<< Adam Bede: Conclusions? | Front Page | Cartoon Centennial >>
Monday, August 18, 2008
Holy Gavagai, Batman! Letterman’s Out of Control
Blockbusters: In the wake of being dissappointed both by Iron Man and The Dark Knight and by the critics who love them, David Bordwell speculates on the rise of super-hero movies. He refuses to blame the zeitgeist - being generally against such explanations - and instead lists eleven contributing factors, including a changing hierarchy of genres, rise of franchises, shock and awe, hambones eclipsing stars, “the mecha look,” and “the dark.” But Hellboy II gives him hope.
Language: Willard van Orman Quine is widely known for an argument that starts with some linguist encountering a native informant who utters “gavagai“ in the presence of a passing rabbit. Quine then adduces a passal of arguments as to why the linguist is not justified in concluding that gavagai means rabbit and why it should be extremely difficult for him or her to make any sense out of what the informant says. Here’s a story about the late Kenneth Pike, a highly skilled and experienced linguist, in a similar situation. “Barely fifteen minutes into the elicitation, he has the rudimentary alphabet, presumably a phonemic script of some sort, figured out; he has vowels, consonants, word initial stress; and due to the strange behavior of the plural suffix k, he begins to guess that this is a vowel-harmony language. Moreover, he has taken steps to discover and to describe the morphology of the language, which—lacking gender and number agreement—seems to be an agglutinating Asian tongue.” What did Pike know that Quine didn’t?
Fiction: Authors may create their characters, but they don’t always control them, a brief note, with citations, by Keith Oatley.
Performcance: Finally, an hour-long video in which Charlie Rose interviews David Letterman in which Letterman talks his career, about doing a show, about Johnny Carson, and about executives. At roughly 29 minutes in Letterman explains how the best things are utterly unplanned and unexpected, how they start out small, and blow up to illuminate the rest of the show.
Comments
Bill, I think Quine and Pike are doing different things, ‘tho we call each “translation” (which is sort of Quine’s point).
Quine’s example shows us that translation is impossible because we can never start from ground zero with a new language and be sure we “mean” the same thing a native user “means” by the same words. Ultimately, this also implies that when we first learn our native language, there’s no reason to assume that the teacher and learner share something in common we can call “meaning.” We just know that when we use words in certain ways, we can count on normative behavioral responses.
So Pike, for example, might learn that in English, it goes, “One egg, two eggs, three eggs” and so on. So he induces and deduces various laws of grammatical and morphological transformation, and he begins to accumulate a lexicon, but he never knows if he and the native mean the same thing by “egg.” If we think about it as a Venn Diagram, we can say that Pike and the native discover a sliver of shared meaning (the sound “egg” refers at times to the white thing from a chicken’s hoo-haw) but that each speaker has a large area of unshared meaning. Quine would say that we might as well get rid of this talk of “meaning” and say that under certain conditions, Pike can predict that using a certain word will lead to a certain normative behavioral response in the native and vice versa.
This is why some would say that the “meaning as intention” argument is bound to fail. The connotative scatterplot in my brain as I use a set of words will never mirror that in another’s brain when they hear those same words. No one will ever really know what I mean, in that sense.
Still, there’s a way in which Quine could be tempered by a certain falliblism. His gavagai example is, like Hume’s extreme examples of the sun rising from day to day, a limit case of induction. Pike’s work shows the importance of deduction from a set of reliable experiential and academic givens about language. We might say that the Venn diagram is more one of shared spaces, with thin outlying areas of private association, and thus redeem the idea of meaning. Quine is right that, at the limit, we could never know for certain if one translation is right. But pragmatically, we can say that given the relative stability of the universe and the cultures from which we receive language, chances are there is less private divergance than Quine might suppose.
On a separate note, I love the Bordwell piece. I would quibble with him about zeitgeist readings. Sure, simplistic ones are bad—“After 9/11, we all need heroes.” But his own list of shared characteristics points toward a zeitgeist in which viewers privilege simulation and spectacle over individual style and meaning—which is exactly the zeitgeist Jameson describes formally and thematically as postmodernism.
It may simply be that Pike was interested in actuality (reality, things, objects) rather than simply in scholastic angels-on-the-head-of-a-pin arguments. It may be that Quine’s mistake was talking about a situation that might possibly be actual, and that his successors determined themselves never ever to make that mistake.
If someone actually wants an answer to the question “What does strange person X speaking unknown language Y mean when he or she says Z”, there are actual steps he or she can take to find out. But only if he or she wants an answer.
Quine also probably failed to realize that the extreme differences between cultures and languages are located in certain specific sensitive places, not everywhere, and don’t make all communication impossible.
But of course I’d say that.
Bill, I think Quine and Pike are doing different things, ‘tho we call each “translation” (which is sort of Quine’s point).
