Abstract
Research questions:
The Moral Foreign Language Effect is a phenomenon in which individuals exhibit lower moral engagement (i.e., are simultaneously less deontological and less utilitarian) when using a foreign language compared to their native language. This paper reports on three experiments involving bilingual participants to investigate the impact of foreign language on moral judgment in self-sacrifice dilemmas.
Design:
In three experiments, we asked N = 425 participants whether they would self-sacrifice to save other people. In Experiment 1, the people to be saved were strangers. In Experiment 2, they were relatives. Experiment 3 replicated the first two with the addition of the control condition with the option to sacrifice a stranger instead.
Data and analysis:
Experiments 1 and 3 showed no effect of language on willingness to self-sacrifice. In Experiment 2, the participants using a foreign language were less willing to sacrifice themselves.
Conclusions:
Language has no consistent effect on participants’ willingness to sacrifice themselves or others.
Originality:
Contrary to previous studies on the issue, we controlled both for the beneficiaries (strangers vs. relatives) and the method of saving them (self-sacrifice vs. sacrifice of a stranger).
Significance:
Our results cast doubt on the robustness of the moral Foreign Language Effect.
Limitations:
We limited our research design to two language pairs. Failure to replicate Experiment 2 results in Experiment 3 could be an effect of the different language pairs used in both experiments.
Keywords
Introduction
The foreign language effect
Chances are that you have heard of a trolley dilemma before: there is an oncoming train that will kill five innocent people on the tracks unless you manage to stop it or divert it somehow. In the best-known version, you find yourself next to a switch. If flipped, it would divert the train onto a different track with one person on it. It would then kill one individual, instead of five. You are faced with a difficult choice: would you flip the switch or let the five people die? (Foot, 1967). This hypothetical scenario has been used to study people’s moral intuitions for over 50 years. But did you know that if you try to solve this dilemma in a foreign language (FL), you might be less apprehensive to let the one individual die? That would be caused by the phenomenon referred to as the moral Foreign Language Effect (Costa, Foucart, Hayakawa, et al., 2014; Geipel et al., 2015). Previous work showed that the effect might arise due to increased emotional distance (Costa, Foucart, Arnon, et al., 2014), weaker social norms activations (Geipel et al., 2015), or changes in the strength of moral inclinations (Hayakawa et al., 2017; Muda et al., 2018). In the current research, we focus on the last aspect, that is, we test how changes in the strength of moral inclinations (i.e., deontology and utilitarianism) impact judgments when using an FL. Our contribution to understanding the moral Foreign Language Effect is based on a novel design (i.e., self-sacrificial moral dilemmas) in which both moral inclinations lean toward the same choice in such scenarios, making it impossible to confound a decrease in deontology with an increase in utilitarianism. We believe our design is the most suitable to detect whether using an FL affects moral judgment and decision-making regardless of which of its two components it affects. To more deeply explain our assumption, below, we describe how the understanding of the moral Foreign Language Effect has evolved and how our research can answer still unresolved issues.
Initial research on the Foreign Language Effect found that people using an FL were more willing to accept harm to maximize outcomes—consistent with utilitarianism (Cipolletti et al., 2016; Costa, Foucart, Arnon, et al., 2014; Geipel et al., 2015). Researchers initially theorized that such behavior may result from increased preference for utilitarianism, which results from greater involvement in information processing and greater focus on the consequences of an action, or from decreased emotional reactions to causing harm. When using classically presented moral dilemmas, such as the switch dilemma presented above, it is impossible to distinguish which mechanism plays a greater role and influences the decision change. This is because, in dilemmas of this type, the deontological and utilitarian choice is arbitrarily negatively correlated, that is, maximizing results (supporting utilitarianism) always involves causing harm (violating deontology). More utilitarian choice may be the result of both a greater importance of outcomes (which would indicate greater preference for utilitarianism), and the effect of a lower aversion to causing harm (which would indicate a reduced preference for deontology), or it may be the sum of any changes in both tendencies acting in the final effect. As such misconception may arise, forcing participants to decide between deontology and utilitarianism lacks real-life relevance and exhibits artificial characteristics (e.g., Bauman et al., 2014; Kahane, 2015). Thus, more recent research presents a more nuanced view of moral decision-making and, in turn, of the influence of an FL on the responses in moral dilemmas.
