overall language abilities in the L1, we compared reading times to the first and second sentence (introduction) in the ongoing baseline block for both groups. Results indicated no significant differences between monolinguals and bilinguals (all ps > .05).
Data analysis
Our results are organised into two broad sections: 1) comparison between monolinguals and bilinguals in the shared L1, and 2) comparison between the L1 and L2 contexts, only for the bilingual group. Within each of these sections, subsections referred to whether the comprehension ongoing task, the PM task, or an index for ON-PM switching, were considered.
Ongoing text comprehension
This was assessed by calculating reading times for each text and averaging them for updating (early and late) and PM (baseline, focal and non-focal condition). In addition, we calculated mean accuracy in and response times to the comprehension questions for ongoing trials as a function of conditions. Thus, for these measures, we averaged across trials defining the updating (early and late) and PM conditions (baseline, focal and non-focal) for each group in ON trials. Note that ON trials were cloze questions where the PM cue was not presented among the response options. In order to equate the number of PM and ON trials, we selected the ON trials that preceded PM trials. Thus, for each PM trial in the focality conditions (a total of 6), the previous ON trials (6) were considered for comparison (see Cejudo, Gómez-Ariza & Bajo, Reference Cejudo, Gómez-Ariza and Bajo2019, for a similar procedure).
PM performance
This was assessed by analysing response times and accuracy for PM trials in L1 for each focality condition and group (or L1/L2 language in bilinguals). For these analyses, averaging was done for PM trials (trials where the cloze questions contained the PM cue) and considering the updating (early vs late) and focality conditions (focal vs non-focal) for each group (monolingual vs bilingual) or language (L1 vs L1 condition in bilinguals).
ON-PM switching
This was assessed by the subtraction between ON and PM trials. In this third analysis, we aimed to explore the processes of monitoring and switching that take place during the implementation of the prospective intention. As mentioned, in order to equate the number of trials in the ON and PM condition, for these analyses we also selected the responses to the questions of the ON trials that appeared before the PM trials.
For all the analyses, data trimming was performed by removing participants with accuracy or response times greater than three times the interquartile range in the ON task for at least two levels of prospective load conditions. This resulted in the removal of two monolinguals and four bilinguals.
For simplicity, in the results section, we only included significant effects and interactions. In Appendix A, we detail all the statistics for significant and non-significant effects and interactions.
3. Results
Monolinguals vs Bilinguals in L1
Online text comprehension: Text reading times in ON trials
We averaged total reading times in the text per participants and condition and submitted them to a 3 × 2 × 2 mixed factorial ANOVA with Group (monolingual vs bilingual), Prospective load (baseline vs focal vs non-focal) and Updating (early vs late) as factors (for means and standard deviations per condition see Table 3A). The results of this analysis indicated that the main effect of group was marginally significant, F(1,59) = 3.631; p = .062; 
$\eta _p^2 \;\,$ 0.058, indicating that, in general and independently of the prospective load and updating condition, reading times in the bilingual group were faster (M = 3159, SD = 968) than in the monolingual group (M = 3682, SD = 1474).
Mean score and standard deviations in behavioural data for the monolingual and bilingual group in L1 as a function of the experimental conditions.
