Languages (most kids learn English, Chinese, French or Spanish, and an elective language which can things like Japanese, Russian, Arabic or Sign Language). Students are required to be fluent in 4 languages by age 16. Most begin at 4 or 5. A large portion of this training looks like “Rosetta Stone”… then students get into more conversational classes (vocabulary drills are all on the tablets). By age 12, many students are simply taking classes in other languages.
That old status shibboleth. 4 natural languages! No, a rationalized system would be more like ‘you learn English if it isn’t your native language’, and the extra time is used for other things (say, programming languages).
Learning 4 other languages is perhaps excessive, but I’d think it would be a good idea for everyone to learn at least one other language well enough to be able to think in it without too much trouble, preferably one that carves concept space up very differently from your native language. Otherwise it’s too easy to make mistakes like confusing incidental groupings in your language with natural categories. Of course it could be a language optimized for easy learning instead of a natural language, let alone multiple romance languages.
Otherwise it’s too easy to make mistakes like confusing incidental groupings in your language with natural categories.
I’m quite sure that explicitly learning the relevant knowledge about the process of category formation in humans and its implications towards accurate thinking would take a lot less time and be more effective than learning another language.
I don’t think that learning languages is an optimal method to gain any ability other than actually understanding and speaking them (and even that, in the world of the future, might be better accomplished by buying a good translation program).
I’m quite sure that explicitly learning the relevant knowledge about the process of category formation in humans and its implications towards accurate thinking would take a lot less time and be more effective than learning another language.
More effective than just learning another language chosen at random and never giving the differences much thought, sure. But I’d guess a combination would be considerably more effective than either. Most people don’t learn well without examples (and intuitively understood real world examples would be a lot better than abstract theoretical examples), and I don’t think most people are good enough at divorcing themselves from their perspective without the extra help of a different point of view.
(and even that, in the world of the future, might be better accomplished by buying a good translation program)
Maybe I misunderstand you, but if you genuinely believe translations that allow as good an understanding of the original as a close to native level understanding of the language are possible (at similar message length) in all or even most cases you don’t understand how much language constrains you.
But I’d guess a combination would be considerably more effective than either.
Well, it seems like a case of clashing intuitions. Learning another language might give some cognitive benefits and it will provide real-world examples of language not being a perfect representation of reality, but I think those benefits will be weak and not worth the effort. And I’m very doubtful of the idea that there are some general insights that can be acquired only through language-learning.
And on the second point: I don’t believe that perfect translations are possible. By mentioning the possibility of future translation software, I wanted to point out the fact that the author of quoted article kind of ignores technological progress (well, except that the future will be full of iPads and they shall be rollable!). And perfect translations are unnecessary for the purpose of practical communication anyway (and hey, if I’m speculating about the awesomeness of future AI, I might as well postulate that the translators will be smart enough to alert you whenever a possible linguistic confusion comes up).
I’m quite sure that explicitly learning the relevant knowledge about the process of category formation in humans and its implications towards accurate thinking would take a lot less time and be more effective than learning another language.
That’s like arguing that learning anatomy will do more to keep you fit then doing sports.
Otherwise it’s too easy to make mistakes like confusing incidental groupings in your language with natural categories.
Is that actually true? My understanding was that Sapir-Whorf had little experimental backing for anything that really mattered, and then there was that recent search showing extreme regularity in how color names were assigned...
Poly-linguals keep claiming that Sapir-Whorf is really true. Here’s Eric Raymond responding to a comment I left earlier today,
And I must tell you that based on my experience as a crib bilingual who has spoken four languages I don’t think strong Sapir-Whorf is a mistake at all. I have actually felt the effect on my thoughts when I code-shift between languages, and if the reports I’ve heard from other polylinguals are to be believed this is an experience we all have. It is also relevant that I can think things in mathematical notation or Python that I cannot think in any natural language, or vice-versa.
Poly-linguals keep claiming that Sapir-Whorf is really true.