Recent research on moral judgments is based on two models that measure deontology and utilitarianism as independent inclinations. The two models are: Process Dissociation model (PD) (Conway & Gawronski, 2013) and the Consequences, Norms, Inaction model (CNI) (Gawronski et al., 2017). Both models consider sensitivity to harm and sensitivity to outcomes as separate parameters, rather than two ends of one moral continuum. Sensitivity to harm is conceptualized to guide deontological choices (in the PD model this is reflected with the D parameter and in the CNI model this is reflected with the N parameter). Sensitivity to consequences of one’s actions is conceptualized to guide utilitarian choices (in the PD model this is reflected with the U parameter and in the CNI model this is reflected with the C parameter). The critical difference between PD and the CNI is that in addition to the sensitivity to norms and to consequences, CNI also includes the inaction (I) parameter. This new parameter is conceptualized to capture the preference for inaction in moral dilemmas, that is, the preference to accept the option that does not require taking any action.
In both models, the two relevant moral inclinations are personality traits and are relatively stable over time. However, they are differently cued by a decision problem, so that peoples’ decisions vary across formally identical but semantically different moral problems. The research conducted using the PD and CNI models found that participants using an FL, across large sets of scenarios, made on average similar decisions as in their native language (NL). However, particular components of each decision changed: in some studies, FL participants were less concerned both with the consequences of their actions and with moral norms (Białek et al., 2019; Hayakawa et al., 2017; Muda et al., 2018). Other times, only one of the inclinations decreased in the FL: sensitivity to norms (Hennig & Hütter, 2021) or concern for consequence (Barabadi et al., 2023).
Despite these advancements, moral dilemmas used in the moral Foreign Language Effect research typically have the structure of a trolley dilemma and artificially pit concern with moral norms (i.e., deontology) against concern with consequences of one’s actions (i.e., utilitarianism). Thus, such studies weigh the relative strength of the deontological and utilitarian moral considerations.
Another issue with prior research is that most of it presents participants with uniquely the footbridge dilemma (Thomson, 1976), limiting the generalizability of the Foreign Language Effect. The footbridge dilemma differs from the switch dilemma described above. Instead of flipping a switch, it involves pushing a heavy man on the tracks to stop the trolley. Using just this dilemma is problematic because different dilemmas activate deontological and utilitarian considerations to different degrees, a feature which Baron et al. (2012) call difficulty. Some dilemmas—like the switch scenario—activate utilitarian considerations more, and thus produce overwhelmingly utilitarian decisions. Some other scenarios—like the footbridge scenario—activate deontological considerations more and produce overwhelmingly deontological decisions. Already the first study that introduced the moral Foreign Language Effect reported that it occurs only in the footbridge dilemma, but not in switch (Costa, Foucart, Hayakawa, et al., 2014).
Nevertheless, people do not make moral decisions solely based on contextual factors; they obviously have their own moral preferences and activate corresponding deontological or utilitarian inclinations/considerations more eagerly. In line with our reasoning and previous findings (Białek et al., 2019; Hayakawa et al., 2017; Muda et al., 2018), using an FL should decrease the importance of both inclinations to a similar level, thus producing no effect (see Figure 1). However, because certain scenarios (e.g., footbridge scenario) activate deontological inclination more, using an FL will affect this inclination more than utilitarian inclination. Consequently, using an FL might lead to the deontological choice being made less often (see Figure 1).
The influence of foreign language usage on moral decision-making.