Off-line text comprehension: Accuracy to the comprehension question
Averaged ON responses to the cloze questions in ongoing trials were submitted to a 3x2x2 mixed ANOVA with Group (monolingual vs bilingual), Prospective load (baseline vs focal vs and non-focal), and Updating (early vs late) as factors (see Table 3B). Results showed that the main effect of group was significant, F(1,59) = 4.465; p < .05; 
$\eta _p^2\,$ = 0.070, with greater accuracy in bilinguals (M = .88, SD = .15) than in monolinguals (M = .82, SD = .19). Moreover, the main effect of prospective load, F(1,59) = 4.944; p < .05; 
$\eta _p^2 \,$ = 0.077 was significant, indicating that text comprehension in the baseline block led to more accurate responses (M = .87, SD = .18) than comprehension during the focal (M = .82, SD = .18) and the non-focal blocks (M = .84, SD = .18); and the main effect of updating, F(1,59) = 22.719; p < .0001; 
$\eta _p^2 $ = 0.278, where early updating led to better performance (M = .89, SD = .15) than the late updating condition (M = .82, SD = 0.20). Overall, these results showed that PM instructions had a cost in text comprehension as suggested by the higher accuracy in the baseline condition. More importantly, however, the pattern of data demonstrates better performance of bilinguals over monolinguals independently of the prospective load or difficulty of the updating process. This pattern suggests that bilinguals compared to monolinguals might monitor more efficiently the context for appropriate cues independently of the difficulty of the prospective task.
Off-line text comprehension: Response times to the comprehension question
We averaged the response times (for correct responses) per participants and condition and submitted them to a 3x2x2 mixed factorial ANOVA with Group (monolingual vs bilingual), Prospective load (baseline vs focal vs and non-focal), and Updating (early vs late) as independent variables (see Table 3B). A main effect of prospective load, F(1,59) = 15.972; p < .0001; 
$\eta _p^2 $ = 0.213) showed faster response times in the baseline (M = 3033, SD = 1357) when compared to focal (M = 3540, SD = 1414) and non-focal (M = 3752, SD = 1726) conditions. Interestingly, the interaction prospective load by group was also significant, F(1,59) = 4.129; p < .05; 
$\eta _p^2 $ = 0.065, indicating that the difference between monolinguals and bilinguals was significant in the non-focal condition (monolinguals: M = 4171, SD = 1901; bilinguals: M = 3258, SD = 1280; t(59) = 1.952; p < .05; d = 0.51). This suggests that, in general, bilinguals were faster answering to the comprehension question. However, these differences between groups did not appear in the focal (monolinguals: M = 3679, SD = 1390; bilinguals: M = 3376, SD = 1445; t(59) = 0.883; p = .381; d = 0.23) and baseline (monolinguals: M = 3153, SD = 1499; bilinguals: M = 2891, SD = 1168; t(59) = 0.787; p = .434; d = 0.26) conditions. Thus, these results indicated greater efficiency of bilingual people adapting their monitoring abilities to perform the more resource-demanding PM activity. Nevertheless, in the blocks where monitoring was not required (i.e., focal and baseline) both groups performed in a similar way.
Prospective performance accuracy.
Average PM responses to the cloze questions containing the prospective cue were submitted to a 3x2x2 mixed ANOVA with Group (monolingual vs bilingual), Focality (focal vs non-focal), and Updating (early vs late) as factors (see Table 3C). These results showed a significant main effect of group, F(1,59) = 15.129; p < .0001; 
$\eta _p^2 $ = 0.240, with greater accuracy in bilinguals (M = .78, SD = .30) than in monolinguals (M = .51, SD = .40). Interestingly, the interaction between focality and group was also significant, F(1,59) = 9.539; p < .05; 
$\eta _p^2 $ = 0.139, indicating that bilinguals were equally accurate in the focal than in the non-focal condition, t(27) = −1.106; p = .396; d = −4.32, while monolinguals showed greater accuracy for focal cues than for non-focal cues, t(32) = 3.136; p < .05; d = 1.63.
Prospective performance response times.
We averaged response times per participants and condition for prospective cloze trials and submitted them to a 3x2x2 mixed factorial ANOVA with Group (monolingual vs bilingual), Prospective load (focal vs non-focal), and Updating (early vs late) factors (see Table 3C). The results of this analysis indicated that group was the only significant effect, F(1,59) = 7.618; p < .05; 
$\eta _p^2 $ = 0.114, indicating that, in general, bilinguals were faster responding to the PM cues (M = 2847, SD = 1212) compared to monolinguals (M = 3722, SD = 1758).
