Sapir Whorf Hypothesis

Edward Sapir and his student Benjamin Lee Whorf developed a theory of linguistics which claims that language shapes thought. This idea lies behind the LinguisticDeterminism of LeftWing and PostModern philosophers.

Whorf wrote (original emphasis):

: We cut nature up, organize it into concepts, and ascribe significances as we do, largely because we are parties to an agreement to organize it in this way - an agreement that holds throughout our speech community and is codified in the patterns of our language. The agreement is, of course, an implicit and unstated one, but its terms are absolutely obligatory; we cannot talk at all except by subscribing to the organization and classification of data which the agreement decrees.
: (Whorf, B. L. (1940): Science and Linguistics, Technology Review 42(6): 229-31, 247-8.)

Or see at much greater length http://sloan.stanford.edu/mousesite/Secondary/ThoughtReality.htm.

Very relevant 2003 paper on this topic: http://www.mpi.nl/world/pub/lang&mind_final1.pdf (very balanced, lots of empirical evidence; strengthens SWH, but also points out its limits).

"A language that doesn't affect the way you think about programming, is not worth knowing." - Alan J. Perlis, Epigrams in Programming

So, seeing the determinism in this idea, and the happy consonance with CulturalMaterialism?, in order to control the thoughts of the population LeftWing political and cultural institutions seek to enforce a particular vocabulary on the population. Hence PoliticalCorrectness.

Of course, there is also Politically Right language. For example, it's not nice to call people Racist. And what about CollateralDamage?? We have FreedomFighters?. They have Terrorists. And don't mention asylum seekers, speak of Suspected Illegal Immigrants.

With leftist moral equivalence, you might as well go back to the stone age and discard the concept of language. They turns saints into Hitlers, fighting against Nazism into terrorism, etc. How about I turn the argument around to its extreme and say 'why shouldn't people be racist against blacks, one black in 1973 did something wrong for an hour in a small rural town therefore the entire world is justified in slaughtering all blacks' - do you agree with that logic? If not, why?

Ultimately, these approaches lead to CulturalRelativism?. You don't want to get any of that on you.

Interesting. While I might agree with Whorf's actual words, above, I don't see any logical connection between those words and the implications usually drawn from them; that is, "Language determines perception" (as claimed on LinguisticDeterminism). Even "which claims that language shapes thought" is actually a very weak claim, because "shapes" can be taken as weak or strong. --AlistairCockburn

You probably have to be a social "scientist" to follow this kind of argument. Take it up with your friendly neighbourhood professor of cultural studies...

Hmm. That last seems to assume the SapirWhorfHypothesis ...

Be aware that using one language to directly translate foreign cultural concepts is fraught with problems. For instance, although Ilongot "liget" is roughly analogous to "anger" [see <http://www.humanas.unal.edu.co/psicologia/docentes/sierra/emociones/Russell.htm>], it is not the same - and should you ask "What is the Ilongot term for 'anger'", you have already incorrectly framed your result.

There are universal concepts, but they exist far below the level of "Culture". [Pure linguistic relativists disagree, but they're both rare, and _wrong_ ;-]. Be aware that even such apparently basic and universal concepts as basic shape terms and basic colour terms are culturally relative.

Some languages conflate shape and size - Ewe has multiple words for "round", depending on whether the object is as big as a golf ball, or a basket ball, for instance. Still others break "long" up into such categories as "long and thin", "long and made of wood", "long and flat", etc.

RussianLanguage has separate "light blue" and "dark blue" basic colour terms, and mauve is apparently becoming its thirteenth (Note that EnglishLanguage has eleven basic colour terms). Other languages have as few as two, but which colours these two cover varies. How many colours in a rainbow? Not necessarily seven...

Same is true in Spanish: "light blue" = celeste and "dark blue" = azul. These (or their likeness) are also English words; "celestial" = sky blue and "azure" = purplish blue.

I was taught that while not all languages have the same amount of colour terms, there is a standard order in which they appear. First black and white, then red, then (IIRC) blue and green, then orange, yellow and purples. So if a language has only three colour terms, say Ack, Oink and Sniss, 'Ack' may mean white but would also be used for light colours like yellow, 'Oink' might cover not just black but other dark colours as well and 'Sniss' would be red and everything not covered by Ack and Oink, but if a NativeSpeaker? was asked to show a sniss object, he would show a red object, so the default, or prototype'' if you will, would be red.

Yes and no... The first two steps are pretty much fixed. It's still a bit messy, though because which colours are in the warm set/cool set/macro-red set can vary. Then it all gets really messy =) Yellow or Green or Grue (Green/Blue composite) splits off... They may divide our "yellow" differently - one segment of our yellow may clump with their grue, and another segment of their yellow may clump with their macro-red... To put that another way, there are languages that have only two words for colors. Given a choice of many colored tiles, speakers of these languages reliably indicate that these words correspond to what english speakers call black and white. There are languages with three words for color. In all such languages known, these words correspond precisely to black, white and red. There are other languages with 4,5,6 color terms. Again, they have black, white and red, plus some terms from a set of 3 colors (or so). Exactly which of the 3 they have varies, but all 6 of those colors will be present in any language that has 6 or more terms for colors. And so on. (Roughly speaking. Don't quote me.)

There might be a language which has no color terms, namely the Piraha language mentioned below. It may be that colors are described on the spot (i.e. a red thing might be described as blood-like).