To evaluate our predictions, we decided to test the moral Foreign Language Effect in dilemmas where one gets the option to sacrifice oneself to save several people or do nothing and let them die. In a regular trolley dilemma, the decision to sacrifice one person is typically seen as a more utilitarian choice (i.e., choosing to save more people), whereas the decision not to sacrifice the one person can be seen as deontological (as it follows the “do not kill” deontological social norm). In the self-sacrifice variant of the dilemma that we use in the experiment, the situation is more complex. Both moral inclinations—utilitarian and deontological—lead toward the same choice: self-sacrifice (Polacek, 2017). However, they differ in their reasoning: For utilitarianism, self-sacrifice is obligatory (the life of one person cannot outweigh the lives of five). For deontology, it is a supererogatory act (one that is not required but is morally good). Importantly, neither moral inclination prohibits the act of self-sacrifice. Nevertheless, individuals will often refrain from self-sacrifice due to self-interest and the natural desire for survival. In this scenario, moral inclinations are pitted against self-interest (Figure 1).
There are compelling reasons to expect that both moral inclinations (utilitarian and deontological) will become less salient when presented in an FL (Białek et al., 2019; Hayakawa et al., 2017; Muda et al., 2018). In typical sacrificial dilemmas, such as the switch scenario, this decrease in both inclinations often results in no observable effect of using an FL. This is because the two inclinations oppose each other, and their simultaneous reduction tends to cancel out any overall effect.
Our introduction of self-sacrificial scenarios provides a unique and insightful approach to this problem. In these scenarios, both moral inclinations cue the same response (self-sacrifice), so a decrease in both inclinations should produce a measurable main effect—specifically, a reduced likelihood of choosing self-sacrifice in an FL compared to the NL. This experimental design allows us to distinguish between three possible outcomes: (a) a decrease in both moral inclinations when using an FL, which would be evidenced by a reduced tendency to choose self-sacrifice; (b) no change in moral inclinations when using an FL, which would result in no observable effect across any of the scenarios; and (c) increase in utilitarian inclination, which would be evidenced by an increased tendency to choose self-sacrifice. Thus, our self-sacrificial dilemmas offer a more sensitive test of the Foreign Language Effect on moral decision-making, potentially revealing effects that might be masked in traditional sacrificial dilemmas.
Experiments
Overview and methodological approach
We conducted three experiments (total N = 425). Data and materials for all experiments are available at https://osf.io/9r8qf/?view_only=f45fdd1d7a654ed38c539c0d5cd52ba5. In each experiment, we presented proficient bilinguals (in Experiments 1 and 2—Polish native speakers, in Experiment 3—Ukrainian native speakers) with a set of moral dilemmas concerning a situation where one can self-sacrifice to save other people—in Experiment 1, to save strangers; in Experiment 2, to save relatives. In Experiment 3, there were three sets with (1) self-sacrifice to save strangers, (2) self-sacrifice to save relatives, and (3) regular moral dilemmas, that is, without the option to self-sacrifice and instead with situations where strangers would die unless some action was taken, which would instead kill one individual. The participants’ task was to rate whether they would be willing to self-sacrifice. In Experiment 1, with a Yes/No answer; in Experiments 2 and 3, with answers ranging from 1 to 7, where 1 = No, I definitely wouldn’t do it and 7 = Yes, I definitely would do it. In Experiment 1, contrary to our hypothesis, the language of presentation did not influence the participants’ moral choices. In Experiment 2, we observed the main effect of language—the participants were less likely to sacrifice themselves in an FL. Experiment 3 showed no significant difference between languages apart from the footbridge scenario.
There were two previous studies on the topic of the Foreign Language Effect on self-sacrificial dilemmas: Fernández-Sanz et al. (2023) and Romero-Rivas et al. (2022). In Fernández-Sanz et al. (2023), the researchers presented Spanish-English bilingual children with moral dilemmas in either L1 or L2. The dilemmas varied in utilitarianism, aversiveness, and whether they allowed for self-sacrifice. The children’s responses were analyzed for utilitarian choices and willingness to self-sacrifice. The authors concluded that children were more willing to self-sacrifice in L2 (and made more utilitarian judgments). In Romero-Rivas et al. (2022), the researchers presented Spanish-English bilingual students with one of the two moral dilemmas (either the footbridge or the switch) in either L1 or L2. They had three options to choose from: do nothing, sacrifice someone, or sacrifice self. The participants also completed an empathy scale along the two dilemmas. The conclusion from the study points to how people are more willing to sacrifice themselves in their FL (likely due to a lowered self-distance).
