Bilingual memory advantage: Bilinguals use a common linguistic pattern as an aid to recall memory

Participants

The bilingual population in this study consisted of 20 participants who came from the United States of America, but who were located in Cambridge, England, when the data collection was carried out. They were all born in the USA and, for each, both parents were Spanish speakers. They all grew up speaking only or mainly Spanish at home or privately with family. They were schooled mainly in English, although some (3 out of 20) had bilingual education. The mean age of the population was 24 (range: range 22–26). Four of the participants were married to Spanish-speaking partners. Two participants had one child each and were speaking to their children in Spanish only, although the children were attending English-speaking nursery schools. Seven of the bilinguals spoke an additional language but only at a beginner or low intermediate level.

We also have two baseline populations for comparison, 20 English monolingual speakers who were students at the time at either the University of Cambridge or the University of East Anglia (mean age 20, range 19–25, 9 male, 11 female) and 20 Spanish monolingual speakers (mean age 21, range 18–26, 8 male, 12 female), all university students at the University of Zaragoza, Spain. Our monolingual populations were not completely monolingual since they all learned a second language during their education. However, they did not take their language further than the secondary school level and were using it only or mainly when travelling.

All participants performed a short memory test that included looking at two blocks of 20 objects for 20 minutes and listing the items after a 5-minute break for each block (https://faculty.washington.edu/chudler/puzmatch1.html). The recall rate difference was 10% maximum among the participants (the difference in objects remembered was one or two at the most), which indicates that there were no substantial differences in general memory among the participants. Bilingual participants also did an online language proficiency test for Spanish (https://www.transparent.com/language-resources/tests.html). There were no significant differences among the bilingual participants in proficiency for either language. They were included in the study only if their score in each language was 95% and above correct.

Stimuli

The materials and method in this study were identical to those used in Filipović (2018). This would enable subsequent comparisons between second language learners of English and Spanish (Filipović, 2018) with speakers who have equal proficiency in both these languages (current study). The stimuli consisted of video clips filmed with a Sony DCR-HC18E digital video camera, and the experiment was run on a portable PC laptop using Microsoft Office PowerPoint. Each target video clip depicted either an intentional or a non-intentional causation event (e.g. a girl popping a balloon on purpose versus a girl playing with a balloon that popped accidentally, clearly surprising her). There were 10 target videos and 10 filler videos. The filler videos depicted non-causation events, for example, a man drinking coffee. All the target videos were matched for action type (e.g. both intentional and non-intentional breaking events were witnessed by all participants). All the videos were pilot-tested by two native speakers of each language in order to ensure that they were uniformly judged as either intentional or non-intentional. The target videos are described in Table 1.

OPEN IN VIEWER

Table 1.

Experimental video stimuli

Intentional events	Non-intentional events
Girl pops an orange balloon with a needle	Blue balloon pops while a girl is playing with it
Girl pushes a Barbie doll off the bed	A cup is pushed off the sink while washing hands
Woman crushes a plastic cup in her hand	A wooden toy tower gets crushed by a woman falling on it
Woman drops a magazine onto the floor	A pen is dropped during writing
Woman knocks a box off the table using her elbow	A bottle is knocked off the table while rummaging through things located on it

Procedure

The procedure was the same as that followed in Filipović (2013, 2018). Participants were shown the target videos and were asked to describe them in either English or Spanish. They were given three practice videos as a warm-up. Subsequently, after a short intervening break, they were given a recall task in which they were asked whether they had seen particular events.

For the video description, they were told that they would be watching videos of the duration between six and nine seconds, depicting various actions, and that their task was to verbalize what had happened after each video clip (verbalization stage). The bilingual participants heard and used both their languages throughout the experiment. Each performed half of the experiment verbalizing in English while hearing the instructions in Spanish and then the reverse for the other half of the experiment, that is, verbalizing in Spanish while listening to the instructions in English. They were randomly assigned to English-first or Spanish-first groups.

The stimuli were balanced between the two bilingual groups (Spanish-first and English-first respectively) in the following way. One list was created for both groups. The Spanish-first group verbalized four target items and six filler items in Spanish and six target items and four filler items in the second block in English. The English-first bilinguals did exactly the opposite: four target + six filler items in the first block in English and then six target and four filler items in Spanish in the second block. The items were randomized within each block for both groups to avoid recency bias in items.

The intervening distractor task consisted of a 120-second task in which a 10 × 10 grid of randomised letters was shown on the screen, and the participants were asked to count how many of the letters M, N and Z they could see.