Yes, they are doing different things. Pike is doing translation while Quine is talking about translation. If Pike’s situation were like Quine says it is, then he’d have gotten nowhere.
Bill, I think that misses Quine’s point, which is that Pike can never be sure that what he’s doing is what we commonly think of as translation. Which is to say, Pike can never be sure that “gavagai” means “rabbit,” even if it seems to perform the same function most of the time as the word “rabbit” does. Quine’s point is that translation cannot be thought of as a matter of meaning or synonymity because there’s simply no way to know completely what a word “means” in that sense—or what we really mean when we talk about “meaning” in that sense.
In Spanish, “casa” means “house” and “home” in English. We have no way of translating “casa,” really. And to a particular speaker, “casa” means something fuller, richer than it means to another speaker. If we think about communication as a matter of meaning, of mind-states, then for Quine, communication must always fail. I can’t get you to see what I “mean” by “home.” There’s a limited Venn diagram space where two speakers might agree. They might go on for weeks or years thinking they have agreed. But at some point, it becomes clear that the two speakers mean something slightly different. (This happens in marriages all the time. Words like “attention” or “romantic” work until they don’t work.) That’s why Quine’s not just talking about translation—his thought experiment works for language learners in one language as well.
I’m not sure I see the value of Quine’s insight if it doesn’t make any discernable difference except in extreme examples.
You point out that in Spanish, “casa” means something slightly different than “house” or “home” in English, and that in English, “romantic” can take on different meanings for different people. Because of this you say “we have no way of translating casa, really.”
But what does that mean? In the vast majority of real world scenarios, we DO have a way of translating “casa,” don’t we?
CH - Consider the famous saying, “Mi casa es tu casa.” We commonly translate that as, “My house is yours,” or, less commonly, “My home is your home.”
But “casa” in one word contains the connotative differences between “house” and “home” in English.
So the American visitor to the Spanish home might feel that all the amenities of a house are being shared. Feel free to lounge on the sofa, to eat the plums in the food box, make yourself comfortable.
But the other offer that might also be on the plate, above and beyond material amenities, might be something deeper: “While you’re hear, consider this your home.” And with that word “home” comes all the connotations of rootedness, stability, and safety that Americans associate with the word “home.”
Quine seems to suggest that it is only through further discourse, back and forth, that the visitor might begin to glom a sense of what offer is really being made. But each of those supporting discourses are subject themselves to further ambiguities of translation.
Some theorists connect Quine on the indeterminacy of translation to Derrida’s notion of interminable hermeneutics, of how the chain of signification extends outward at every moment of a piece of discourse. There’s no final sentence, from which we can go back and tie up the discourse into a neat, meaningful package that is now open to translation.
The problem, Quine seems to argue, is that we commonly think of meanings as ideas in people’s heads, and a translation is when a word from one mental-set can map simply onto the same ideas as a word from another mental-set. Instead, Quine continues, it’s best to see how words tie to behaviors, how they get things done. Words that achieve the same behaviors can thus be translated as such.
So over the course of a visitor’s stay at your house, you negotiated back and forth over what the original offer could mean. When the visitor’s behaviors are those that the Spanish speaker sees as the true effects of “mi casa es tu casa,” then we can say that something like translation has taken place. But it’s not a simple word-to-word recoding system, nor is it a matter of putting new lexicons into new grammatical arrangements. It’s a matter of words being understood in such a way that the proper way of life modeled by those words is adopted by the other.
Gavagai might mean rabbit quite easily in one situation: “I want to catch a gavagai for dinner.” But that doesn’t mean that we’ve exhausted the other possibilities of behavior that might follow from a gavagai-centered discourse. Say, for example, a warren of three black gavagai has deep spiritual significance. Pointing to a dinner platter and saying “three gavagai” might mean something terribly different to the native speakers.
That’s my sense of the issue at stake here.
Machine translation in action:
http://languagelog.ldc.upenn.edu/nll/?p=301
Luther understands Quine here.
I’d say the simplest answer is that Quine is talking about semantic deduction and Pike is doing syntactic deduction. As far as the semantic aspect, Pike is using simple objects that were picked because he felt that it’s reasonable to assume that the concepts surrounding them are similar to his own. He has no guarantee of that, but it’s a reasonable assumption.
Quine’s argument is not so much that translation is impossible, but that infallible, isomorphic type translations are impossible. Functional, behaviorally deduced translations are all we have to work with.
"Quine’s argument is not so much that translation is impossible, but that infallible, isomorphic type translations are impossible. Functional, behaviorally deduced translations are all we have to work with.”
Thank you, and the same to Luther. I think I get the point now.
Holy Gavagai! “Thank you,” in a Valve comment thread? Whoa! Can the apocalypse be far behind?
BTW, here’s some remarks on translation by Martin Kay, one of the grand old men of machine translation:
http://www.stanford.edu/~mjkay/CurrentState.pdf
He seems to think of translation as a negotiation between the two languages.