Overgeneralization. I do not know how many languages does one need to know to be considered polylingual, but if 6 is enough, then let me say this:
Yes, some concepts are carved differently in different languages, and it feels… strange… when one encounters this for the first time. Sometimes it is kind of scary when you realize that you can express a concept in a foreign language, but there are no words in your native language that would describe it with equal power, and you can only use a sentence or two to describe what in the other language you could have said directly by using one word, or just slightly modifying an inflection.
This effect is often exaggerated ad absurdum, probably because it is useful for some theories or even politically (“if we could make people use different words, we could build a perfect society”) and for polylinguals it is good for signalling (“man, I speak foreign languages, and it totally gave me magical powers”). I predict that this effect is stronger in predominantly monolingual societies.
As an example, it is often said that people would be less sexist if they stop using words “he” and “she”, and find some gender-neutral pronoun. If you believe this, you should go to Hungary, because in Hungarian language there is no word for “he” or “she”, there is only a gender-neutral pronoun… but the Hungarian society is more or less the same as in surrounding countries. So I consider this theory experimentally disproved on sample of millions of people—and yet I predict that most LW-ers will not update zir / eir / vis / shklir opinion on this.
Also, there is a question of how many time should one invest to these polylingual insights. Let’s say that after 2000 hours of study of foreign languages you have found dozen occasions where the wordspace is divided differently in different languages. Should you invest another 2000 hours to find another dozen examples? You will probably hit diminishing returns, because languages are similar, they evolved from each other. What is the cost/benefit analysis of this? If I tell you that in Russian there is no word for “blue”, only words for “dark blue” and “light blue”, how much enlightened do you feel? How many hours of your time would you like to pay for this mystical knowledge?
Learning languages is good if you want to communicate with other people. And as a side-effect it gives you a few insights, but if your goal are only the insights, then you can use your time much better.
EDIT: And to be less of a hypocrite, I admit that learning languages can also be a beautiful hobby. But there is no need to rationalize it by pretending it gives you more that it does.
That said, once a linguistic structure has been culturally associated (however arbitrarily) to a particular ideological position, using that linguistic structure within that culture signals (however inaccurately) one’s association with that position. For example, the signaling effects of using gendered and ungendered pronouns are more or less independent of the social-engineering effects of those pronouns.
And, of course, signaling choices have social-engineering effects of their own. My using, or not using, linguistic structures that are (however arbitrarily) culturally associated with sexism (once those structures are known to exist) has an effect on how acceptable I am perceived as considering sexism to be, which in turn has an effect on how acceptable sexism is perceived to be more generally.
So I consider this theory experimentally disproved on sample of millions of people—and yet I predict that most LW-ers will not update zir / eir / vis / shklir opinion on this
You just convinced me that removing grammatical gender does not make society less sexist. (I still believe that using gender-neutral pronouns or phrases instead of a single gendered pronoun does, and that having a gender-neutral pronoun at all makes society less biased against genderqueers.)
I parsed your comment as: “using genderless pronouns reduces sexism; but using genderless language does not reduce sexism”.
Then I imagined a graph where x-axis is how large part of the language is gendered, and y-axis is how much this makes people sexist. So at the beginning the curve starts going down (because this is what you believe), and at the end it jumps back right to the original level (to catch the new data point). Sounds ad-hoc to me.
Or let me put it in different words: If you think that language structure has an effect on sexism, how exactly would you structure the language to produce minimum sexism possible (ideally zero)? For some reason, removing gramatical genders completely seems not optimal. -- So if you would start with a language that does not have gramatical genders, would you actually add them to the language to make it less sexist?
Note: You have an option to defy my data, because I actually don’t have a proof that Hungarians are not less sexist. It honestly seems to me so, but that is not completely reliable.
Another option is to say that the influence of language on sexism is not strictly f(language), but more like f(language.today, language.yesterday), so it is not precisely the absence of single-gendered pronouns, but rather their recent (or ongoing) removal that reduces sexism. (They I would say that it is backwards: non-sexism removes pronouns, pronouns don’t remove sexism.)
I previously believed that removing some amount of mentions of gender in language (e.g. replacing “he or she” with “they”) would reduce sexism. If I met a society where white people were referred to as “whe” and black people as “ble”, I would think “Wow, that society is pretty racist”. By analogy, I think our society is pretty sexist. There are reasons why removing some gendered language, like pronouns, or all grammatical gender, would make people less sexist; for example, they would not be constantly primed to think of gender and thus act in gender-dependent ways. Hungary disproves this (no need to defy your data, I believe you).
