Monday, August 15, 2005

Mentalese - Reading 1

So, here we go with Reading One for my course this semester - Intro to Linguistics. Since it is my first semester, I am using the course book they give to me with little change. Reading One is Mentalese, a chapter in Steven Pinker's nonfiction bestseller The Language Instinct. Before we get into it, here is a discusion of Thomas Jefferson.

This comes from an NPR interview with someone who had written a book about the American Founding Fathers. He made the point that it is too simple to look at Jefferson's flaws and simply pronounce him a hypocrite. On the one hand, he writes of the equality of all men, but on the other owns slaves. Horrible, right? Well, what is weird about it is, in this author's terms, the language we use to condemn Jefferson is due in large part to Jefferson himself. In other words, it is the thought of this hypocrite which let's us see his flaws. He helped persuade us all of man's equality and without him, we might not see slave owning as wrong at all.

Take that as you will.

The topic of this reading is "what language do we think in?" As an English speaker, do I think in English? And does an Apache speaker think in Apache? Pinker starts and closes the chapter looking at Orwell's Newspeak (1984 - that's a book, not a citation), which, according to Orwell, is a carefully designed language created so that thoughts not approved by the Government would be impossible. Pinker also mentions reform movements in language to help improve our thought. And then the popular perceptions, associated with the names of Sapir and Whorf, that Eskimos think differently because they have 100 words about snow or such. All of these people are saying that we literally use language to think. But, according to Pinker, "it is wrong, all wrong." (P, p.46).
Pinker says there is no scientific evidence "that languages dramatically shape their speakers' ways of thinking." This concept of linguistics determinism (language determines our very thoughts) is associated with the names of Sapir and Whorf, though Pinker largely praises Sapir's work on Native American languages and curses the arguments that his student, Whorf, made from them. Quoting Whorf, a la Pinker:

We dissect nature along lines laid down by our native languages. The categories and types that we isolate from the world of phenomena we do not find there because they stare every observer in the face; on the contrary, the world is presented in a kaleidoscopic flux of impressions which has to be organized by our minds - and this means largely by the linguistics systems in our minds.

So Whorf does OK here, until he makes the mistake in the last sentence. Yes, our mind has to organize the sensory input, but why is this using a linguistic system?

Pinker goes on to show how many of Whorf's claims are faulty. For instance, Whorf claimed that Apache people think differently because when you looked at their language, it was structured differently than English. But Pinker mentions that Whorf's translations are deliberately awkward. You could just as well translate them into a more familiar sounding sentence, and, moreover, you can express English in a terribly obtuse way as well, but that doesn't mean we think differently. Pinker then looks at color words. Not all languages slice up colors in the way English does. "Latin lacks generic "gray" and "brown"; Navajo collapses blue and green into one word." There are even languages with only black and white, and nothing else. Does this mean they perceive color differently? In the end, no, we all see color about the same due to the way our retina's are constructed (Pinker doesn't deal with color blindness). Moreover, there are patterns common across all languages about what colors will be missing. Finally, they asked a New Guinean group whose language only had black and white, to identify reds and the fire engine red we all know was considered by them to be the most red of all the reds shown.

Whorf also just had some plain false claims. He claimed that the Hopi had no words about time, and that they thought of time differently than English speakers. However, Hopi actually do have words about time and they are quite common. No one knows how Whorf came up with this.

Then finally the myth of the Eskimo's hundred words for snow turns out to be pretty much bogus. He traces how the urban myth developed. It turns out that the Inuit don't have many more words than we do for snow. Then, he makes the bigger point, quoting an article by Geoffrey Pullum, that, in the end, so what if they did? Printers have hundreds of words for fonts. Horse breeders name different types of horses. Interior decorators have names for a variety of mauves. So what? Does that mean that we all think differently than printers, horse breeders, and interior decorators? (There is an equivocation, or at least loose thinking, here that Pinker is making which we will come back to.)

Pinker then turns to various experiments which have purported to show that language changes how we think. The most dramatic is Alfred Bloom's experiment. English grammar has a "subjunctive" mode, which is for things that might happen, but didn't. You use it by saying things like "would, might, may, were to" etc. Not all languages have this. Chinese is one that does not. So Bloom constructed a story with tons of subjunctive sentences, which he had both Chinese and American students read, in their native language. Then he asked them if the things descibed actually happened. 98% of the Americans said 'no'. Chinese students only got it right 7% of the time. However, there turn out to be problems in the experiment. First, it was really bad Chinese that he had provided to the Chinese speakers. Second, there were in fact ambiguities in the story that the Chinese students all picked up on, while the Americans missed it. When you fix these problems, the differences disappear.

