Learn a foreign language to reduce bias?

by AShepard1 min read22nd Apr 201237 comments


Personal Blog

Interesting new paper (anyone have a link to an ungated version). Abstract (emphasis added):

Would you make the same decisions in a foreign language as you would in your native tongue? It may be intuitive that people would make the same choices regardless of the language they are using, or that the difficulty of using a foreign language would make decisions less systematic. We discovered, however, that the opposite is true: Using a foreign language reduces decision-making biases. Four experiments show that the framing effect disappears when choices are presented in a foreign tongue. Whereas people were risk averse for gains and risk seeking for losses when choices were presented in their native tongue, they were not influenced by this framing manipulation in a foreign language. Two additional experiments show that using a foreign language reduces loss aversion, increasing the acceptance of both hypothetical and real bets with positive expected value. We propose that these effects arise because a foreign language provides greater cognitive and emotional distance than a native tongue does.

Speakers of multiple languages: have you noticed a similar pattern in your own lives?
37 comments, sorted by Highlighting new comments since Today at 1:03 AM
New Comment

XiXiDu's comment made me realize what the really important implication of this result is: it's not that learning a foreign language is a way to improve rationality (though maybe it is); it's that if you want other people to be rational, don't speak to them in their native language.

Kind of makes me wonder whether the Sequence translations are such a good idea after all...

Kind of makes me wonder whether the Sequence translations are such a good idea after all...

Of course they are! Native English speakers deserve a chance to discuss the Sequences rationally, too. :D

XiXiDu's comment made me realize what the really important implication of this result is:...

I have my doubts that I am neurotypical or otherwise close to the norm, e.g. socially or culturally or when it comes to an education. So I wouldn't assign much weight to my perception here. Although the fact that someone else agreed seems to corroborate what I wrote.

Anything that activates System 2 (for instance, writing questions in a difficult-to-read font) has the same effect.

Sounds interesting; got a citation?

Here's a quote from Kahneman's Thinking, Fast and Slow:

The experimenters recruited 40 Princeton students to take the CRT (Cognitive Reflection Test). Half of them saw the puzzles in a small font in washed-out gray print. The puzzles were legible, but the font induced cognitive strain. The results tell a clear story: 90% of the students who saw the CRT in normal font made at least one mistake in the test, but the proportion dropped to 35% when the font was barely legible. You read this correctly: performance was better with the bad font. Cognitive strain, whatever its source, mobilizes System 2, which is more likely to reject the intuitive answer suggested by System 1.

Is there really a 'system 1' and 'system 2' anyway? Brain is a real-time system, and has to provide gradual de-rating when the time is shorter. When facing a really complicated problem that would take a lot of time to solve, there's "system 2" type reasoning being not even wrong if not given enough time.

"System 1" and "System 2" aren't meant to represent separate sub-minds; they're just a helpful categorization of mental processes. If you want a fuller disclaimer on this useful oversimplification, Kahneman has an excellent one in Thinking, Fast and Slow.

As a trilingual, I would still advise against doing learning a newer language for the sole sake of gaining rationality skills. There are much better and cost-efficient ways to improve rationality besides learning a newer language, which would likely take years, or depending on the language, decades, to fully master.

To answer your question, strangely enough, now that this topic has been brought up, I do have a slight change in personality whenever I switch to a different language conversationally, although I might have to experiment further if it affects whether I am more or less rational.

[-][anonymous]9y 3

I do have a slight change in personality whenever I switch to a different language conversationally

It also happens to me.

It would be interesting to find out how the bias-reducing effect is related to the mastery of the language. I guess that fully mastering the language is not necessary for this effect, and even could reduce it.

My model is that by speaking in a foreign language we have to pay more attention to our thoughts, and we have to make more things explicit. Thus it is more difficult for some biases to be unnoticed. Fully mastering the language would make the biases easier too.

(I am not sure about the cost-efficiency, because I don't know how strong is the effect, and how strong are effects of alternative methods. If the effect is strong, perhaps spending two weeks learning Esperanto using Anki would be worth doing. But I guess the effect is not very strong.)

... but honestly learning languages early on is a good investment for several other reasons. Not only for social and economical reasons, but also due to it becoming a lot harder to learn languages later on in life (please do correct me if this is a myth).