In sum, the analyses of the prospective task indicate that bilingual participants were not affected by the difficulty of detecting the prospective cues given their high performance in both the focal and non-focal conditions. However, overall, monolinguals showed less accuracy. These results could indicate that bilingual participants may have a better ability to involve cue detection than their monolingual counterparts. This was also reflected in their overall faster response times to the PM.
Switching between ON and PM tasks accuracy.
The switching cost index was submitted to a 2x2x2 mixed ANOVA with Group (monolingual vs bilingual), Focality (focal vs non-focal), and Updating (early and late) as factors (see Table 3D). The main effect of group was marginally significant, F(1,59) = 3.802; p = .056; 
$\eta _p^2 $ = 0.061, with less cost in bilinguals (M = .10, SD = .39) than in monolinguals (M = .26, SD = .46). The other two main effects were significant: focality, F(1,59) = 7.628; p < .05; 
$\eta _p^2 $ = 0.114, indicating greater cost in the non-focal (M = .25, SD = .44) than in the focal block (M = .13, SD = .42); and updating, F(1,59) = 15.198; p < .001
$; \;\;\eta _p^2 $ = 0.205, with greater cost for early updating (M = .26, SD = .44) than for late updating (M = .13, SD = .44). More importantly, the focality by group interaction was significant, F(1,59) = 19.177; p < .0001; 
$\eta _p^2 $ = 0.245, indicating that monolinguals showed greater cost for non-focal than for focal cues, t(32) = −4.368; p < .001; d = −2.15, whereas this effect was not significant for bilinguals, t(27) = 1.030; p = .276; d = 7.96 (see Table 3C).
Switching between ON and PM response times.
We averaged response times per participants and condition and submitted them to a 3x2x2 mixed factorial ANOVAs with Group (monolingual and bilingual), Focality (focal, and non-focal), and Updating (early and late) (see Table 3D). The main effects and interactions did not reach significance.
Overall, when considering switching from the ON task to the PM tasks bilinguals seem to overcome the cost of task switching more efficiently than the monolinguals, and they did so in both focal and non-focal conditions (no difference between conditions), whereas monolinguals evidenced greater cost in the more difficult non-focal condition. This pattern, however, was only evident when looking at the accuracy data and not to the response times.
Bilinguals L1 vs L2
With the aim of exploring the language effect in monitoring cost, prospective performance and switching processes, we ran the same analyses on bilinguals and compared their performance when the task was done in the L1 and the L2.
Online text comprehension: Text reading times in ON trials
We averaged reading times per participants and condition and submitted them to a 3 × 2 × 2 within participants ANOVA with Language (L1 vs. L2), Prospective load (baseline vs. focal vs. and non-focal), and Updating (early vs. late) as factors (see Table 4A). The main effect of language was significant, F(1,27) = 32.366; p < .0001; 
$\eta _p^2 $ = 0.545, indicating faster reading times in the L1 (M = 3159; SD = 868) than in the L2 (M = 4324; SD = 1085).
Mean score and standard deviations in behavioural data for the bilingual group in L1 and L2 as a function of the experimental conditions.
Overall, the language in which the task was executed modulated online text comprehension. Thus, when participants were reading texts in their L1 they were faster independently of the prospective load and updating conditions.
Off-line text comprehension: Accuracy to comprehension questions.
The number of correct responses to the ON task in the cloze questions were averaged per subject and condition and submitted to a 3x2x2 within ANOVA with Language (L1 vs. L2), Prospective load (baseline vs. focal vs. non-focal), and Updating (early vs. late) as factors (see Table 4B). The main effects of language, F(1,27) = 6.532; p < .05; 
$\eta _p^2 $ = 0.195, and updating, F(1,27) = 35.781; p < .0001; 
$\eta _p^2 $ = 0.570, were significant. Importantly, there was a significant interaction of language by updating, F(1,27) = 7.692; p < .05; 
$\eta _p^2 $ = 0.222, indicating that although the updating effect (i.e., higher accuracy in the early than late conditions) was significant in both the L1,(t(27) = −3.070, p < .05, d = −1.70), and the L2 (t(27) = −4.957, p < .0001, d = −4.38), the effects differed in size – that is, there was larger effect size in the L2 than in the L1.