Pinker moves next to showing all of the ways that we think without language. Babies do all sorts of amazing thinking before they learn to speak. Monkeys can identify clanship and extended causality through time. He discusses people who have never spoken with anyone but clearly still think. People think in images as well, all sorts of scientists and poets have discussed their thinking in images. Morever, there have been experiments where it is clear that people are mentally rotating an image in place to solve a problem.

Pinker then presents the Turing machine, the invention of the all around CS guru Alan Turing. The Turing machine can, given enough time, solve any problem by taking inputs and restucturing them into an output following a set of rules. No individual item within the Turing machine is intelligent, but the rules governing the manipulation of the symbols creates intelligence. Then Pinker comes to a bold, and now quite controversial, claim:

"This, in a nutshell, is the theory of thinking called "the physical symbol system hypothesis" or the "computational" or "representational" theory of mind. It is as fundamental to cognitive science as the cell doctrine is to biology and plate tectonics is to geology." (P, p. 68)

So, now Pinker has told us how our minds really work, how we really think, and it is through the manipulation of symbols. Are these symbols for English speakers in fact English symbols? Actual English words, which our brain moves around to reason? For a list of reasons, the answer is clearly no.

Pinker finishes by coming back to Orwell and saying that NewSpeak will not be the end of all thought, since we use language to get our thoughts out, but we don't think in it. Languages change, and the children of Newspeak speakers will change the language to do what they need.

Most of this is all fine and good. Here are some problems.

1) The simple symbol manipulation theory is wrong. I'd recommend Andy Clark's Being There to see some useful alternatives to Pinker's theory, especially as he addresses neural networks and embodied thinking.

2) Pinker has clearly set up Whorf to be absurd - typical straw man argument. While Whorf's original claim, quoted, goes too far, he obviously doesn't believe we are all literally brain dead without language. In the quote he even throws in the "largely" caveat - "largely by our linguistic system". So the fact that monkeys and babies have a mental life doesn't show that language has no impact upon our thinking. In fact, it would be quite surprising if it had no impact at all. Learning, learning anything, is literally a restucturing of the brain. Much of the restucturing is small - learning a single word to go with the 20,000 we already know doesn't rewire large portions of our brain, but some learning is quite large. There are even gross experiments where they send visual input to a cat's auditory areas of the brain and through time the brain rewires itself so that the hearing parts of the brain transform into being capable of some form of sight. Or just thinking in folk philosophy terms, we learn to be who we are by our experiences, largely in childhood but beyond as well (yes, yes, of course mediated by our biology in quite constrained ways; see Pinker's Blank Slate). A traumatic experience with a banana and a monkey makes you afraid to go into an ice cream store. Why would language be unique in having no impact whatsoever on our thoughts and beliefs, and thus on the way we think. So Pinker's basic mistake goes hand in hand with the symbol hypothesis. In his conception, we have a processor, which is universal to all humans. This is Pinker's "thinking". Then we have a bunch of things we think about, which have no impact on the processor. That processor is how we think, and it does not change, Pinker appears to assume. The problem is that there is no absolute distinction between how we think and what we think.

3) The equivocation is on this basic word "think". There is no such single thing. Remember the interior designers and printers from above with tons of words for things that all seem the same to us? I would argue that in fact they do "think" differently than us - about interior design and printing. Of course they do. Theoretical physicists think differently about stars than I do too, because they are trained on this topic. This only becomes alarming if we think there is one thing "think" and "oh no it is now different! I have no hope of ever understanding a word a printer says ever again!"

(Reminds me of a personal event several years back. I had spent a semester studying in China, and some family members wanted me to make a comment on "the Chinese mind". I think there are books with titles like this. I couldn't think of anything useful to say. How do you categorize 1.2 billion people? As far as I could tell they "think" they same as I do. So I said, "I have nothing intelligent to say on that." Now that answer was a good one if the point of conversation is to pass information across. Of course, that is not the entire point of conversation. It is also a social even, and on that account my answer was terribly rude. Hopefully, I'd do bettern now.)

So Whorf and Pinker make a similar mistake. Whorf sees that language has an effect on parts of our behavior and argues that language is thought. Pinker sees the universality of thinking apart from language and claims language has no impact on thought whatsoever. Both are wrong.

6 comments:

kristybox said...

I am embarrassed to admit that I read this entry in its entirety, and enjoyed it.

Silentknyght said...

I read the entry in its entirety (like kristybox), but it's difficult to make general, observational comment on such a long post. It's as if I were attempting to interject questions into someone's run-on sentence. However, I will do my best.

Yes, both are wrong, but I found that conclusion unenlightening and almost predictable (sorry, I generally predict people gravitate towards the middle of an argument rather than the side). Anyhow, that's not what was interesting.

It would agree that both language and symbolism affect perception, however it is a more interesting question (to me, and perhaps to a linguist as well), if the communication of those thoughts is dependent more upon the symbolism or upon the language.