The recall task asked two questions about each of the witnessed events (e.g. Did you see a girl with a Barbie doll? Was what happened in that video accidental or on purpose?). The participants were asked to mark their answers on an answer sheet by circling YES if the action in the video was on purpose (intentional) and NO if it was not on purpose (i.e. it was accidental/non-intentional). They were also told that they should not guess and that they should leave a question unanswered if they were not able to recall the relevant information. Responses were classified as incorrect if the participants circled the wrong answer (e.g. intentional instead of non-intentional) or if they failed to give any response (i.e. they left both options unmarked because they could not recall the crucial piece of information; this was the case in 17% of all responses for all groups). Further post-hoc verbalizations were collected from all bilingual participants in order to obtain equal numbers of verbalizations for each video to those of both languages as monolinguals (since the bilinguals had described only four to six of the videos in each language). Bilinguals verbalized each video again in the language different to the one used for the given video during the experiment.

Hypotheses

Following the thinking-for-speaking hypothesis (Slobin, 1996, 1997, 2003, 2006) and the previously reported language-specific effects on monolingual episodic memory (Filipović, 2011, 2013), I expected the Spanish monolingual speakers to have better recall of whether the causation was intentional or not than the English monolingual speakers. This is because the Spanish lexicalization pattern requires speakers to draw the relevant distinctions with regard to intentionality while English leaves this piece of information largely ambiguous (as shown in Filipović, 2013).

I also hypothesized, based on previous research (Filipović, 2013), that both constructions ‘She pushed the bottle off the table’ and ‘The bottle fell off’ would be used by English monolingual participants to refer to either intentional or non-intentional events indiscriminately. By contrast, Spanish monolingual speakers were expected to use intentional and non-intentional constructions consistently for intentional and non-intentional events respectively in order to signal clearly when an event they witnessed was intentional and when it was not.

I based the central hypothesis regarding the bilingual speakers on the recent model CASP for Bilingualism (Filipović & Hawkins, 2018), which predicts that bilinguals would use a shared mechanism for expressing and remembering causation events that would work in both languages (i.e. they would maximize common ground in CASP terms), thus having the same recall outcomes regardless of the language explicitly used for verbalization. In other words, the prediction was that bilinguals would not be influenced by the language of operation and that their memory for causation events would not be less accurate when they verbalized events in English, indicating that resources from both languages are employed as aids to memory regardless of which language is used for speaking. If, on the other hand, the language of operation was a strong factor, we would then expect to see more detail in the description of events and better recall when the bilinguals were speaking Spanish than when they were speaking English.

Verbal descriptions

The monolingual speakers verbalized in line with the predictions: English monolinguals tended not to draw regular and explicit distinctions between intentional and non-intentional events and left their descriptions ambiguous (as in (1a) and (1b)). They explicitly specified events as non-intentional in only 26% of the cases and as non-intentional in 20% of verbalizations (see Appendix). By contrast, the Spanish monolingual participants consistently distinguished between intentional and non-intentional events, as in (2a) and (2b), using SVO constructions for the former in 100% of cases and the se construction for the latter in 100% of cases (see Appendix for more details about responses across all groups):

Bilinguals showed clear differentiation between intentional and non-intentional events in both English and Spanish. Their verbalizations of non-intentional events in Spanish closely resembled that of the Spanish monolinguals – they used se-constructions 100% of the time, either with the affective dative le (‘se le cayó’ = ‘to him/her it fell’; 65% of instances) or without it (‘se cayó’ = ‘it fell; in 45% of the cases). In English, they described non-intentional events by either using SVO constructions with added adverbial modification (49% of instances) or via the use of inchoative constructions (‘The balloon popped’; 47%). For intentional actions, they only used the SVO constructions (100%) in both languages. They added adverbial modification in English in 41% to indicate intentionality (e.g. ‘on purpose’ or ‘forcefully’). In most cases when they did not use additional adverbials, the verbs they used gave a clear indication of intentionality because they referred to actions performed with a considerable force (e.g. ‘The woman crushed/destroyed the cup’ or ‘She threw the magazine onto the floor’; see Ibarretxe-Antuñano, 2012 on the relationship between force and intentionality in lexicalization). The examples of bilingual verbalizations are given below:

Recall

Recall errors concerning the intentionality of the video action from monolinguals and bilinguals in each language were turned into proportions incorrect, because monolinguals had described all 10 videos in their language, while bilinguals had described four videos in one language and six in the other. These proportions were transformed into arcsin scores for the purposes of statistical analyses. I carried out multiple 2 (Group) x2 (Condition-intentional/non-intentional) ANOVA analyses, comparing the two monolingual groups as well as bilinguals speaking Spanish with both monolingual groups and bilinguals speaking English with both monolingual groups. There was a main effect of language in the monolingual English-Spanish comparison (F = 9.90; df = 1.00; p = .003) and a significant Intentionality x Group interaction (F = 13.28; df = 1; p = .001). Independent sample t-tests were significant for group differences in the non-intentionality condition only (t = 4.103; df = 38; p < .001). In the comparison of bilinguals speaking English and English monolinguals there was also a main effect of group (F = 11.68; df = 1.00; p = .002) and a significant interaction of Group x Condition (F = 11.14; df = 1.00; p = .002;). The t-tests revealed that, again, the difference between the groups lay in the recall for non-intentional items only (t = -3.89; df = 38;p <.001). Bilinguals speaking Spanish also differed from the monolingual English speakers in the number of recall errors. There was a main effect of group (F = 10.78; df = 1.00; p = .002) and a significant interaction of Group x Condition (F = 14.96; df = 1.00; p <.001). In this case as well the independent sample t-tests reached significance only for the non-intentional condition (t = -4.33; df = 38; p <.001). No other comparison resulted in significance (p >.05 for bilinguals when speaking Spanish versus Spanish monolinguals, for bilinguals when speaking English versus Spanish monolinguals, and for bilinguals speaking English versus bilinguals speaking Spanish). The means for errors in recall of intentionality (in arcsin scores) are given in Table 2.

OPEN IN VIEWER

Table 2.

Mean percent errors values for recall

Descriptions →Groups	English intentional	Englishnon-intentional	Spanish intentional	Spanishnon-intentional
English monolinguals	12.00	49.55
Spanish monolinguals			15.20	12.45
Bilinguals	13.75	14.20	13.75	10.70

Interpreting monolingual and bilingual results

I investigated whether different patterns in the lexicalization of causation in English and Spanish resulted in differences for recall memory in monolingual and bilingual populations. The results show that Spanish and English monolinguals speakers differed in both how they described and how they recalled events. Spanish speakers use two distinct constructions: one for intentional and one for non-intentional events as predicted by the lexicalization pattern (and as illustrated in the Appendix). English speakers use both SVO and inchoative constructions across the different events, not discriminating explicitly between intentional and non-intentional meanings. They also only very occasionally disambiguate the SVO constructions with adverbials (e.g. ‘She dropped the cup accidentally’). Recall that differences between the two monolingual populations were also found with respect to non-intentional items in the experimental stimuli, while the intentional events did not elicit any significant differences. Intentional events can be described using the same means in both languages (simple transitive SVO constructions) while the unintentional events are described differently in the two languages (typically the same SVO constructions in English vs. se-constructions in Spanish). Our results support findings from previous research that indicated that the differences in both verbalization and memory between English and Spanish participants pertain to the non-intentional event stimuli in particular (Fausey & Boroditsky, 2011; Filipović, 2013). The reason for this may be because prototypical transitivity does indeed involve intentionality, as Hopper and Thompson (1980) point out. A transitive agent is normally an intentional instigator or a causer of the action. There is a strong relationship in English between the subject function (and its sentence-initial position) and the agent role, reflected in the fact that subjects are indeed most frequently agents and, as Fausey and Borodistky (2011) have shown, the focus on agents is more pronounced in English than in Spanish. Thus, SVO structures in English may inspire an agentive interpretation more often than the non-agentive one, and that could have interfered with recall for non-intentional events.

We also may have to allow for the possibility that the mere nature of the stimuli is positively skewed towards the salience of intentionality. For example, in our stimuli, a girl approaches the bed on which a Barbie doll is positioned, looks at the doll and pushes it off the bed, whereby the intention to perform the action is clear. On the other hand, non-intentional events all had an element of surprise and unexpectedness, which may affect the amount of detail that is recorded, especially if language does not encourage the encoding of certain details (such as whether the action was on purpose or not). In the case of non-intentional event stimuli there was more information to be integrated and more inferences to be made in order to encode them fully in memory (e.g. judge the behaviour of the agent before and after the balloon popped). Having clear labels for distinguishing intentional and non-intentional events means that this encoding in language and memory was facilitated for Spanish speakers and Spanish–English bilinguals. Finally, some verbs may have more inherent intentionality in them than others (e.g. push vs. drop; see Filipović, 2018, forthcoming for more discussion). This is a matter for further exploration with an experimental design that would specifically target such contrasts and control for the differences in default interpretations in individual speakers.

It is interesting to point out that bilinguals in this study did not perform differently based on the language of the operation (i.e. the one they were verbalizing in). When we look at the ways bilinguals verbalized during the experiment, we can see that they were drawing explicit distinctions between intentional and non-intentional events in both languages (see Appendix).

It is worth noting that the effect we report here may be weaker or disappear altogether if bilinguals are in a monolingual mode (Grosjean, 2001), due either to the specifications of a communicative situation or to experimental conditions (e.g. when access to one language is blocked through interference). However, when access is uninhibited and especially when both systems are highly active in a communicative situation (or entrenched through habitual use of both), bilinguals are more likely to draw on the relevant lexical, grammatical and conceptual differences from both languages when speaking either of their languages.