A critical difference between the two studies and ours was that the two presented three choice options which proved troubling methodologically: sacrifice self, sacrifice a stranger, or do nothing, while our study only had two: sacrifice self or do nothing. Choosing to sacrifice a stranger despite the option to self-sacrifice is something different from sacrificing a stranger given no other option to save the five others. It combines utilitarian considerations (rejecting the do-nothing option) with self-interest (favoring other-sacrifice over self-sacrifice). This three-option design weighs the relative importance of deontology, utilitarianism, and self-interest. We believe our design, which does not provide an option to sacrifice others, more robustly tests the moral Foreign Language Effect by contrasting any moral considerations against self-interest.
Experiment 1
The first experiment used four dilemmas pitting the life of five strangers against the life of the decision-maker.
Participants
There were N = 114 (59 female, 55 male; mean age = 20.6 years; age range = 18–76 years) participants. All of them were native speakers of Polish. Participants were recruited through an external research firm and received 20 PLN (about 5 USD) for participation in the 10-minute study run in laboratory settings. We excluded additional N = 19 participants because they had low self-assessed proficiency scores (below 5 on a 10-point scale), lived in an English-speaking country for 10 months or more, were not native speakers of Polish, or had parents who were not native speakers of Polish. These exclusion criteria roughly follow selection and exclusion criteria in the past work (e.g., Białek et al., 2020; Costa, Foucart, Arnon, et al., 2014; Muda et al., 2023). The reason for establishing such exclusion criteria was to minimize the impact of confounding variables that influence the perceived emotionality of language (e.g., Pavlenko, 2004, 2012). They include FL proficiency, age of FL acquisition (Harris, 2004; Harris et al., 2006), frequency of FL use (Degner et al., 2012), order of FL acquisition (Dewaele, 2010), and the context of acquisition (Dewaele, 2004, 2008). To control for the potential impact of these variables, we excluded data from individuals who might have developed strong emotional associations with their FL (e.g., living in an English-speaking country; Opitz & Degner, 2012) or who acquired their NL in unusual contexts (e.g., as dual language learners having parents who speak different languages at home, which may influence the formation of cognitive, emotional, and executive functions; Barac et al., 2014; Hammer et al., 2014). Detailed information regarding the participants’ experience with the FL can be seen in Table 1.
Details of the participants’ self-reported English proficiency and age of foreign language acquisition.
Materials and procedure
Participants were randomly assigned to perform the task either in Polish or English. We informed participants that the study involved decisions concerning moral dilemmas and gave them an opportunity not to participate if they felt uncomfortable (there were no dropouts due to this fact). The participants were presented with four moral scenarios presented in random order. For each scenario, the participants were asked whether they would decide to sacrifice themselves (Yes, I would/No, I wouldn’t). The dilemmas were translated into Polish and backtranslated by the bilingual authors of the study. An exemplary dilemma reads as follows:
You are negotiating with terrorists to save a group of five tourists that have been captured. The leader of the terrorists gives you the choice: if he shoots you, the five tourists will be safe; if you decide not to self-sacrifice, the terrorist will kill five tourists and you will be safe. Would you decide to self-sacrifice? [Yes, I would./No, I wouldn’t]
Results
We conducted a Generalized Mixed Model analysis (Table 1). The decision (yes/no) was a dependent variable, while scenario (transplant/hospital/terrorist/metro) and language (NL/FL) were factors. We clustered them by the participants’ ID. On average, 36% of participants decided to self-sacrifice when using their NL while 30% when using an FL. The analysis showed that there was no main effect of language, F(1, 106) = 0.89, p = .348, but there was an effect of the scenario, F(3, 336) = 7.72, p < .001 (see Figure 2). Critically, the interaction showed to be nonsignificant, F(3, 336) = 1.29, p = .279. To conclude, language did not affect the willingness to self-sacrifice for a group of strangers.