Off-line text comprehension: Response times to the comprehension question
Response times (for correct responses) were averaged per participant and condition and submitted to a 3x2x2 mixed factorial ANOVA with Language (L1 vs. L2), Prospective load (baseline vs. focal vs. non-focal), and Updating (early vs. late) as factors (see Table 4B). The result of this analysis showed a significant main effect of language, F(1,27) = 19.728; p < .0001; 
$\eta _p^2 $ = 0.473, indicating that response times were faster in the L1 (M = 2950, SD = 1261) than in the L2 (M = 3865, SD = 1224).
Altogether, these results indicate that language modulated accuracy and response times to the comprehension questions. Comprehension was faster in the L1 than in the L2. In addition, processing in the L2 led to a larger updating effect – this is to say, L2 comprehension seems to be especially impaired in the most linguistically complex late updating condition.
Prospective performance accuracy
Average PM responses to the cloze questions with prospective cues were submitted to a 2x2x2 within ANOVA with Language (L1 vs. L2), Focality (focal vs. non-focal), and Updating (early vs. late) as factors (see Table 4C). The interaction of language by focality was significant, F(1,27) = 4.278; p < .05; 
$\eta _p^2 $ = 0.137, showing that in L1 there was no difference between focal and non-focal cues (t(27) = −1.106; p = .278; d = −0.24), whereas in L2 the difference was marginally significant (t(27) = 1.966; p = .060; d = 8.00), indicating greater accuracy in focal cues than in non-focal cues (see Table 4C).
Prospective performance response times
We averaged response times per participants and condition and submitted them to a 2x2x2 within factorial ANOVA with Language (L1 vs. L2), Focality (focal vs. non-focal), and Updating (early vs. late) as factors (see Table 4C). There was a significant main effect of language, F(1,27) = 7.125; p < .05; 
$\eta _p^2 $ = 0.209, indicating that, in general, L1 responses to the PM cues were faster (M = 2847, SD = 1212) compared to L2 (M = 3982, SD = 2643). The main effect of language was modulated by the language by updating interaction, F(1,27) = 4.509; p < .05; 
$\eta _p^2 $ = 0.143. This interaction indicated that there was no difference between the late and early updating conditions in the L1 (t(27) = −0.618; p = .541; d = −11.60), while in the L2 this difference was significant (t(27) = 2.103; p < 0.05; d = 5.46), with slower response times in the late than in the early updating condition (see Table 4C).
In sum, in the L1, bilinguals seemed to be able to overcome the difficulties associated with the focality of the cue and the updating requirements of the text. However, when they performed the task in their L2, the cost associated with processing in their less dominant language produced focality effects in accuracy and updating effects in response times.
Switching between ON and PM tasks accuracy
The switching cost index calculated from the PM and ON questions were submitted to a 3x2x2 within ANOVA with Language (L1 vs. L2), Focality (focal vs. non-focal), and Updating (early vs. late) as factors (see Table 4D). The main effect of updating was significant, F(1,27) = 22.409; p < 0.001; 
$\eta _p^2 $ = 0.454, reflecting that the difference between ON and PM trials was smaller in the late updating (M = −.01; SD = .46) compared to the early updating condition (M = .12; SD = .36).
Switching between ON and PM tasks response times
We averaged response times per participants and conditions and submitted them to a 2x2x2 within ANOVA with Language (L1 vs. L2), Focality (focal vs. non-focal), and Updating (early vs. late) (see Table 4D). The main effects and interactions did not reach significance.