What evidence is there for or against the SapirWhorfHypothesis? -- SethGordon

As I recall TheLanguageInstinct presented some solid evidence against it. -- PhilGoodwin

As I recall TheLanguageInstinct presented solid evidence for an alternative theory (that language is built in to our brains, which could be interpreted as thought determines language). A key point to this theory is NoamChomsky's idea of deep structure - that the brain works on a fixed grammar that is translated to our natural language.

I am not sure that this directly contradicts the SapirWhorfHypothesis, because it seems to be related to grammar. The SapirWhorfHypothesis could apply to vocabulary; it seems that the main target of PoliticalCorrectness is vocabulary and not grammar. Also, even when it comes to grammar, the two theories are only contradictory if taken to the extreme (that is, thought is the only thing determining grammar/language is the only thing determining language.

Altogether I would not say that TheLanguageInstinct presents solid evidence against it. -- ChrisBrooking?

While TheLanguageInstinct as a whole is a general overview of modern linguistics, there's one particular section that presents the arguments against the SapirWhorfHypothesis. -- MossCollum

Chapter 3 of TheLanguageInstinct is largely a refutation of LinguisticDeterminism and the SapirWhorfHypothesis. Here is a summary of the main points:

We are often unable to put our thoughts into words, or we say things that don't accurately reflect what we are thinking.
When we try to remember something that we heard or read, we rarely remember the exact words. Instead, we remember the "gist", and attempt to re-construct the original from it.
Whorf's assertions that people who speak other languages think differently are based upon word-for-word translations of utterances in those other languages into English. For example, an Apache sentence meaning "He invites people to a feast" is translated into English as "He, or somebody, goes for eaters of cooked food." To Whorf, examples like this proved that the Apache mind had to be fundamentally different from his own.
Whorf's claim that the Hopi have no concept of time does not stand up to scrutiny. Studies by others have shown that Hopi speech has tense and that they have units of time, quantifiers of time, a calendar, ceremonial day sequences, timekeeping devices, and time-based record keeping.
Eskimo speakers do not have an incredible number of words for snow. They have about the same number as English speakers (who have snow, sleet, slush, blizzard, avalanche, flurry, etc.). They may even have less words, english has borrowed from so many places. But there is not a 1:1 relationship between the words they have and the words we have.
- Sanskrit has a very diverse vocabulary for snow as well IIRC. -- TheerasakPhotha
People who are raised without a language, and people who lose the power of language (stroke victims or aphasiacs, for example), exhibit the capability of thought. When language is introduced or restored to such people, they are able to recount what they were thinking during the period when they did not have language. Similarly, other "languageless" creatures like babies and monkeys demonstrate the ability to reason about space, time, objects, number, rate, causality, and categories.
"Visual thinking" is influenced by language very little, if at all.
- An observation that lasts until the moment you meet visual programming languages and start learning graph-based approaches to problem-solving.
The TuringMachine shows that symbolic, logical reasoning can be carried out mechanically, without any meaning ascribed to the symbols or the operations. Thus, language is not a prerequisite for this kind of thought. This is not thought, and it does require a formal language.
People are capable of learning one another's languages. The primary difficulties in doing so are related to learning the words and grammar, not in attempting to understand a different way of thinking.
- Perhaps you have not learned any languages with methods of thinking significantly different from English, or perhaps you have not learned them well enough to speak in those languages without constructing sentences like a foreigner; I assure you that there are such languages, and that if you spoke them well you would percieve a difference in thought process.
  - {Or perhaps the author is implying that learning a new way of thinking comes for free, and is, therefore, not a primary difficulty. Not that I'm agreeing. But I do agree with: "A language that doesn't affect the way you think about programming, is not worth knowing." - Alan J. Perlis, Epigrams in Programming, #19}

One is left with the impression that Pinker considers Whorf to be a quack who had no business pretending to be a scientist (there is a whiff of AdHominem attack). Pinker does not deny that language can have some influence on thought (he calls this the "weak" form of the hypothesis), but the influence is not as dramatic as LinguisticDeterminism claims.

Well, yeah, I agree with the 'weak' version of the hypothesis wholeheartedly. We can look at the 'language' of mathematics for instance, that is, the use of physical objects and writing as a means of representing mathematical concepts. Even before the advent of modern algebraic notation, mathematicians have always used some kind of representation of symbols and numbers to organize their thoughts, although they were generally not as efficient as our modern, purely symbolic written conventions. It is interesting to note that the ancient Chinese performed Gaussian elimination with pieces of bamboo to represent quantities of grain. -- TheerasakPhotha

One can have thought without language, but our ability to distinguish between unnamed concepts is relatively weak. In an unannotated domain, that is, where we have not given names to things, we can still think, but the things we are thinking about do not have defined boundaries. Only once we give things names do they have distinct identity. Until then, there isn't even a fixed number of concepts, because what is one thing from one point of view will morph into two or twenty from a different point of view and you won't even know it is happening because you don't think of these things as having distinct identity.