Results of Experiment 1.
Discussion
Contrary to our hypothesis, the language of presentation did not influence the participants’ moral choices, even though the willingness to sacrifice varied between scenarios. We decided to switch the five strangers from Experiment 1 to five relatives in Experiment 2. In such a case, utilitarianism and deontology should still favor the same choices (i.e., self-sacrifice), while the motivation to act should increase. This is because almost all accounts of morality suggest people have a special moral duty toward their kin, either because of kin selection (Hamilton, 1964), or because of a greater valuation of kin lives (Engelmann & Waldmann, 2022). Thus, in Experiment 2, we hypothesized that the participants would choose to self-sacrifice less in the FL than in the NL condition, while overall the tendency to do so should be higher than in Experiment 1.
Experiment 2
Participants
There were N = 123 participants (67 male, 56 female) with a mean age of 21.2 years (age range = 19–29 years). As in Experiment 1, participants were recruited through an external research firm and received 20 PLN (about 5 USD) for participation in the 10-minute study run in laboratory settings. They were all native Polish speakers. In addition, there were 61 participants whose answers were not considered in the analysis because their self-reported proficiency was lower than 5 on a scale of 1 to 10, their NL was not Polish, or they lived in an English-speaking country for at least 10 months (the same exclusion criteria as in Experiment 1). Detailed participants’ FL demographics can be found in Table 2.
Details of the participants’ self-reported English proficiency and age of foreign language acquisition.
Materials and procedure
The procedure and materials were the same as in Experiment 1—we used a set of four moral dilemmas based on those from Experiment 1 modified in such a way that now instead of five strangers, it was five relatives who were going to die unless a self-sacrificial action was undertaken. In addition, instead of Yes/No answers, participants rated their willingness to self-sacrifice ranging from 1 to 7, where 1 = No, I definitely wouldn’t do it and 7 = Yes, I definitely would do it. We hoped this scale would better capture the potential influence of language on moral judgments.
Results
We ran a Linear Mixed Model (Table 3). In the analysis, we took the value (1–7) as a dependent variable, we took group (native/foreign) and variable (scenario—transplant/hospital/terrorist/metro) as factors, and we clustered them by the participants’ ID. The analysis showed a main effect of the scenario (p < .001) as well as a significant effect of the language (p = .011). On a scale from 1 to 7, mean scores were as follows: in the NL, participants were inclined to self-sacrifice by 5.64 (SD = 1.42) points on average, while in the FL, they were inclined to self-sacrifice by 5.12 (SD = 1.65) points on average (see Figure 3). None of the other effects or interactions proved to be significant.
The results of the three experiments.
Results of Experiment 2.
Discussion
There was a main effect of language; this time, as expected, the participants were less likely to sacrifice themselves to save five relatives, when deciding in an FL. Experiments 1 and 2 showed the results to be inconsistent, so to obtain a better view of the situation, we conducted Experiment 3 that combined their design. This time, the participants saw either the dilemmas where they dealt with strangers or those where they dealt with relatives. In addition, we decided to include the original footbridge scenario, alongside three other scenarios without self-sacrifice. This would allow us to compare the response patterns in scenarios with and without the option to sacrifice the self.
Experiment 3
Participants
We analyzed data from N = 188 participants (126 female, 62 male) with a mean age = 19.6 years (age range = 16–36 years). As in Experiments 1 and 2, participants were recruited through an external research firm and received 20 PLN (about 5 USD) for participation in the 10-minute study run in laboratory settings. This time, all participants were native Ukrainian speakers. There were 27 additional participants whose responses were excluded from the analysis, because they were not native speakers of Ukrainian, their parents were not native speakers of Ukrainian, self-reported proficiency in English lower than 5 on a scale of 1 to 10, or lived in an English-speaking country for 10 months or longer (the same exclusion criteria as in Experiments 1 and 2). Detailed participants’ FL demographics can be found in Table 4.
Details of the participants’ self-reported English proficiency and age of foreign language acquisition.