All in all, our results showed smaller differences in accuracy between ON and PM trials in the more demanding late updating condition. This effect was however not evident in response times. Interestingly, the effect of updating in accuracy was independent of the language in which the ON task was performed (and language did not yield significant effects), suggesting that switching was not influenced by the language in which the task was performed.
4. Discussion
In the present study we examined the influence of bilingualism on prospective memory. First, we explored possible differences between monolinguals and bilinguals in a PM task, which varied in monitoring demands (baseline –without PM instructions-, focal PM or non-focal PM tasks) and the linguistic requirements (early updating vs. late updating). Second, we compared bilingual PM performance when the task was carried out in the first (L1) and second (L2) language. To this end, monolinguals and bilinguals performed an event-based task in which the nature of the PM cue (focal vs. non-focal) and the linguistic requirements of the ongoing task (early updating vs. late updating) were manipulated. Additionally, bilingual participants performed the PM task in both their L1 and L2. Below, we discuss the findings of our study in the same order of presentation used in the Results section.
Monolinguals vs Bilinguals in L1
Bilinguals showed better performance in text comprehension during the ongoing task with faster reading times (online text comprehension), higher response accuracy and faster response times (off-line text comprehension) to the comprehension questions. In general, this advantage was independent of the monitoring requirements of the prospective task and the linguistic difficulty of the ongoing activity, suggesting that overall bilinguals were more efficient L1 comprehenders than monolinguals when they faced inferential revision during text comprehension. However, regarding response times to the comprehension question, we found that bilinguals were faster compared to monolinguals only in the more resource-demanding PM activity. This suggested a higher ability of the bilingual participants facing activities where monitoring is required. Notice that, this does not necessarily suggest that each of the underlying reading comprehension processes are affected by bilingualism, but that higher-levels processes such as inferencing, monitoring and revision, involved in our text comprehension task were modulated by language use.
Studies comparing differences in text comprehension between monolinguals and bilinguals, Teubner-Rhodes, Mishler, Corbett, Andreu, Sanz-Torrent, Trueswell, and Novick (Reference Teubner-Rhodes, Mishler, Corbett, Andreu, Sanz-Torrent, Trueswell and Novick2016) have also found that bilingual readers show better performance than monolinguals in their offline comprehension of syntactically ambiguous sentences, also suggesting that their previous language experience may enhance general performance in language comprehension. Similarly, Afsharrad and Sadeghi Benis (Reference Afsharrad and Sadeghi Benis2017) also showed that successful L2 learners outperformed unsuccessful L2 learners classified as monolinguals, in a reading comprehension task, and they attributed this reading comprehension advantage to bilinguals’ better use of metacognitive strategies. Moreover, Filippi, Leech, Thomas, Green, and Dick (Reference Filippi, Leech, Thomas, Green and Dick2012) studied the effect of interference in an auditory sentence comprehension task finding that bilingual speakers outperformed their monolingual peers in the more interfering condition, suggesting that better cognitive control abilities in bilinguals might allow them to control the interference. Hence, our results add to the evidence suggesting that the bilingual experience modulates high-level comprehension processes, leading to better performance than the monolinguals.