If you on the other hand stay learning languages during your youth, becoming fluent in a language takes less and less time (at least according to my own experience and that of all of the polyglots I know).

Also I'd say that being able to call on different aspects of your personality is a very valuable trait. If you combine the use of standard associations exercises (and hence have an above average control of your emotional state) with keying different associational patterns to different languages you'll achieve a capacity for holistic problem solving and idea generation that I'd expect to be soaring high above the average levels.

Speakers of multiple languages: have you noticed a similar pattern in your own lives?

The most obvious difference to me is that I can tolerate and even enjoy some things in English that I am unable to tolerate in German. For example, when I hear certain kinds of Hip hop or gangster rap in German I am instantly disgusted. When I hear similar lyrics in English then I am merely amused.

I am also much quicker to dislike people who speak German than people who speak English. For example, if I listen to some of what American politicians say I am just amused. If people utter similar things in German then my emotional reaction is much stronger.

FWIW, I'm from Germany and feel the same way.

I wonder if folks from other countries experience comparable phenomena? Or is German just an unusually annoying language?

I'm a native Finnish speaker and I too have a stronger revulsion towards ugly songs when they're sung in Finnish.

My guess is we just don't mind it as much when it's some other language that's being used to create ugliness. Shortcomings by the (perceived) home team often feel like something that might be held against you, hence the need to signal disapproval more strongly.

[-][anonymous]6y 2

I'm half-Ukrainian, half-Russian, and I find pop songs in English basically 'comfortable noise', in French 'beautiful exercises in phonetics' (since French 'r's are strikingly different to what I am used to), in Russian 'depressingly monotonous/how do they even think this is to my tastes?', and in Ukrainian... I either think of them as 'derived from Russian', or wait for my favorite lines without paying attention to the rest:)

This is interesting because I find it counterintuitive. I would have thought that the tendency would be the opposite: that, for a given thought, expressing that thought in German (your native language) would tend to make it more sympathetic to you rather than less (all else being equal).

Data such as this may force me to reevaluate my predictions of the effect one might have on an audience by speaking to them in their native language.

The effect in question is actually attenuation through mediation: most people can attest to the fact that foreign swear words are not perceived as strongly as the native ones, even if the meaning and usage is identical. Having an intermediate step of translation reduces emotional impact. Not surprisingly, the effect wears off after living long enough in the new environment. This has little to do with framing.

I don't know what you mean by "framing" here, but the takeaway for me is that speaking someone's language doesn't necessarily earn you "points".

E.g. XiXiDu doesn't say, "Well, this may be gangster rap, but at least it's in German!"

(In retrospect, I've had opportunities to notice this before. For instance, Amanda Knox seemed to get less "credit" for speaking Italian than I would have expected.)

It sounds like you're not treating foreign language speakers as members of your society, so when they say violent or politically abhorrent things, you don't get worked up about it because they're not plausible threats. They're part of a different community, they're someone else's problem.

Regarding music, I certainly find I can accept mindless bubblegum pop a lot better if it's in Japanese than in my native English. That might be a different phenomenon though.

Regarding music, I certainly find I can accept mindless bubblegum pop a lot better if it's in Japanese than in my native English. That might be a different phenomenon though.

Or simply a different kind of 'threat'.

My native tongue is also German, and I can confirm your statements.

Well, it's just anecdotal evidence, but here are my observations:

When I'm writing text in English, I take more time to express my thoughts, rewriting passages until I think they actually express what I mean. This might make my statements more precise and thought-through than otherwise. I have especially noticed that effect on IRC: While, when I'm in a German channel, I just write my thoughts down as they come, in #lesswrong, when I'm trying to make a point, writing stuff takes longer, and the discussion might have already flown on once I'm ready for submitting my ponderings.

When I'm writing text in English, I take more time to express my thoughts...

I think that I might be an outlier in that respect.

I was really bad in school (or didn't care / couldn't deal with the setting (probably complex reasons)) which resulted in a noticeable deficit when it comes to punctuation and spelling in German. Which would result in a lot of criticism and ridicule and therefore caused me to largely avoid textual discussions in German.

So how come I ended up taking part in English discussions instead?