However, I still believe that removing unbalanced mentions of gender (e.g. replacing epicene “he” with “he or she”) reduces sexism. (In the aforementioned society, there’s a proverb that goes “An Englishwhite’s house is whis castle”, and if asked everyone tells you that it also applies to blacks and other races.) They don’t just constantly remind people that gender exists, they also remind them that there is a default gender for people to be and anything else is a special quirk. As efforts to remove these mentions usually come from people who fight sexism in general, it’s very hard to test whether there is any causation flowing this way rather than the other one.
I also believe that having a pronoun applicable to genderqueers helps, because it contributes to making it a recognizable category (wearing black, not wearing a clown suit). This is probably swamped by other ways to make the existence of more than two genders noticeable.
While I was offline, I realized that there are two aspects of this situation. First, whether gendered words exist in given language or not. Second, whether they are used correctly or incorrectly (such as using male pronoun “he” in situations where female person is also possible). These are partially independent (even in a language without gramatical genders it is possible to use constructs such as “director-man” and “director-woman”, and then incorrectly use the word “director-man” in situations not limited to male persons).
Now in your comment I see the third aspect; whether there is a gender-neutral pronoun for people who don’t identify as “he” or “she”. Construct “he or she” is bad, because even when it tries to describe a superset, does it by enumerating subsets. I don’t see any problem with singular “they”, but I am not a native English speaker.
I am afraid that inventing a new pronoun specifically for genderqueers would lead to infinite discussions… such as whether the same pronoun should be used for all kinds of genderqueers, or whether there should be different pronouns for different groups, and everyone would accuse everyone of insensitivity because they use last-year pronouns instead of the most current update. (Evidence: there is still not consensus on the gender-neutral pronoun, and that should be much easier task.)
Yup, those are the three aspects I’m talking about.
I’ve never seen anyone demand a pronoun that didn’t cover binary people, only gender-neutral ones. Having to make do with “he”, “she” and combinations thereof is insufficient.
Even if he genuinely is thinking in the other languages, that doesn’t go very far in showing usefulness. I bet if I spent years memorizing Scholastic metaphysics, I’d fall into occasional Scholastic patterns of thought or vocabulary even if it was completely useless. (And let’s not talk about the size of the marginal returns, if we can’t even decide on the sign.)
I do ‘feel’ a bit different depending on which language I’m thinking in. (On the other hand, it would be impossible for me to perform a double-blind test to find out how much of that is due to placebo.) But sufficiently strong forms of the SW hypothesis sound like bullshit to me. (Most of the times someone points out some exotic feature of some language purported to influence its speakers’ thought, English has that feature too. And I mentally classify anything starting with “X language has N words for Y” as rubbish, trash, junk, garbage, waste, refuse and litter.)
I’m bilingual in Finnish and English, and don’t recall ever consciously getting nontrivial insight from the conceptspace variations. Any idea what I could be looking for to notice if this is actually happening?
Are there examples of bilingual people making less of some conceptual mistake unilingual people make for some pair of languages?
That would be unfair for non-native English speakers, who’ll have to learn one more language than native English speakers. And as for the advantages of bilingualism (in addition to what FAWS says), see http://languagelog.ldc.upenn.edu/nll/?p=3175 -- though IIRC those effects require that you actually speak both languages on a daily base for years. But I agree that four languages is usually overkill. (On the other hand, the efforts needed to learn an n-th language when you are already fluent in (n − 1) languages decreases with increasing n, so if for some reason you’ll have to learn Finnish at some later point in life, e.g. because you want to move to Finland, having learnt lots of languages before will definitely help.)
Anyway, using living languages among the most influential ones is a vast improvement over the Latin and Ancient Greek that some Italian high schools teach. (And the idea of teaching a sign language in schools is one that had never occurred to me, but which is not obviously wrong.)
That would be unfair for non-native English speakers, who’ll have to learn one more language than native English speakers.