Possibly I'm subjecting myself to mysticism or maybe just seeing what I expect to see, but I perceive that speaking a certain language makes thinking a certain way easier. For example, the GermanLanguage almost always literally describes function with names, and I think that native German Speakers appear to have a more scientific bent than other cultures. I realize that this isn't evidence, but I'd like to hear other people voice (type) their perceptions of this. -- ShaeErisson

The problem is that German culture is at least as likely to influence German people as much as the details of the German language - Germans prize technical excellence and scientific details, so those qualities get emphasized in German personalities. For a contrast, look to the Japanese, whose language constructs differ greatly from German, but who also have a strong technical/scientific bent to their current culture -- PeteHardie

I think that language can influence thought, but we all know thought came before language. All language really is is a way of reifying our thoughts both for our own use (e.g., chronicles), and for transmission to others (e.g., news radio). In those cases where a language was an inadequate means of expression, it was upgraded accordingly. Consider the massive importation of French and Latin words into EnglishLanguage, and Sanskrit and Pali words into ThaiLanguage. Some of this is a matter of prestige; Anglo-Saxon obviously has a word for year (gieru?), but many texts written in Anglo-Saxon begin with Latin, like so: Anno 1066. But that was not the crux of it; the crux of it was the need to express technical and ecclesiastical thoughts. And if Latin, French, Sanskrit, and Pali were not there to furnish these words, then they would have been invented independently. Chinese texts tend to obstinately cling to their own native vocabulary, even in the Buddhist canonical texts, where words like 'sutra' are replaced with the native 'jing'. Are the Romans and Chinese unique in being able to formulate their own vocabulary? Certainly not.

I definitely do not think anything like NewSpeak is sustainable in the long term, at least not in a pure form. It would fall apart in the unguarded prole quarters and then influence Party members afterwards. :) -- TheerasakPhotha

Hmmm, SapirWhorfHypothesis might have the causality wrong, maybe the culture determines the form of the language. Language is fairly plastic over relatively short periods of time. Consider how relatively easy it is to place the approximate decade of a random sample of twentieth century speech, you'll notice that the more popular culture tends to change relatively quickly whereas high culture changes only slowly: (totally made up example) teen slang varies from week to week while classical radio announcers sound the same for decades. -- LarryPrice

The idea is not that language doesn't change but that your conceptual leaps are constrained by the language. What you can think in ten years is not what you can think now. Consider the following exchange between a member of an alien culture and a human space traveler: "What is this thing that you earth people call...love?" (sorry I just couldn't resist) -- BrianEwins

But are your thoughts constrained by language, or is your language constrained by your thoughts?

I believe that we can rewrite the quotation at the top, substituting OO for "agreement" (and cutting out a lot of probably irrelevant chaff):

: We view the world in an object oriented (OO) manner as we do, because we agree to. Our language reflects this OO view. The OO View is absolutely obligatory; we cannot talk at all except in OO terms.

Right, onto my analysis. First of all I believe that Worf is contradicting himself here because in the first sentence he says that we 'Agree' to abide by the rules of an OO view, and in the third sentence he is saying that the view is obligatory.

This aside however this statement doesn't say that thought is controlled by speech, it says, clearly, that our speech is controlled by our thoughts, and that these thoughts are constrained by society. It also says that once this agreement is made it is unbreakable.

I would disagree with the last part of the statement. Some forms of meditation practice breaking down the objective/subjective barriers and this is definitely not an OO thought form. Zen Buddhism is also an attempt at breaking the OO view of the world (Zen Koan Paraphrase: Which is moving, the flag or the wind? Neither!).

Non-OO forms of art, such as some variations on Jazz and abstract paintings are also methods of escaping this system. Schizophrenics may also have transcended this view, possibly through damage to the portion of the brain responsible for categorization.

Anyway, this is probably enough for you all to rip me to shreds on, so I'll shut up now ;) -- Bryan Dollery

But nobody really talks about non-OO things except in an OO way. When was the last time you heard somebody say what Tao was, except by saying it isn't any of such-and-such objects? When was the last time you heard somebody describe an abstract painting without just categorizing shapes, colors, styles, evoked emotions, and other such objects?

I would agree that language constrains thought: Things that are hard to verbalize are hard to think about in any depth for any long period of time.

But, we can change our language, and we do so continuously. So I would prefer to say it as... "Language enables thought." A good choice of language can enable and ease discussion of topics of importance to you. That's why we have technical terminology in our business: Developing this terminology makes it easier for us to develop and work with concepts that help us succeed. (No, I'm not talking about memes, or Darwinian selection of ideas! ;-)

For example: I found the terms/concepts of "coupling" and "cohesion" very helpful in developing my understanding of how to make good modular programs. I, and others, seem to also have found value to this funny term/idea called "normalization" of data. -- JeffGrigg

Here's an example that I'll be using tomorrow during my philosophy exam, I'm sure. Without language, you could not have the sensible thought, "Fred is going to Uruguay," without knowing Fred and without having been to (or actually in) Uruguay. So, if I continued the story, "While he is there, he will be taking many photographs of the street culture," you wouldn't be able to comprehend that string of symbols (or sounds) as meaningful--as I'm sure you just did--if you didn't comprehend "Fred is going to Uruguay," as it stood. So, some thoughts are definitely built out of language. -- SunirShah

I think that you've got it backwards: the words "Fred" and "Uruguay" were invented to describe things that were already understood. I assure you that my five month old doesn't have language but is perfectly capable of conceptualizing: "Mommy is coming here. While she is here I will get something good to eat." Which I think is semantically equivalent to your example. So it is perfectly reasonable that we are able to think without words. I certainly think that words help us to think. But I think it's going too far to say that they enable us to think -- PhilGoodwin

Are you sure babies conceptualize anything at all? As I understand it, babies don't really have any barriers between themselves and experience- they don't think "I'm hungry;" they just are hungry.