A relevant fact here is that the first online game I began to play, back in the year 2000, was mainly played by people from the U.S. and Australia. I played it for over 5 years. In addition to my basic knowledge gained at school, that time allowed me acquire a vocabulary and reading comprehension that would serve as a plateau for reading more advanced English content and later books.

The result was that I ended up with good enough English to feel confident to take part in some online discussions. Most of that confidence came from the fact that English wasn't my first language. It was an excuse that I didn't have when writing German.

My German punctuation and spelling is still bad. But I am planning on improving it up to the level that is necessary to write quickly and accurately. That is not a priority though. Math is more important.

So you might actually take more time to express your thoughts in English. But due to the above I am not sure if that is true for me as well. Another reason I doubt this is that not thinking things through is generally one of my biggest shortcomings. But I am trying to fix that.

My personal experience is that I've attached different parts of my personality to my different languages. I'm currently fluent in three and passable in two others. I notice that the part of my personality I attach to a given language depends on a variety of factors, but I can have a marked impact on it by focusing on my associations while learning the language. This is particularly true when I take the language from the classroom to an an experience of immersion.

In this regard I've also noticed that languages of the same root (for me: Danish, English and German) have a tendency to invoke related frames of minds whereas new language families (French, Arabic and Polish) allow me greater freedom in what kind of thoughts are likely to occupy my mind when I use them and hence what kind of person I am in this tongue. It's worth noting that the Polish me is closer to the French me than both the Arabic and Danish parts. I suspect a likeliness in grammar and syntax to have played a role in this.

Now the funny thing is that I didn't really notice this phenomena myself. While living in France I was visited by my parents. Both my dad (who was fluent in French) and my mum (who isn't and wasn't) noticed that everything from my tone of voice, to my subject matter to the way I gestured and moved changed when I changed between languages. American friends, independently, later made remarks in the same direction.

I must admit though that I haven't tried it regarding decision making. It's definitely worth a try and as it is, I'd predict that thinking in French will help me visualise a given situation, Danish would be preferred for extracting the emotional response, English for both estimating it's likeliness of happening as well as turning the given situation into a joke whereas my Polish and Arabic are not quite yet at the same level of abstraction. I'm prone to being extremely flirtatious in Polish and my Arabic has a tendency to call on my altruistic and carefree sides.

Quoting myself on MR:

  1. Does using a foreign language reduce decision-making biases?

Not too surprising. This sounds like the usual cognitive disfluency effect forcing people to shift into type-2 thinking. If they can’t solve it quickly and have to think carefully about each piece, then they’re more likely to reflect and use abstract thinking – which will usually increase normative scores.

Someone here will probably know this better than I do (unfortunately :( ), but I believe there have been a series of studies done on the interplay between language and culture within certain countries--that certain constructions within the language allow for the development of differing types of cultural standards. The example that is coming to my mind is a language like German when compared to English. In English, you tend to be able to interrupt people more often, because you know where a sentence is going (You can go fuck--), whereas in German (I believe), since verbs come at the end (you yourself can go fuck--I know this isn't how it would be set up exactly), you can't interrupt nearly as much. Perhaps there is some ground in the idea that since learning languages forces you to subconsciously accept new modes of communicating, your way of looking at rational choice changes.

Of course, there's probably also an effect of simply needing to be more rigorous in a language that's not your own. My ability to be lazy in English is in part because I've spent a lot of time in this language and feel comfortable assuming things that I probably should not be. Whereas if I began to learn Mandarin, I would have to think about everything I say and do.

It would be difficult to separate cause and effect here. A culture where interrupting people was taboo in the first place might develop a similar syntax.

Speaking of rigor, I've come to love Spanish's subjunctive mood because I can inject doubt into a statement by changing a single syllable, whereas in English I need to mess about with cumbersome disclaimers to say the same thing. Also, if I'm making a subjective value judgment, I can clearly indicate that it's a fact about me, not a fact about the world.

True. It might be possible by seeing how language changes over time in comparison to what evidence of cultural change we have, but it might also be a lot of speculation. In either case though, the frame of the language still arises such that it is significantly different, and you have to work your way through the frame before you can feel comfortable communicating in the language.

I have two native languages which I speak regularly and fluently. I also understand basic Spanish, but I haven't tried thinking about bets or probabilities in that language.