Multiple languages has always meant someone is going to be screwed over in some sense. (Pity the Papua New Guinean whose language is not spoken even the next valley over, much less worldwide...) The main thing is to minimize the damage done.
That’s interesting, but your caveat aside it’s not all that great: one has a less than 50% chance of Alzheimer’s so the expected time-value of 5 years of protection is 2.5 years, at the least valuable time of your life, in exchange for let’s say a year of your youth? Not a slamdunk case for bilingualism (much less tri or quadlingualism, which is just plain signaling).
That old status shibboleth. 4 natural languages! No, a rationalized system would be more like ‘you learn English if it isn’t your native language’, and the extra time is used for other things (say, programming languages).
Learning 4 other languages is perhaps excessive, but I’d think it would be a good idea for everyone to learn at least one other language well enough to be able to think in it without too much trouble, preferably one that carves concept space up very differently from your native language. Otherwise it’s too easy to make mistakes like confusing incidental groupings in your language with natural categories. Of course it could be a language optimized for easy learning instead of a natural language, let alone multiple romance languages.
I’m quite sure that explicitly learning the relevant knowledge about the process of category formation in humans and its implications towards accurate thinking would take a lot less time and be more effective than learning another language. I don’t think that learning languages is an optimal method to gain any ability other than actually understanding and speaking them (and even that, in the world of the future, might be better accomplished by buying a good translation program).
More effective than just learning another language chosen at random and never giving the differences much thought, sure. But I’d guess a combination would be considerably more effective than either. Most people don’t learn well without examples (and intuitively understood real world examples would be a lot better than abstract theoretical examples), and I don’t think most people are good enough at divorcing themselves from their perspective without the extra help of a different point of view.
Maybe I misunderstand you, but if you genuinely believe translations that allow as good an understanding of the original as a close to native level understanding of the language are possible (at similar message length) in all or even most cases you don’t understand how much language constrains you.
Well, it seems like a case of clashing intuitions. Learning another language might give some cognitive benefits and it will provide real-world examples of language not being a perfect representation of reality, but I think those benefits will be weak and not worth the effort. And I’m very doubtful of the idea that there are some general insights that can be acquired only through language-learning.
And on the second point: I don’t believe that perfect translations are possible. By mentioning the possibility of future translation software, I wanted to point out the fact that the author of quoted article kind of ignores technological progress (well, except that the future will be full of iPads and they shall be rollable!). And perfect translations are unnecessary for the purpose of practical communication anyway (and hey, if I’m speculating about the awesomeness of future AI, I might as well postulate that the translators will be smart enough to alert you whenever a possible linguistic confusion comes up).
That’s like arguing that learning anatomy will do more to keep you fit then doing sports.
More like saying that if your goal is muscle mass it’s better to hit the gym than to take up figure skating.
Is that actually true? My understanding was that Sapir-Whorf had little experimental backing for anything that really mattered, and then there was that recent search showing extreme regularity in how color names were assigned...
Poly-linguals keep claiming that Sapir-Whorf is really true. Here’s Eric Raymond responding to a comment I left earlier today,
http://esr.ibiblio.org/?p=4139&cpage=1#comment-372073
And he responds to more criticism here.
Overgeneralization. I do not know how many languages does one need to know to be considered polylingual, but if 6 is enough, then let me say this:
Yes, some concepts are carved differently in different languages, and it feels… strange… when one encounters this for the first time. Sometimes it is kind of scary when you realize that you can express a concept in a foreign language, but there are no words in your native language that would describe it with equal power, and you can only use a sentence or two to describe what in the other language you could have said directly by using one word, or just slightly modifying an inflection.
This effect is often exaggerated ad absurdum, probably because it is useful for some theories or even politically (“if we could make people use different words, we could build a perfect society”) and for polylinguals it is good for signalling (“man, I speak foreign languages, and it totally gave me magical powers”). I predict that this effect is stronger in predominantly monolingual societies.