If that is true, how can you understand Uruguay separate from experiencing it?

My cat, who has very limited oral and written linguistic abilities, seems to find ways of explaining exactly what she wants, especially at 4:30 am when I was out all day and most of the night before and she didn't get her hour of play in. Such communications include loud purring and stepping on my face to awaken me, jumping off the bed and looking back at me once I'm awake, and as I follow her around my apartment, she continuously checks behind her to be sure I'm following, and then lying down on the toy she would like to play with. She is very persistent, and never fails to make her intent known. -- EvanCofsky

To argue that some thoughts are constrained by language and that others are aided by language is to argue nothing. This is a fact and is self-evident. The statement at the top of this page is that the SapirWhorfHypothesis states that language shapes thought. This statement is untrue; the hypothesis states no such thing, it states that

: our view of the world is so entrenched that our language is constrained by it.

That is, we can only talk about the things that we can think about.

If it states the opposite, that we can only think about things using language then this would be trivial to dismiss because of all the concepts that we can not put into words. Have you ever tried to explain to your partner how much you love them? Or tried to describe an abstract painting by Picasso? -- Bryan Dollery

Your proposed restatement of their words is quite different from the quote actually cited at the top, so I'd like to know where you get your version of their hypothesis.

The words about this hypothesis that I have seen so far are either obviously incorrect or so hopelessly weak as to be vacuous. I'd be interested in seeing something substantive.

Looking at the quote at the top: "We cut nature up ... largely because we are parties to an agreement ... that ... is codified in the patterns of our language." The word largely here makes the sentence vacuous. It ends up only saying that words are agreements, and we sometimes happen to chop the world into the same pieces. But guess what? We sometimes also don't.

Continuing: "...we cannot talk at all except by subscribing to the organization and classification of data which the agreement decrees." Both obvious and false. Obvious because of course we can't tell someone else what we mean without using words that have essentially the same meaning to them. And false because we actually do generate new meanings for words as we talk, shifting the meanings through examples and body language. I offer as an example ThomasKuhn's use of the word "paradigm".

In general, people construct the world in ways that have no analog in words, and do that all the time. They are then faced with communicating that construction. Some go to drawings, some to music, some to paintings, some to acting, some simply invent new phrases. So these people do "talk", and do so without "subscribing to the organization and classification of data which the agreement decrees." -- AlistairCockburn

What if any significance is there in applying the SapirWhorfHypothesis (the simplest description of which I have seen is that "language limits thought") to Programming Languages and concepts? Does thinking in machine language, Assembler, COBOL, Fortran, Smalltalk, C, C++, VB, Java etc limit your thoughts? Can you think different thoughts working in one programming language versus another? Can you think these programming thoughts in English or another natural language? I know that the language I am working in does at least influence the design and programming thoughts I think.

How significant are DesignPatterns in thinking of computer design issues? If there were not names like Adapter or Proxy would these concepts be harder to think about? If you think about such things are your thoughts visual or do you think about them in words? Personally, I believe I mostly think about such things visually, but find the names most useful for communication.

Are metaphor and simile, which are critical to the extension of natural language, also important in extending design and programming thoughts? This is *like* that, or this *is* one of those? This seems critical to me in designing software.

Are some computer languages better at allowing the programmer to express useful thoughts, and if so, does this characteristic necessarily lead to better code, more productivity or some other benefits? -- GlennWilson

I think so; one example is figuring out that continuations/closures exist, but not knowing how to express them because you've never learned Smalltalk or Lisp. -- ShaeErisson

In the light of this conversation, it has been interesting to compare this page and the book SketchesOfThought (which argues for an ability to manipulate as-yet-undemarcated ideas) with my ongoing research into what a project consists of. It is clear that once a concept gets named, it is easier to parse my experience to call upon that concept (but this has been known since I took perceptual psychology in 1973). I have been watching (internally) how seemingly new ideas form themselves, pulling some particular shape of experience out of a morass of criss-crossing experiences. That idea-formation seems to support SketchesOfThought on the one hand, and also the notion that once we name a concept or perception, we can use it with fluency to parse and speak experience. (Same is true with programming languages, by the way). -- AlistairCockburn

My description of reality is this: We each live in glass spheres separate from each other, and because we want to understand and be able to compare we paint graph paper like lines on the inside of our spheres... the problem is that each line we paint covers up some part of the raw truth.

Language is a form of life (Lebensform, said by LudwigWittgenstein in his PhilosophicalInvestigations?). There does not pre-exist any kind of language before our life. But as languages shape itself from our life, it can definitely affect our life combined with almost infinite factors of our life. and it affects back again to language; ad infinitum as like two mirrors facing each other. So we can say there's some kind of correlation between the two, but cannot say direct causal relationship.

One interesting point that shows the correlationship is between what we call the word order and our dispositions concerning mental focus. English is HeadInitial? but some Asian languages (Korean, Japanese,...) are HeadFinal?. See the difference in the definition of "language" in each dictionary :

: Um... In both English and Japanese, at least, modifiers precede nouns (red book, akai hon), and the same is probably true for Korean. Japanese and Korean do differ from English in that their basic word order is Subject-Object-Verb, whereas English is Subject-Verb-Object. But this means nothing, since the basic word order of many European languages is Subject-Object-Verb (the most common order among the world's languages) and many Asian languages (including all the varieties of Chinese that I know of) are Subject-Verb-Object (the second most common order among the world's languages). -- SteveConley

English dictionary

Language: the system of sounds and words used by humans to express their thoughts and feelings (Oxford Advanced Learner's Dictionary 5E, Oxford, 1996)

Korean dictionary

Language: human thoughts or feelings/to express/ used/sounds or words/ system (Han Korean Dictionary, Sungahndang, 1998)

Do you see the almost reversed word order between the two? In Korean, we start from the boundary informations whereas in English they start from the central informations.