I suppose I'm taking a pragmatic viewpoint--but from my position, the importance of linguistics doesn't lie in the perception of the world, just in the conveyance of the perception of the world.

But, perhaps you saw that conclusion, too.

pacatrue said...

Welcome back, silentknyght. One of the problems I am having with these posts is that I want to explain things to people who don't study linguistics, because if I can't explain it, then I don't really understand it. At the same time, how do I explain the key concepts in a 30 page chapter and keep it short?

Anyway, knyght asks the question:

"I would agree that both language and symbolism affect perception, however it is a more interesting question (to me, and perhaps to a linguist as well), if the communication of those thoughts is dependent more upon the symbolism or upon the language."

I think it always has to do with exactly what we are talking about. So let's take perception. I know hearing best because I focus on the sounds of language. Humans have the raw capability to hear amazing nuances in the fluctuations of a sound wave. In tests, they can literally distinguish between hundreds of different pitches within a single musical octave under controlled conditions. But most people cannot identify more than a handful in a musical scale. In language, we all have the raw ability to learn any sound in any language in the world if we grow up that way; however, that ability can be severely compromised as we learn.

So, take Chinese, which is called a tonal language because, apart from the consonants and vowels, words are distinguished by a pitch pattern. So ma with a high level pitch is mother. Ma with a falling then rising pitch is horse. And ma with a falling pitch is "to scold". All of us, with the exact same genetics, would hear this no problem if we grew up hearing Chinese. But as English speakers, they become almost impossible to hear, and it can take weeks if not years of training as an adult to hear this. They have done neuro-imaging studies of English and Chinese speakers listening to tone. It turns out that English speaker use a different part of the brain than Chinese speakers do in hearing tone. English speakers use parts of the brain which also fire with intonation and melody. Chinese speakers's brains fire more on the linguistic naming side of things.

The point? Due to our experiences learning language, our brains literally hear things in different ways. (Incidentally, once English speakers learn tone, their brains function, for this aspect, very similar to the Chinese speakers as well.)

Does this mean we think differently? Yes and no. I would argue there is no one thing "think" which is to be the same or different. And since there is no single "think" we can't answer if it is particularly using language or symbols.

Killer Llama said...

Yo, Yoda,

I agree with Silentknyght on the conclusion. People do gravitate towards the "middle way." I find I do it all the time as well. Given the complexity of just about everything, usually truth lies far from the edges that define a problem. However, it is also a bit lazy to simply proclaim that both are wrong... far more interesting would be to hear what you propose is the correct answer.

I also find you guilty of making the same error that you accuse Pinker of making. In one spot you write:

"Pinker has clearly set up Whorf to be absurd - typical straw man argument. While Whorf's original claim, quoted, goes too far, he obviously doesn't believe we are all literally brain dead without language. In the quote he even throws in the 'largely' caveat - 'largely by our linguistic system' . . ."

Then, later, in you write:

"Whorf sees that language has an effect on parts of our behavior and argues that language is thought. Pinker sees the universality of thinking apart from language and claims language has no impact on thought whatsoever."

Though you do not explicitly state that Whorf argues that thought is exclusively defined through language, by setting this up as a dichotomy with Pinker who "claims language has no impact on thought whatsoever," the implied position is that Whorf thinks that language is necessary for thought.

Personally, I found the most enlightening bit of this piece to be when you told the story of your personal conversation with your family about the Chinese mind... One could even argue that the sole purpose of that conversation was social, and anything you said to engage their attention would have sufficed. If that's true, then one would have to separate the study of communication of ideas with the study of communication of ... communinity, I suppose. Perhaps language is designed for the communication of factual information; emotions, personal ties, mean of interacting with other humans is probably best conveyed - and thus studied - completely independent of the meaning of words.

Maybe this then comes back to your original discussion. Perhaps logical, rational thinking is shaped by language. After all, you cannot write a computer program if you don't know what a computer is. But irrational, emotive, and creative thinking is language independent.

Killer Llama said...

BTW, in case my last comment came of too critical, let me also say that you did a great job of explaining a complex subject in a compact form. I think that's evident by the fact that at least three folks were able to digest the subject and comment intelligently on the matter... so good job :)

Silentknyght said...

Hmm. Your argument based on hearing is falling on deaf ears, so to speak. Haha.

As an individual, I am a visual person as opposed to an auditory person; that is to say, not only is it is difficult for me to understand your position well enough to comment on it, but also that my original comments were drawn from thoughts of visual archetypes.

Of course, this reinforces your notion that "there is no single "think'".

I like llama's take home thought:
"Perhaps logical, rational thinking is shaped by language... But irrational, emotive, and creative thinking is language independent." That seems to appeal to me.