As an example, it is often said that people would be less sexist if they stop using words “he” and “she”, and find some gender-neutral pronoun. If you believe this, you should go to Hungary, because in Hungarian language there is no word for “he” or “she”, there is only a gender-neutral pronoun… but the Hungarian society is more or less the same as in surrounding countries. So I consider this theory experimentally disproved on sample of millions of people—and yet I predict that most LW-ers will not update zir / eir / vis / shklir opinion on this.
Also, there is a question of how many time should one invest to these polylingual insights. Let’s say that after 2000 hours of study of foreign languages you have found dozen occasions where the wordspace is divided differently in different languages. Should you invest another 2000 hours to find another dozen examples? You will probably hit diminishing returns, because languages are similar, they evolved from each other. What is the cost/benefit analysis of this? If I tell you that in Russian there is no word for “blue”, only words for “dark blue” and “light blue”, how much enlightened do you feel? How many hours of your time would you like to pay for this mystical knowledge?
Learning languages is good if you want to communicate with other people. And as a side-effect it gives you a few insights, but if your goal are only the insights, then you can use your time much better.
EDIT: And to be less of a hypocrite, I admit that learning languages can also be a beautiful hobby. But there is no need to rationalize it by pretending it gives you more that it does.
Absolutely agreed, as far as it goes.
That said, once a linguistic structure has been culturally associated (however arbitrarily) to a particular ideological position, using that linguistic structure within that culture signals (however inaccurately) one’s association with that position. For example, the signaling effects of using gendered and ungendered pronouns are more or less independent of the social-engineering effects of those pronouns.
And, of course, signaling choices have social-engineering effects of their own. My using, or not using, linguistic structures that are (however arbitrarily) culturally associated with sexism (once those structures are known to exist) has an effect on how acceptable I am perceived as considering sexism to be, which in turn has an effect on how acceptable sexism is perceived to be more generally.
You just convinced me that removing grammatical gender does not make society less sexist. (I still believe that using gender-neutral pronouns or phrases instead of a single gendered pronoun does, and that having a gender-neutral pronoun at all makes society less biased against genderqueers.)
I parsed your comment as: “using genderless pronouns reduces sexism; but using genderless language does not reduce sexism”.
Then I imagined a graph where x-axis is how large part of the language is gendered, and y-axis is how much this makes people sexist. So at the beginning the curve starts going down (because this is what you believe), and at the end it jumps back right to the original level (to catch the new data point). Sounds ad-hoc to me.
Or let me put it in different words: If you think that language structure has an effect on sexism, how exactly would you structure the language to produce minimum sexism possible (ideally zero)? For some reason, removing gramatical genders completely seems not optimal. -- So if you would start with a language that does not have gramatical genders, would you actually add them to the language to make it less sexist?
Note: You have an option to defy my data, because I actually don’t have a proof that Hungarians are not less sexist. It honestly seems to me so, but that is not completely reliable.
Another option is to say that the influence of language on sexism is not strictly f(language), but more like f(language.today, language.yesterday), so it is not precisely the absence of single-gendered pronouns, but rather their recent (or ongoing) removal that reduces sexism. (They I would say that it is backwards: non-sexism removes pronouns, pronouns don’t remove sexism.)
That’s not what I said. Let me rephrase:
I previously believed that removing some amount of mentions of gender in language (e.g. replacing “he or she” with “they”) would reduce sexism. If I met a society where white people were referred to as “whe” and black people as “ble”, I would think “Wow, that society is pretty racist”. By analogy, I think our society is pretty sexist. There are reasons why removing some gendered language, like pronouns, or all grammatical gender, would make people less sexist; for example, they would not be constantly primed to think of gender and thus act in gender-dependent ways. Hungary disproves this (no need to defy your data, I believe you).
However, I still believe that removing unbalanced mentions of gender (e.g. replacing epicene “he” with “he or she”) reduces sexism. (In the aforementioned society, there’s a proverb that goes “An Englishwhite’s house is whis castle”, and if asked everyone tells you that it also applies to blacks and other races.) They don’t just constantly remind people that gender exists, they also remind them that there is a default gender for people to be and anything else is a special quirk. As efforts to remove these mentions usually come from people who fight sexism in general, it’s very hard to test whether there is any causation flowing this way rather than the other one.