: But dictionary entries are not complete sentences and fail to capture the distinction you are trying to describe because they lack head verbs. And the location of "information" in a sentence doesn't necessarily correspond to the basic word order of the language. For example, fronting objects and adjuncts for emphasis and/or topicalization: "That book, I have read." or "Quietly, we tiptoed down the hall."

There was an interesting report from a psychologist, RichardNisbett?, shown in the New York Times. He conducted several experiments to compare the thinking habits between Westeners - American students - and Easterners - Koreans and Japanese. In one experiment, the subjects were shown a picture with a largest fish in the center and some small fish around it in a lake. Most American students first started to tell about the largest focal object, the big fish, when asked about the picture afterwards. On the other hand, most japanese students first explained at the start more about the background environment and relations between objects rather than the objects per se.

Surprisingly, when the Japanese students were shown a picture with the same largest fish but with different background, seldom could they recognize the fish as same, while most American students could say it was the same.

Nisbett concludingly said, Easterners tend to focus on the context and relations and think holistic but Westerners focus on the objects and try to analyze. -- JuneKim

: Whatever Nesbitt's conclusions, I doubt it has anything to do with language. The dominant culture in East Asia has always been the Chinese, yet the various Chinese languages do not follow the Japanese/Korean pattern at all. On the other hand, many European languages are Subject-Object-Verb like Japanese and Korean, and some of them (Finnish and Hungarian, for example) are even agglutinative (I'm speaking geographically, not in terms of language families). Futhermore, English shares a great deal of similarity with Chinese, more so than any other European language I know of (Subject-Verb-Object, almost no inflectional morphology, fairly strict word order), so the language has nothing to do with it. Maybe the Japanese students in Nisbett's experiment just needed glasses. -- SteveConley

http://www.boston.com/dailyglobe2/045/nation/Debate_opens_anew_on_language_and_its_effect_on_cognition+.shtml (BrokenLink: 2002/07/18)

: "In English, time rushes forward. In Mandarin Chinese, it moves down. The past lies above, and the future lies below. So is the mind of a Mandarin speaker different from the mind of an English speaker? The question is one of science's loaded topics, a politically charged theory with a racist past. But researchers now say they are uncovering proof that it may be true."

uh uh...

: Another researcher has found evidence that languages which have many terms for color, such as English, give their speakers an advantage in remembering them.
: Critics say the findings are all small effects, well short of profound. "What happens to these neo-Whorfians is they keep backing off," said Lila Gleitman, a professor at the University of Pennsylvania. "Their position then becomes sufficiently weak that it holds no interest."

This last seems completely wrong to me. I can testify from personal experience that language is overwhelmingly important for conscious recall. Not having the language to describe songs is why it is impossible for me to consciously recall anything about them. I can recall words without any problems, but I cannot recall sounds or colours. Really?? I can recall sounds, tunes, colors, images, temperature, orientation, emotion, and a host of other things that don't express well as words. My brother finds it hard to understand that people don't recall spacial/temporal frameworks -- something I have to work at -- but I have no problem with components. -- gh

Above, many people talk about how one can think without words. This is certainly an exceptional case. It is an extremely rare occasion when my thoughts consist of anything other than words and spatial relations. In fact, words and spatial relations are so important (and everything else so unimportant) that many people believe them to be the essence of consciousness. (See JulianJaynes) -- RichardKulisz

I'm not versed in the literature, but it seems to me as follows. In order to consciously recall an image, sound, smell or concept, you need something else to remind you of it. Once you have a smell or color you can easily remember various related things, but really the only available ways to keep several things in place at once are sentences and diagrams. Sound can be done but only sequentially, and in practice I can remember things with songs but can only start at certain key points. So Jaynes' point should just be an obvious comment on the nature of the senses we have available to us so far, right?

However, if you can come up with temporary place-holding words quickly and easily, wouldn't that be enough to invalidate Sapir-Whorf? When you run into a smell, think of it as snippid, and after a few tries it should be as easy to recall as lemon.

That only works if you encounter the distinct smell often enough, in closely related situations. And it must be easily distinguished as a unique smell instead of as a combination of them. If you only encounter combinations of smells, then the lack of a language with which to discriminate and categorize them all means you can't divide and conquer the problem.

Sure, it would take some work to be able to discuss smells properly, but understanding smells well enough to recognize combinations or effects would take some work anyways. Most people don't work with smells very much, just passively receive them, which is why we think little of them and have few words describing them. But is it definitely the second which is preventing the first? As far as I can tell, the main effect of the limited vocabulary is that nobody can explain smells to me, and I haven't bothered to figure them out for myself.

This is easier to see with something like music. Over the centuries people have built up a large vocabulary of musical terms, which make it easier to teach others how to play or even to compose music. But a few people have also figured out how to become musicians on their own, maybe making up their own terms, or maybe just leaving certain concepts unverbalized. Of course, if they ever try talking to someone else they will have to agree on words.

