In Eva Hoffman's memoir of being a bilingual and bicultural immigrant to America from Poland, she describes two languages competing and commanding in her head:

“Should you become a pianist?” the question comes in English. No, you mustn’t. You can’t.

“Should you become a pianist?” the question echoes in Polish. Yes, you must. At all costs.

These quotes are striking because we typically regard language as conveying information, not changing it. In the last decade, however, research has shown that answers to questions can depend on the language of the question. For example, when Chinese-English bilinguals were randomly assigned to answer a self-esteem questionnaire in Chinese, they received scores indicating lower self-esteem than those who answered the same questionnaire in English. In this case, cultural differences appear to be the cause. When reading self-esteem questions in English, bicultural respondents are cued to adopt the American self-enhancing bias.

When reading questions in Chinese, respondents may draw on the traditional Chinese virtue of modesty.

It has also been shown that answers on a questionnaire can be influenced by whether the questions are presented in one's native language. Boaz Keysar of the University of Chicago and his colleagues have dubbed this the "foreign language effect." Using examples from economic decision making, including the phenomenon psychologists call loss aversion (see a summary here), they found that bilingual speakers make slightly more rational decisions when evaluating vignettes written in a foreign language.  

New research has now extended these findings into the realm of morality. Albert Costa teamed up with Keysar to study the classic to decisions involving the well-known “trolley” dilemma. A trolley is speeding toward five workmen who will not see the trolley in time to leap out of harm’s way. You are standing nearby, close to a fork in the tracks. You have time to flip a switch to divert the trolley onto a different track, where only one workman is stationed. Is it permissible to flip the switch, thus killing one workman and saving the lives of the five workmen on the other track?

Philosophers and psychologists have long used variations on this vignette to examine different types of moral reasoning, including utilitarian decision making. Utilitarianism specifies that saving more of the workers will amount to the greatest good; hence, it is morally permissible to kill one in order to save five. Most people (close to 70 percent in some studies) say that it’s permissible to flip the switch. When Joshua Greene presented this and similar dilemmas while people underwent brain scans, he found that executive function areas of the prefrontal cortex were active, suggesting that people were consciously reflecting over this decision.

A related dilemma requires taking “personal” action to kill one person in order to spare five. In the "footbridge" dilemma, you are positioned on a footbridge over a train track with the trolley speeding toward five workmen. You are standing next to a large man wearing a heavy backpack. If you push this man off of the footbridge onto the tracks below, he will die, but he and his heavy backpack will stop the trolley, thus saving the five workmen. This dilemma is called "personal" because you are faced with the choice of saving five lives by killing someone with your own hands. Conversely, the flip-switch trolley dilemma is labeled "impersonal" since you are not required to kill anyone directly. Respondents of the footbridge dilemma usually report aversive feelings, such as feelings of horror at the request to push someone on to the tracks to certain death. Using a human being as the means to an end feels wrong, even if the end has the positive consequence of saving lives. Typically, permissibility ratings for these personal dilemmas are far below 70 percent, in the range of 12 to 20 percent. In Greene's neuroimaging study, emotional areas of the brain, rather than the prefrontal cortex.

Based on these findings, we can imagine the results of evaluating trolley dilemmas in a foreign language. If using a foreign language does not involve the brain's emotional centers as strongly as using a native language, then we can expect increased utilitarian responding when presenting the footbridge (“personal”) dilemma in a foreign language. This is what Costa, Keysar and colleagues found. When reading and responding in a native language, only 20 percent of respondents said that personally pushing someone off of the footbridge to save five workmen was permissible. When evaluation was done in a foreign language, the permissibility rate climbed to 33 percent. Native-foreign language pairings included Spanish/English (recruited from the U.S.), Korean/English (Korea) English/French (France) and Spanish/Hebrew and English/Hebrew (both in Israel).

In a second study, Costa and colleagues directly compared responses on the footbridge (personal and thus emotional) and flip-switch (impersonal and, therefore, less emotional) dilemmas. To ensure that cultural norms associated with a specific language were not at play, they compared respondents for whom Spanish was the native language and English the foreign language, and respondents with the reverse pairing (English the native language, and Spanish the foreign language). In every comparison of native to a foreign language, a lower percentage of native language utilitarian responses (average 18 percent) was obtained compared to the foreign language condition (44 percent). However, no foreign language effect was found for the impersonal dilemmas, with “permissible” responses occurring for 71 percent of respondents in both native and foreign language conditions. This suggests that the foreign language effect is specific to moral dilemmas with a strong emotional component.

This research is important for its contribution to two areas of inquiry. The first area is the research tradition of testing bilinguals’ reports that their first language feels more emotional. Do these reports have implications beyond private feelings? The findings of Costa et al. raise the question of whether moral reasoning in a native vs. foreign language differs in real-word situations, beyond the evaluation of vignettes. The second area is decision making. Books such as Kahneman's Thinking Fast and Slow have brought a scientific lens to classic distinctions between emotion and logic and between intuition and deliberation. A crucial question is how language develops so that a native language apparently engages emotion and intuition more readily than does a foreign language.

Although Costa, Keysar and colleagues were admirable in sampling from a variety of native/foreign language pairings, one cultural explanation for their findings has not been examined. Using a native language plausibly induces the feeling that one is reasoning about in-group members. Conversely, a foreign language could signal that the scenario is relevant to strangers, foreigners, or out-group members. Indeed, studies have shown that varying the wording of trolley dilemmas to imply that the workmen on the tracks are of a different ethnicity changed the pattern of responses in complex ways.

The research also raises a question: What about reasoning in a second language that is not foreign? Sojourners who reside for years in a foreign country often become fluent in its language and may ultimately feel as proficient and as emotionally engaged in that language as in their native language. The children of immigrants who grow up with two languages–one spoken at home with family, the other learned at school–frequently become balanced bilinguals. Would these bilinguals resemble the foreign language learners studied by Costa et al. and demonstrate reasoning that is less emotional and more utilitarian using their second language? Although these studies remain to be done, my bet is “no.” I suspect that the foreign language effect is confined to foreign languages and reflects the reduced emotionality that accompanies foreign language learning. Therefore, for balanced bilinguals, or for those who have extensive immersion experience in their second language, I predict little or no difference in moral judgments.

I make this prediction based on my own research. Balanced bilinguals typically report that both of their languages feel emotional. In one study of Spanish-English bilinguals, I recorded electrodermal activity while the bilinguals listened to emotional phrases in their first or second language. For immigrants and sojourners who did not have extensive immersion experience in their second language, skin conductance amplitudes were reduced for their second language. For balanced bilinguals, on the other hand, the skin conductance amplitudes were similar, I concluded that languages reliably elicit emotional arousal and the subjective experience of emotionality when they are learned and used in emotional contexts. Second languages used extensively in an immersion context are subjectively experienced as emotional and elicit autonomic responding consistent with the subjective report of emotional arousal.

Since many readers of Mind Matters are bilingual or multilingual, let me pose this question: Have you had experiences in which your evaluation of a situation was influenced by the language used to explain the situation?

Are you a scientist who specializes in neuroscience, cognitive science, or psychology? And have you read a recent peer-reviewed paper that you would like to write about? Please send suggestions to Mind Matters editor Gareth Cook, a Pulitzer prize-winning journalist and regular contributor to NewYorker.com. Gareth is also the series editor of Best American Infographics, and can be reached at garethideas AT gmail dot com or Twitter @garethideas.