I also believe that having a pronoun applicable to genderqueers helps, because it contributes to making it a recognizable category (wearing black, not wearing a clown suit). This is probably swamped by other ways to make the existence of more than two genders noticeable.
While I was offline, I realized that there are two aspects of this situation. First, whether gendered words exist in given language or not. Second, whether they are used correctly or incorrectly (such as using male pronoun “he” in situations where female person is also possible). These are partially independent (even in a language without gramatical genders it is possible to use constructs such as “director-man” and “director-woman”, and then incorrectly use the word “director-man” in situations not limited to male persons).
Now in your comment I see the third aspect; whether there is a gender-neutral pronoun for people who don’t identify as “he” or “she”. Construct “he or she” is bad, because even when it tries to describe a superset, does it by enumerating subsets. I don’t see any problem with singular “they”, but I am not a native English speaker.
I am afraid that inventing a new pronoun specifically for genderqueers would lead to infinite discussions… such as whether the same pronoun should be used for all kinds of genderqueers, or whether there should be different pronouns for different groups, and everyone would accuse everyone of insensitivity because they use last-year pronouns instead of the most current update. (Evidence: there is still not consensus on the gender-neutral pronoun, and that should be much easier task.)
Yup, those are the three aspects I’m talking about.
I’ve never seen anyone demand a pronoun that didn’t cover binary people, only gender-neutral ones. Having to make do with “he”, “she” and combinations thereof is insufficient.
Even if he genuinely is thinking in the other languages, that doesn’t go very far in showing usefulness. I bet if I spent years memorizing Scholastic metaphysics, I’d fall into occasional Scholastic patterns of thought or vocabulary even if it was completely useless. (And let’s not talk about the size of the marginal returns, if we can’t even decide on the sign.)
I agree with you, he was, after all, responding to my calling “General Semantics” an extension of the “strong Sapir-Whorf mistake”.
I do ‘feel’ a bit different depending on which language I’m thinking in. (On the other hand, it would be impossible for me to perform a double-blind test to find out how much of that is due to placebo.) But sufficiently strong forms of the SW hypothesis sound like bullshit to me. (Most of the times someone points out some exotic feature of some language purported to influence its speakers’ thought, English has that feature too. And I mentally classify anything starting with “X language has N words for Y” as rubbish, trash, junk, garbage, waste, refuse and litter.)
To jump on the anecdotal evidence bandwagon, is there any evidence that learning a second language increases one’s skills with their natural language?
I’m bilingual in Finnish and English, and don’t recall ever consciously getting nontrivial insight from the conceptspace variations. Any idea what I could be looking for to notice if this is actually happening?
Are there examples of bilingual people making less of some conceptual mistake unilingual people make for some pair of languages?
I can’t think of any good examples right now, but that does happen to me (though I might be setting the bar for “nontrivial” lower than you).
That would be unfair for non-native English speakers, who’ll have to learn one more language than native English speakers. And as for the advantages of bilingualism (in addition to what FAWS says), see http://languagelog.ldc.upenn.edu/nll/?p=3175 -- though IIRC those effects require that you actually speak both languages on a daily base for years. But I agree that four languages is usually overkill. (On the other hand, the efforts needed to learn an n-th language when you are already fluent in (n − 1) languages decreases with increasing n, so if for some reason you’ll have to learn Finnish at some later point in life, e.g. because you want to move to Finland, having learnt lots of languages before will definitely help.)
Anyway, using living languages among the most influential ones is a vast improvement over the Latin and Ancient Greek that some Italian high schools teach. (And the idea of teaching a sign language in schools is one that had never occurred to me, but which is not obviously wrong.)
Multiple languages has always meant someone is going to be screwed over in some sense. (Pity the Papua New Guinean whose language is not spoken even the next valley over, much less worldwide...) The main thing is to minimize the damage done.
That’s interesting, but your caveat aside it’s not all that great: one has a less than 50% chance of Alzheimer’s so the expected time-value of 5 years of protection is 2.5 years, at the least valuable time of your life, in exchange for let’s say a year of your youth? Not a slamdunk case for bilingualism (much less tri or quadlingualism, which is just plain signaling).