I think knowing two different languages so intimately, especially because my second language, Telugu, is from an Eastern culture, helps me see how values shape vocabulary. Group thinking, applause lights, and tribal politics are more salient when I switch languages, and especially when I try to find the exact word in one language to match to an important concept in the other.

For example, in Telugu, "My son" is used as an insult on par with "Bastard" or even "Motherfucker," something I can't even explain to monolingual English speakers. The idea is that parents have absolute authority over their children, so the speaker is claiming a huge amount of status over the other person, essentially asserting the right to do anything and everything to them, as one would to property. "You're like dirt to me." "Your only value is your use to me, and you have displeased me." (Interesting parallels to "Who's your daddy?", although in the West they tend to emphasize dominance through greater skill rather than parental ownership.)

A very obvious example of tribalism: "Westerner" is used as an insult with the weight that "heretic" or "apostate" would have had 100 years ago in the West. Noticing that people in India spoke this way allowed me to notice the alarming frequency with which people in the US accuse others of being "un-American" or "un-democratic." Looking back, this was probably part of what helped me admit that vague notions of "democracy" may not be the pinnacle of human government, though Enlightenment republics have worked out better than any system so far. I don't think India achieved the peak of cultural perfection back when they codified the word "Westerner" to mean "Bad Person," and it's only a few hundred years later, so I doubt the US has achieved it, either.

Other things I've noticed: "homely" has a very positive connotation, meaning essentially this, which is considered the picture of female perfection. "I" tends to disappear when I'm speaking in Telugu; there is less need for it. "I'm sleepy" becomes "Sleep is coming"; "I'm hungry" becomes, roughly, "Hunger came/was applied to me." Saying "I feel" and "I think" is awkward. I noticed that it might enhance the Mind Projection Fallacy, although I haven't tried. (It could also encourage less selfish thinking.)

I feel a step removed from both cultures as a result of being steeped in them both, which is exactly the place I want to be as an aspiring rationalist. It's a great exercise trying to translate connotations and associations across languages.

For example, in Telugu, "My son" is used as an insult on par with "Bastard" or even "Motherfucker," something I can't even explain to monolingual English speakers.

American English does have something somewhat similar

It's not clear from this description what level of fluency is assumed.

Also seems plausible that additional fluency would actually be harmful, beyond a very basic fluency.

To me the biggest benefit of knowing several languages, especially very different ones, is that it forces one to acknowledge how much of one's thinking depends of small details that would be different in another language, where things are categorized differently.

In French, we use the same word for "city" and "town"; for "politics" and "policy"; so necessarily the way we speak about those will be different. In Chinese, there are different words for brothers and sisters depending on whether they are younger or older, and for cousins depending (among others) of whether they share your surname.

It's a good habit to try to imagine what some arguments would sound like in another language; some depend on etymology and can't even be translated ("how can you call a twelve year old a teenager?"; "Well Democracy comes from the Greek ..."); some work but sound much more awkward (Germany philosophy?).

I acquired a second mother tongue at a young age (English over Mandarin), and I certainly find different courses of thought more comfortable in each language. I also have different cached thoughts stored in each language, which, although usually harmless (I have a small amount of pi and my times tables memorised exclusively in Mandarin, which makes it quite funny to watch me doing arithmetic out loud), has occasionally been the source of quite painful cognitive dissonance (this has got a lot better since LW taught me to decompartmentalise).

I also, unsurprisingly, have different senses of humour and aesthetic judgements in different languages, which makes me able to enjoy speeches, songs, poems etc. in one languages while being barely able to read translations without cringing/laughing/facepalming.

I certainly wouldn't say I'm more rational in either language though, so that's another point to support the "only if you have to concentrate" hypothesis. However, it might help me to run decisions in both languages, if not only because it forces me to re-think the idea from a slightly different angle.

I don't notice this effect, though I haven't systematically tested. Then again, the two languages I speak moderately fluently are both more or less native languages.

[-][anonymous]9y 1

I suspect this effect only works in languages you're not proficient in, not necessarily with all languages you can understand other than your native language.

[-][anonymous]6y 0

Would translating the questions into a foreign language as one goes through the test also help? I mean, suppose a school level introduction to rational thinking was designed, and the pupils told to translate things for themselves into a FL they study? It would be good practice for both tasks, and they might retain the habit of switching to the FL when they have to formalize real-life problems. What do you think?