What I get out of all this is that being given pre-packaged concepts in the form of words makes it easier to think about things. So, of course, does being given them in the form of aphorisms, parables, explanations, or diagrams. That idea seems to me to fall right back into the category of versions too weak to hold interest - more of a SapirWhorfTruism.

It's pretty clear that animals manage to divide the world of perception up into physical objects (that is, they form a mental map of the external world, inferred from sense impressions); and to organize these objects into categories (food objects, fear object, sex object) without recourse to language. --- For RichardKulisz, re: "... language is overwhelmingly important for conscious recall. . . .It is an extremely rare occasion when my thoughts consist of anything other than words and spatial relations. In fact, words and spatial relations are so important (and everything else so unimportant) that many people believe them to be the essence of consciousness. (See JulianJaynes)

Richard, your point exactly undoes your point. You point to JulianJaynes and refer to consciousness. But JulianJaynes himself points out that most of our actual "thinking" is not done in the consciousness. That performance of logic and puzzle solving, happen in the sub-vocal thoughts, not in the vocal, conscious thoughts. He makes it very clear that the speaking-while-thinking portion is the very small subset of our thinking. Here you are getting caught in the catch-22 of consciousness, to wit, consciousness is a little voice speaking into our hearing center, therefore any thought we are conscious of consists of words. But there is much more happening in our brain than a little voice speaking into the hearing center. There are 10^<big number> of other neurons firing concurrently. --AlistairCockburn

New Scientist (http://www.newscientist.com/news/news.jsp?id=ns99996303) quotes a study published in Science Express that appears to lend support to the SapirWhorfHypothesis (although it doesn't mention the hypothesis by name, only with a reference to LinguisticDeterminism). Members of the Brazilian Pirah? tribe, whose language does not represent distinct numerals above two, were given a number of tasks that tested numeric cognition (for example, matching the sizes of sets of blocks, or tapping the floor the same number of times as the proctor). Consistently members of the tribe did well up to numbers around three or four, but tended to have trouble at higher numbers. Of course, to invoke the SapirWhorfHypothesis in analysis begs the question of causality: do they not have those concepts because their language has not evolved to include them, or has their language not evolved to include larger numbers because culturally, socially, and cognitively they do not use them?

It's always interesting to see new material on the subject, but I'm pretty sad about this particular study. It's basically worthless. Everyone is poor at the kinds of tasks they tested (this has been studied to death), and we improve our performance by counting and remembering counts. If you can't count and then duplicate counts, you do poorly -- it doesn't matter why or how the counting was excluded. The authors of this study should never even have published.

Isn't that exactly the point? The tribe does poorly because they didn't count. They didn't count because they have no words for numbers > 2. Ergo, their language shapes their thought (that is, by limiting their ability to recognize larger numbers). Note that the article states that "they are otherwise similar to other adult humans"; that is, it does not extrapolate numeracy to deficiency in intelligence or other areas.

Quick, how many ones here: 11111111111111111? Oh, you didn't do too well, I guess you can't count very well, English must be deficient in number words. This is an easy to grasp exaggeration of the problems that come up with idiotic studies like the one above. If they had very carefully avoided this kind of problem, they would have found nothing to publish.
The key point of the study of this tribe really is that culture and language go together none strictly dominates the other, but both support each other. This is so for every language but the more complex culture and language are the weaker the dependency is. The interesting point of this tribe really is that the culture is not only unusual but also extremely simple and effectively excludes complex meaning systems (history, religion) and the language matches this with alike simple structures and no terms for anything abstract. They don't need it. On the other hand the language has features very useful, e.g. it can be whistled (it carries enough information in tone hights changes to be 'understandable that way).

If the relationship between language and thought doesn't extend beyond practice makes perfect, I think most people would agree the SWH is false. As such, this study doesn't say anything about it.

There is a story about a biblical king who wanted to identify the 'original' language of man. He had three babies taken from their mothers (one hebrew, one greek and one latin) and given to wet-nurses who were under strict instructions not to speak a word within hearing of the babies. Uncontaminated by other speakers, everyone waited eagerly to hear what language the babies would speak. All three babies withered and died.

On the other paw, there are documented cases of communities of deaf people inventing new sign languages: http://www.boker.org.il/english/newsignlanguage.htm or http://www.nerdshit.com/archive/2004/09/17/deaf_kids_in_ni/

The evidence is strong that people left together will develop new languages and jargons. That says more for our need and ability to communicate than it about the relationship betwen language and thinking, which is almost certainly circular.

There are sentances that are hard to translate into certain languages. For example, "by tomorrow, I will have been to Sydney" really maps very badly into Chinese. I know a native Finnish teacher who says that 'some thoughts are easier to think in English; I could not teach what I do in Finnish.'

What matters is - does a language have a word for the concept in question? If not, it takes more MilliEinsteins to express a concept. How well does the word map to the concept? Are there other words that denote related concepts, and sub-types of the concept? If not, it will be harder to be precise. Are there words that describe the wider concept the specific concept is a type of. If not, it will be harder to generalise. The difference between a Terrorist and a Freedom Fighter is significant, especially if the two are one and the same person.

If we don't have a word that means exactly what we want, we have to compose a sentance to explain ourselves. Think of words as methods. Without an exact match, we have to have build up what we want from multiple lower level terms. That distracts us from what we actually want to say. We can introduce new words, but that creates a learning curve for the listener.

We can think without words, but only up to a point. Words encapsulate concepts and enable higher level concepts. We need words to put order to our thoughts.