Interestingly, inferential revision in text comprehension has been linked with different executive functions. For example, a recent study (Pérez, Schmidt, Kourtzi & Tsimpli, Reference Pérez, Schmidt, Kourtzi and Tsimpli2020) showed differences in text comprehension performance due to inhibitory control mechanisms. Specifically, they found that higher compared to lower inhibitory control comprehenders had better performance in a comprehension question when conflicting information was presented in the text. Similarly, inferential text revision requires successful monitoring processes to detect incongruences and to update the mental representation of the text to ensure coherence (Kintsch, Reference Kintsch1998; Perfetti, Stafura & Adlof, Reference Perfetti, Stafura and Adlof2013). Despite the fact that different underlying abilities have been proposed for reading comprehension (Li, Koh, Geva, Joshi & Chen, Reference Li, Koh, Geva, Joshi and Chen2020), we suggest that differences in monitoring abilities are the main factor that explain our findings, given the critical role of monitoring in tracking text coherence during the inferential revision (Pérez et al., Reference Pérez, Cain, Castellanos and Bajo2015) and during the PM task (Hunter Ball & Bugg, Reference Hunter Ball and Bugg2018). Thereby, we interpret that the higher performance found in bilinguals during the execution of the PM trials is explained by greater monitoring abilities. Consequently, it is possible that our bilingual participants engaged the monitoring processes required for high-level text comprehension more efficiently, and therefore, they outperformed monolinguals. This assumption is in line with previous studies in the field of bilingualism and cognitive functions which have shown better monitoring capacity in bilinguals than monolinguals, not only at the behavioral level, but also at the neural level (Abutalebi, Della Rosa, Green, Hernández, Scifo, Keim, Kappa & Costa, Reference Abutalebi, Della Rosa, Green, Hernandez, Scifo, Keim, Kappa and Costa2012; Costa, Hernández, Costa-Faidella & Sebastián-Gallés, Reference Costa, Hernández, Costa-Faidella and Sebastián-Gallés2009; Hilchey & Klein, Reference Hilchey and Klein2011; Morales et al., Reference Morales, Gómez-Ariza and Bajo2013, Reference Morales, Yudes, Gómez-Ariza and Bajo2015; Kousaie & Phillips, Reference Kousaie and Phillips2012, Reference Kousaie and Phillips2017).
As we previously mentioned, this idea is also supported by the findings regarding the prospective task, where bilinguals showed overall faster response times and higher accuracy than the monolinguals. Thus, the bilingual group seemed to overcome the difficulties in prospective performance associated with the monitoring demands of the task, showing similar performance in the more demanding (non-focal) conditions compared with the less demanding (focal) conditions. These findings suggest that bilinguals can adapt their capacities to the demands of the task and engage in prospective processing strategies, enabling them to successfully perform the PM task even in the more challenging conditions. These results support previous studies, similarly suggesting that bilinguals may have better cognitive strategy adjustment than monolinguals (Morales et al., Reference Morales, Gómez-Ariza and Bajo2013, 2015). This pattern of results also resembles previous data on PM and bilingualism that observed how bilingual experience modulates the cognitive processes involved in updating and cue detection to adapt them to the PM task's demands (López-Rojas et al., Reference López-Rojas, Rossi, Marful and Bajo2022). There is also the possibility that the better PM performance in bilinguals (compared to monolinguals) could be explained by a general higher linguistic capacity in the bilingual group and not by their knowledge of a second language. However, given that reading times to the first and second sentence in the ongoing baseline trials did not differ between monolinguals and bilinguals, we consider that L2 learning (and not a general language capacity) is the responsible for their better PM performance.
Similarly, bilinguals also seemed more efficient when switching between activities (ON-PM tasks), as we found smaller switching cost for bilinguals than monolinguals in accuracy. The smaller cost for bilinguals was similar for the focal and non-focal conditions, whereas monolinguals’ cost was affected by the difficulty of the PM task, with greater cost in non-focal than in focal conditions.
The pattern of focality effects observed in monolinguals in their prospective performance and in their switching cost can be easily explained by the Multiprocess framework proposed by McDaniel and Einstein (Reference McDaniel and Einstein2000). According to this framework, focal cues elicit “spontaneous retrieval” of the intention in contrast to non-focal cues that require more costly monitoring processes which in turn results in longer response times and poorer accuracy. However, the lack of focality effects in bilinguals suggested that they overcame the processing difficulties associated with the focality of the cue. These findings support the idea that bilingualism modulates the processes engaged in PM processing.
In sum, when focusing on differences between bilinguals and monolinguals in L1, the general pattern of results showed between-group differences in performance with faster and more accurate responses in bilinguals. More importantly, bilingual participants seemed to be able to overcome the monitoring demands imposed by the nature of the PM cue as compared to monolinguals, who show the usual impairment with increments in monitoring demands.