Notably, the design of this experiment is more complex than that of Experiments 1 and 2 and includes an additional within-subject factor (type of dilemma) and its interaction with language. To compensate for this increase, we increased the number of dilemmas presented to each participant (from 4 to 12) and the sample size by about 50%, hopefully compensating for the additional complexity of modeling.
Materials and procedure
This time we used three sets of moral dilemmas, each consisting of four dilemmas: (1) self-sacrifice for strangers, (2) self-sacrifice for relatives, and (3) regular moral scenarios. Self-sacrificial scenarios are taken from Experiments 1 and 2, respectively. Regular moral scenarios did not include the option to self-sacrifice. Instead, they referred to situations where five strangers would die unless some action was taken, which would instead kill one individual. Participants rated their willingness to self-sacrifice (or in the case of regular scenarios, to sacrifice a stranger to save other people) ranging from 1 to 7, where 1 = No, I definitely wouldn’t do it and 7 = Yes, I definitely would do it.
Participants were randomly assigned to perform the task either in Ukrainian or English. Each participant answered all sets of moral dilemmas (sets presented in random order and dilemmas within sets presented in random order).
Results
We ran a Linear Mixed Model (Table 3). In the analysis, we took the value (1–7) as a dependent variable, we took dilemma type (strangers/relatives/regular), scenario (metro/organs/hospital/terrorist), and language (native/foreign) as factors, and we clustered them by the participants’ ID. The analysis showed the main effects of the dilemma type (p < .001) and the scenario (p < .001), but there was no main effect of language (p = .639). Moreover, the interactions between dilemma type and scenario (p < .001) and scenario and language (p = .026) were significant. None of the other interactions or effects was significant. Figure 4 presents the descriptive statistics.
Results of Experiment 3.
Discussion
We observed no main effect of language. In other words, neither the decision to self-sacrifice for relatives, for strangers, nor to sacrifice a stranger for other strangers was affected by the language used. However, language worked differently for specific scenarios. In fact, there was only one particular scenario where it had significant influence on the responses—the footbridge scenario. What is so unique about this scenario is that it consistently produces the moral Foreign Language Effect, which remains a mystery. This body of evidence suggests we should not be generalizing the effects of FL use from this scenario to all moral decision-making.
General discussion
Across three experiments with 425 bilingual participants, we found no consistent evidence for the Foreign Language Effect. That is, the willingness to self-sacrifice for strangers was never influenced by the language used. Willingness to self-sacrifice for relatives slightly decreased in Experiment 2 but not in Experiment 3. We also observed no moral Foreign Language Effect in scenarios other than the footbridge scenario. To remind our considerations on previous results on moral Foreign Language Effect, regardless of whether we expect a decrease in deontological, utilitarian inclination, or in both, we would see lower willingness to self-sacrifice. Had utilitarian inclination increased, we would have seen greater willingness to self-sacrifice. None of those effects was robustly observed.
Our findings indicate that the moral Foreign Language Effect varies significantly across different moral vignettes. The footbridge scenario, heavily explored in this line of study, may not be representative of the general population of moral dilemmas. Speculatively, using an FL may only affect decisions in very easy or very difficult scenarios—those that strongly cue one inclination over another. To ensure the generalizability of a psychological effect like the Foreign Language Effect, we need to demonstrate its robustness across various stimuli (Yarkoni, 2022). If not, we must refine the theory to explicitly state which problems will and will not be affected.
Studies using a single or only a few similar dilemmas may be unsuitable for drawing far-reaching conclusions about the moral Foreign Language Effect. Their results may heavily depend on scenario specifics, including emotional intensity (e.g., personal action causing harm vs. indirect action; Greene et al., 2009), complexity (e.g., number of people to be sacrificed or saved; Cao et al., 2017), and personal relevance (e.g., characteristics of people to be sacrificed or saved, such as ingroup or outgroup members; Swann et al., 2010).
Instead of using single dilemmas, future research should focus on more varied moral problems, such as designs based on PD or CNI models. These models employ multiple dilemmas, reducing the impact of individual scenario.