Consider: The Romans were great at counting, but useless at higher level mathematics.

The claim that "language shapes thought" seems so obviously true to me and so obviously false to many other people that I have to suspect a definitional mismatch.

You can stop right there, because that's just it. It means different things to different people. In particular, it means almost completely different things to linguists than it does to non-linguists, and even with linguists it nonetheless has a variety of interpretations.
So it's important to be as specific as possible, yet most statements of the SWH are not specific at all (unfortunately, including the several original statements).

Let me suggest an analogy: "roads shape driving". Obviously, roads don't shape driving absolutely. To paraphrase some arguments from above; cars drive in places that roads are not and some cars are very good at it. Many places can not be reached by roads. Not all roads can be driven on. We don't always use the same road to reach the same place. For a road to exist, it has to be built by off-road vehicles. And so on.

Most linguists believe that language does not follow roads, that they're more like dune buggys in the desert -- when it comes to semantics, at least. At an extreme is Jacques Guy, who believes that even syntax (let alone semantics) should not be considered to follow a set of pre-established rules, but that rather, natural language syntax is more like the most frequently trodden paths in a meadow, with nothing to stop one from walking on the grass outside the paths. -- DougMerritt

Driving can transcend roads. Yet the basic statement holds: where we build our roads determines where we will drive; most cars do most of their driving on roads, even cars designed for travelling off-road. Vehicles that build roads tend to be carried on road going vehicles, as are many off-road vehicles such as are trail-bikes. We can drive faster and futher with roads than without. Good roads make for faster, smoother, easier travel and encourage more travel by road, to such an extent that building new roads actually tends to increase congestion. And so on.

There is another aspect to the analogy: where we build roads is shaped by where we want to drive. This, I think, is our definitional mismatch: given that language is an attempt to express thought, how can anyone claim that it is thought that is shaped by language? The metaphoric answer is that it's a two way street.

While adequate roads exist, we use them. Otherwise, we strike cross-country, and if a route is taken often enough, it becomes a track, then a path, then a road. More importantly, it rarely occurs to most people to visit locations unreachable by road, or to build in such locations without also constructing an access road. Even cross-country walkers tend to drive to the start of the trail. We are not absolutely constrained to use roads, but in general, we do, and our range becomes much more limited once we leave the road.

Similarlly, we are not completely constrained by the language we have, but we can drill only so far into new concepts before we become overwhelmed. Our ability to grok nameless things is limited. We embed our understanding of the world in the words we use. Words create order from chaos and using a particular word calls up a particular ordering of reality.

We accept that the physical tools we use shape the way we work (HammerNail?). Why then do we have trouble accepting that the mental tools we use shape the way we think?

--BenAveling

Because most people would then assume (and even academics absolutely did assume, historically) that this means some languages are better mental tools than others, yet this is largely discredited. It's still assumed by many non-linguists; a number of Chinese-born engineers have told me over the years that they think English really is superior for engineering, while Chinese is superior for e.g. poetry. This is certainly fun anecdotal material, but it doesn't prove anything.
For engineering (and science and math etc) it is important to have a specialized jargons with a rich set of technical words. It certainly is obvious that jargon is a helpful mental and communicational tool -- but such jargons evolve naturally in languages where they are needed, including via borrowing (as is infamous in English, French, and Japanese in particular). The existence of a jargon is almost never what people imagine the SWH to be about, however, and it's a fairly trivial side issue.
Outside of technical concepts that the average speaker of a language doesn't know in the first place, it's been fairly well demonstrated that one can communicate ideas just as well in any of the world's languages, albeit more tersely in some than others (it's interesting that the word for "tea" was borrowed fairly directly from Chinese in most languages, and that a large number of languages have borrowed "ok" from English by now).
Bottom line: is it really so obvious that the SWH is true, if all languages are equally good "mental tools"? -- DougMerritt

That language influences individual thinking and culture in general is true beyond any doubt because it has been empirically validated. All totalitarian regimes used language to control the population, both at individual and societal scale. In US you only need to watch the cable news, or listen to talk radio (both left and right), to realize the level of language engineering that is used to shape the thinking of the target audience.

If all languages of the civilized world are equally good mental tools this proves that in absence of major constraints societies evolve equally good languages. In addition we have the modern experience where cultures are in effect communicating vessels, so any other outcome would be surprising. --CostinCozianu

Indeed true, which brings up two issues: first, the varying nature of the individuals; some are more swayed by propaganda, some are less so (in general one would expect a given population to follow a Bell curve); secondly, the inevitable rise of samizdat when those "totalitarian regimes used language to control the population"; the nature and existence of samizdat indicates a pushback (negative feedback) against the attempted controls, at least in a statistical sense, and mutatis mutandis et ceterus paribus, again a Bell curve, but note that the width and location of peak of Bell curve can make all the difference to a society, which I believe is ultimately what described the outcomes of perestroika etc. (I.e. stability vs instability of a system can change merely with a change in the peak and/or width of such a distribution.)

This would seem to be exploring a subset of the SWH, that of control by a totalitarian state, as warned in the novel 1984 amongst others. History seems to show that 100% control of 100% of the thinking of 100% of the population is quite out of the question; there are too many other forces at work, even if the state is largely successful -- and even then I would question whether it was language in particular that shaped the population's thoughts, vs simple social policy, which then required citizens to follow the wording of the official language.