L1 vs L2 in Bilinguals
Language comparisons in online (short-text readings times) and offline (accuracy and response times in cloze-questions) indicated faster and more accurate reading in the L1 compared to L2 comprehension. This is in line with previous studies in the field of reading comprehension in foreign language (Melby-Lervåg & Lervåg, Reference Melby-Lervåg and Lervåg2014). In our data we found an impairment in text comprehension when working in a less dominant language. Although this effect was independent of inference updating and prospective load when focusing on reading times during online text comprehension and response times to the comprehension questions, accuracy during offline comprehension was modulated by the updating condition. Specifically, between-language differences in accuracy increased for the more demanding late updating condition, indicating that L2 comprehension is selectively impaired in difficult conditions. This pattern of results supports the findings by Pérez et al. (Reference Pérez, Hansen and Bajo2019), who suggested that the efficiency of predictive processes and inferential revision (a highly demanding updating process) is reduced in the L2, compared to the L1. Pérez et al. (Reference Pérez, Hansen and Bajo2019) argued that the differences between languages might be because during L2 comprehension, cognitive resources might mainly engage in lower-level features processing and, consequently, less resources might be available to process conceptual features (Horiba & Fukaya, Reference Horiba and Fukaya2015; Segalowitz, Watson & Segalowitz, Reference Segalowitz, Watson and Segalowitz1995; Yang, Reference Yang2002).
Regarding performance in the prospective memory task, we found that when the task was performed in the L2, bilinguals showed a focality effect with better accuracy for the focal than for the non-focal condition. However, when bilinguals performed the task in L1, cue-focality did not have an effect and there were no differences in PM performance between the focal and non-focal conditions. These effects suggest that bilinguals, when working in their L1, seem to be able to adjust their performance to the monitoring demands of the task to overcome the difficulty of non-focal monitoring. However, when the task is performed in the more demanding L2 language, bilinguals may have fewer resources for strategic processing and adjustment; we see this in the standard focality effect, i.e. lower performance in the non-focal when compared to the focal condition. These results are in agreement with previous studies on the role of working memory (WM) in highly demanding inferential reading tasks. For instance, Alptekin and Erçetin (Reference Alptekin and Erçetin2010) reported differences in WM processing depending on the language. Concretely, participants were more accurate in L1 than in L2, and they argued that reading complex texts in L2 comprehension poses higher demands on WM decreasing their performance in this task. Similarly, and due to the fact that our bilinguals were relatively less proficient in their L2 than in their L1, our results are consistent with previous studies reporting differences between poor and good comprehenders in meta-comprehension monitoring when presented with texts that varied in difficulty (Maki, Shields, Wheeler & Zacchilli, Reference Maki, Shields, Wheeler and Zacchilli2005). Thus, Maki et al. (Reference Maki, Shields, Wheeler and Zacchilli2005) found that good comprehenders were more precise than poor comprehenders when making prospective judgments to difficult texts, where more monitoring was required. However, in the easier texts, there were no differences between the two groups. These results suggest that poor comprehenders’ monitoring capacity was reduced when reading difficult texts. In line with this hypothesis, Han (Reference Han2012) observed worse comprehension monitoring for low-proficiency foreign language readers compared to highly proficient foreign language readers (but see Silawi, Shalhoub-Awwad & Prior (Reference Silawi, Shalhoub-Awwad and Prior2020), that failed to find correlations between monitoring in reading comprehension and language proficiency). Altogether, we suggest that the impairment in PM when bilinguals performed the task in their less dominant language was due to reduced cognitive resources and the need to re-allocate attention to the main task.
Finally, analyses comparing the cost of switching between the ongoing and the prospective tasks when working in L1 and L2 indicated that there were no differences between the two languages. Surprisingly, the only observed effect in this analysis involved smaller cost in accuracy in the late updating condition, indicating that while switching was affected by the conditions of the ongoing task, participants were able to disengage from the ongoing task and switch to the prospective task.