We do not know what the true effect is of using an FL on the willingness to self-sacrifice to save kin. Our results are inconsistent across the two experiments presented here: it decreased in Experiment 2 and was unaffected in Experiment 3. For what we can say with greater certainty is that using an FL does not increase the willingness to self-sacrifice. It may, however, within the utilitarian realm, promote self-sacrifice over the sacrifice of a stranger (Fernández-Sanz et al., 2023; Romero-Rivas et al., 2022). Is there a moral Foreign Language Effect? We think so. Previous meta-analyses (Circi et al., 2021; Del Maschio et al., 2022; Stankovic et al., 2022) have demonstrated the robustness of the moral Foreign Language Effect, when the decision was to sacrifice a stranger. Two prior studies reported greater willingness to self-sacrifice as well (Fernández-Sanz et al., 2023; Romero-Rivas et al., 2022). Our largely null findings suggest that using an FL does not produce such clear-cut effects as initially proposed, and the effect itself is a complex phenomenon than originally presented in the literature. Of course, our study has its limitations, so we do not claim our line of results is the only one that is correct. One of them is that we limited our research design to two language pairs. Another limitation may lie in the sample population, which consisted mainly of university students. This demographic may exhibit a tendency toward a more self-focused perspective, potentially influenced by their life stage and relative lack of experiences that foster a broader, other-oriented mind-set, such as parenthood or long-term caregiving responsibilities. Such life experiences have been shown to enhance empathy (Maximiano-Barreto et al., 2020), which could, in turn, affect willingness to engage in self-sacrificial behavior. Consequently, the findings may not fully generalize to populations with more diverse life experiences or age groups. Also, failure to replicate Experiment 2 results in Experiment 3 could be an effect of a different language pair used in both experiments. More research on a greater set of dilemmas, more diverse language pairs, and with larger sample sizes is needed to better understand this phenomenon.
To sum up, we believe that presenting the moral Foreign Language Effect as uniquely causing bilinguals to make more utilitarian choices is an overstatement. A large body of research using more nuanced methods of investigating the moral judgments suggests the moral Foreign Language Effect may result from a larger decrease in sensitivity to norms than in sensitivity to consequences, rather than from increased alignment with utilitarian principles like instrumental harm or impartial beneficence (Kahane et al., 2018). However, this account failed to predict that people’s willingness to self-sacrifice will not be affected. Maybe, both moral considerations indeed became less prominent, but unexpectedly, the motivation for self-interest was also affected by the use of an FL. Future research could consider this possibility and estimate its strength to compare it across languages of presentation.
In a broader context, we believe that research on the FL effect needs serious revisions. Not only in the domain of moral judgments but also in general. Initial research showed that people seem to benefit from using an FL (e.g., Costa, Foucart, Arnon, et al., 2014; Del Maschio et al., 2022; Hadjichristidis et al., 2015; Keysar et al., 2012; Muda et al., 2024). However, using an FL also seems to negatively influence logical reasoning (Białek et al., 2020) and the ability to accurately discern real and fake news (Muda et al., 2023). To further complicate matters, sometimes the Foreign Language Effect fails to arise at all, such as in reasoning problems (Mækelæ & Pfuhl, 2019), the use of information in gambling decisions (Muda et al., 2020), or in gambles with verbally described probabilities (Borkowska et al., 2023). Given this complexity, we should refrain from citing the Foreign Language Effect as driving more utilitarian moral judgments and having a uniformly positive impact on decision-making.
Footnotes
Author contributions
R.M. contributed to conceptualization, methodology, investigation, validation, statistical analysis, writing—review and editing, funding acquisition, and project administration. W.M. contributed to statistical analysis, writing—original draft, and writing—review and editing. A.B. contributed to statistical analysis, writing—original draft, and writing—review and editing. M.B. contributed to conceptualization, methodology, and writing—review and editing.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The current research was supported by grant SONATA BIS 2020/38/E/HS6/00282 from the National Science Centre (NCN, Poland) to Michał Białek. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