By the time Orwell wrote 1984 it was no longer a futuristic warning, it was already a reality in Soviet Union. Language is the mechanism of control, and the most effective one at that. There's no question that language shaped the thoughts and not the other way around. When children are taught in school to refer to a group as the "exploiting class" versus the "working class" what do you expect them to think about it ? 100% control is not even necessary as evolutionary pressures at work are enough if the totalitarian conditions survives in isolation for an extended period. Language both acumulates social experience and knowledge, and facilitates the building of social experience and knowledge in an evolutionary manner. In a free society all kinds of different currents form their own idioms and branches within the language, and survival of the fitest competition ensues. If on the other hand the natural evolution is cut through official policy, the results can be absolutely predictable.
Of course I agree -- one would have to have been living in a cave to be unaware of the truth of that -- it's just that I do not regard this as the same thing as the SapirWhorfHypothesis. Consider that when you say "language shaped the thoughts", yes, true, but if you said "indoctrination shaped the thoughts" or "constant repetition of the same ideas shaped the thoughts", I'd agree with each of those paraphrases, too, and I don't see that the wording using the word "language" is paramount amongst such paraphrases. That's the problem with the SWH hypothesis; it is rarely if ever phrased in a non-trivial way -- usually it is stated in a way that is trivially true, or sometimes trivially false.
Those aren't as interesting as the underlying intent of Sapir and Whorf (the intent is partially recoverable from their writings). An example: a large part of the original intent of the artificial human language LoglanLanguage was to see if humans would think more logically if they used a language with a logical language (Loglan is modelled after formal logic, using predicates rather than e.g. verbs), as an explicit test of (one form of) the SWH. That form of the SWH is not trivially true or false; it may be right, it may be wrong, but the answer isn't trivial. That captures some of the flavor of what Sapir and Whorf meant by their hypothesis. It was originally concerning claims (now considered disproven) that Hopi language caused a radically different world view.
- Oh well. We do not seem to have enough expertise here to justice to the discussion. In addition, trying to do it justice here may turn out to be a copious waste of time. Maybe we should leave it for http://en.wikipedia.org/wiki/Sapir-Whorf_Hypothesis or other more specialized fora. We should perhaps refactor out anything that is already said better on wikipedia. But I think that SapirWhorf? has some special meaning for us that should be left here.
- From my point of view language influences thinking and is influenced by thinking, both at individual and social level. This should be trivial to understand for programmers from comparing Basic/GoTo (unstructured), Pascal (structured), C++ (OO), Scheme/Haskell (higher order), etc. It is well known that in absence of external constraints, languages evolve serving the needs and incorporating the knowledge of the people using them. This is nothing of a big deal. If you have bad tools for some activity you do it poorly, but you also tend to construct better tools. I don't think we need linguistic researchers or philosophers to tell us that, they should come with something more insightful of they can.
- I find it very ridiculous that a bunch of researchers thought they were qualified to design a better natural language. So they designed a "better" language with the idea to test the prediction that speakers of that language will think more logically. But if the prediction fails, the reasonable conclusion is that they are lousy language designers (you can't compete against millennia of language evolution that easily), rather than invalidate SapirWhorfHypothesis. -- CostinCozianu

I'm minded of a woman from Moscow (under the USSR in the 1980s, who later came to the U.S.) who told me that she met her personal State monitor (assigned to listen to her phones, etc) in a food line, and they had a nice chat wherein the monitor revealed knowledge of every aspect of her subject's life... the monitor viewed it as just a job to do, not a matter of ideology, regardless of how her superiors used the results of her monitoring. Weird. -- DougMerritt

Experiential -> Symbolic (Language labels a correlatable set of stimuli)
Symbolic -> Meta-Symbolic (Language as a concept unto itself talking about itself)
Symbolic -> Hypothetical Experiential (Language as a mechanism to think about correlatable sets of stimuli)
Experiential -> Hypothetical Experiential (Stimuli related to other stimuli)

I don't understand why it's one or the other when this sort of feedback loop seems more likely. When thinking in 'language' modes rather than 'non-language' modes there is a richer framework of abstractions to draw on that is for sure but it doesn't stop 'non-language' associations from firing.

Everything shapes thought. The authors are likely overemphasizing language because that is their specialty. I myself am mostly a visual thinker. Visual impressions shape most of my thoughts. Often when discussing something, I first turn language into visuals, process the visuals, produce a visual response, and turn the visual back into language. Sometimes that works well, sometimes not, depending on the subject matter. (I came very close to being in a visual field instead of "IT".) Other people seem to think more "linguistically", which I cannot define because I can't do it for the most part. And also cultural metaphors shape thought. I caught myself using "away mission" to describe my coworker's whereabouts. StarTrek idioms leaked into my communication process. (Trek didn't invent it, but is the heaviest user based on web searches.)

I stand a living counterexample. Due to certain conditions, I have found it necessary to redo the pathing in my own brain. Understand the thought-processes here are somehow lower-level than language communication but carry something that until I am given the language to unlock I cannot express. The necessity of operation depends not on the verbal processing circuits, nor the geometric processing circuits, so the thoughts cannot be spoken nor visualized but they yet exist. Understand if you will the motor commands in my brain are not normal because I have rewired them.

See BashirWorfHypothesis, ProgrammingLanguagesShapeThoughts, LanguageChoiceImposesSocialStructure, LinguisticDeterminism, NewSpeak, PhysiologicalAndPerceptualFactors

CategoryNaturalLanguage, CategoryPsychology