Thereby, the overall pattern of results seems to indicate that L2 processing has a cost in ongoing performance and prospective memory, especially in the more demanding L2 processing conditions and when the PM cue requires effortful monitoring processing. However, once that monitoring results in cue detection, L2 processing does not affect switching to the PM and implementing the action intention.
In sum, our study adds to a wide range of literature suggesting that bilinguals are able to better adjust their cognitive strategies to task demands in comparison to monolinguals (Costa, Hernández & Sebastián-Gallés, Reference Costa, Hernández and Sebastián-Gallés2008; Morales et al., Reference Morales, Gómez-Ariza and Bajo2013, Reference Morales, Yudes, Gómez-Ariza and Bajo2015; Prior & MacWhinney, Reference Prior and MacWhinney2010; see Antoniou, Reference Antoniou2019, for a review). More importantly, our results support and extend previous studies (López-Rojas et al., Reference López-Rojas, Rossi, Marful and Bajo2022) indicating the influence of bilingualism in PM, and the influence of L2 processing on PM performance. Thus, our findings showed an impairment in PM associated with L2 processing. This cost was especially evident in the more demanding conditions suggesting that increments in attentional load due to L2 processing may have impaired the monitoring processes required for prospective remembering.
Prospective memory plays a fundamental role in daily activities. In fact, PM failures (i.e., forgetting to turn off the oven) can result in dramatic consequences when they occur in real life context. Understanding whether and how its underlying cognitive mechanisms can be modulated by bilingualism allows us to accurately address the challenges of today's world. In this line, studies focused on exploring the interaction between different individual characteristics, such us bilingualism and prospective memory, are of especial interest to address different questions. First, these studies add to studies in other domains suggesting that learning a second language (or being immersed in a L1 or L2 context) can impact perception, attention and memory and in turn many real-world phenomenon such as the recall of future intentions (e.g., send a pending email or attending a scheduled meeting). Second, bilingualism can serve as a tool to study in depth the different cognitive mechanisms involved in the prospective recall (e.g., strategic monitoring).
Overall, our findings suggest that some memory processes might vary depending on the linguistic context. Therefore, this research provides a base for future studies about the impact of bilingualism on prospective memory. Further studies are needed to fully understand the dynamic interaction between L1 and L2 processing during recall of future intentions, as well as possible differences of this interaction by comparing different bilingual experiences.
5. Conclusion
In conclusion, the observed results support our hypothesis that differences in prospective processes might be due to bilingualism and the linguistic context in which bilinguals perform a prospective memory task. We observed that in L1 contexts, bilingualism modulates the cognitive processes involved in updating and cue detection to adapt them to the task demands. Bilinguals were able to engage executive control mechanisms to a greater extent than monolinguals, in order to detect and respond to the PM cue. By contrast, recalling future intentions in an L2 context resulted in an impairment of the performance, especially in the more challenging cognitive conditions. These findings suggest the importance of studying how linguistic context modulates certain memory processes.
Acknowledgments
This research was supported by the doctoral research grant FPU17/03378 to Cristina López-Rojas; by grants from the Spanish Ministry of Economy and Competitiveness to Alejandra Marful (PSI 2017-89324-C2-2-P) and to M. Teresa Bajo (PGC2018-093786-B-I00 30B51801); from Junta de Andalucía (A-CTS-111-UGR18 / B-CTS-384-UGR20 / P20_00107) to M. Teresa Bajo; and from MSCA-COFUND Athenea3i to Ana I. Pérez (754446-AT3i). Funding for open access charge: Universidad de Granada / CBUA.
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Competing interest declaration
Competing interests: The authors declare none.
Appendix A
Statistical effects from data analysis in monolinguals vs bilinguals.
Statistical effects from data analysis in bilinguals L1 vs L2.
Open access
