The Puzzling Paradox of Sign Language

http://www.technologyreview.com/blog/arxiv/24521/

by TriinT • 17 years ago

39 39 17 years ago

crux• 17 years ago

This is only interesting if one doesn't realize that sign language is its own language, rather than simply signed English. Syntax and word order are entirely different, and much information is encoded in what one might call 'tone'—eg, the energy or rhythm with which any sign is performed—that in English would be communicated with full words or suffixes.

These facts are patent to anybody with a passing acquaintance of American (or any other sign language). The 'paradox' they're discussing here is tantamount to saying, 'Japanese words tend to be longer than English words. So how do the Japanese communicate as effectively as we do?'

bugs• 17 years ago

This is always what I understood, you have words for letters but you only really use them when you need to spell something out otherwise you use other words. I have never learned any sign language but I assumed it worked in the same way, or else there would be no advantage to signing over writing what you want to say down.

TriinT3• 17 years ago

"This is only interesting if one doesn't realize that sign language is its own language, rather than simply signed English."

Define interesting. Personally, I find the entropic analysis of spoken Engligh vs. sign language pretty interesting, especially so when they relate it to channel capacity and other information-theoretic stuff. Your comment sounds anti-intellectual.

softbuilder2• 17 years ago

>Your comment sounds anti-intellectual.

Not a thing wrong with that. Fwiw, I file Technology Review with Wired and SciAm. That is, fluff. Hardly intellectual material imho.

>I find the entropic analysis of spoken Engligh vs. sign language pretty interesting

That would be interesting if they weren't using so poor an approach. Constraining the analysis to handshapes goes too far. Like constraining English analysis to vowel sounds, or removing prepositions.

Maybe the paper gives the wrong impression but it sounds like the parameters of the study were determined by someone with very little understanding of how ASL actually works.

(edited: formatting)

Tichy• 17 years ago

It would be interesting to discover that some languages are less efficient than others, and what effect (if any) it has on the economy. I think Japanese probably is more complicated that English - so why does it not appear to slow them down? Or maybe it slows them down in some respect (maybe having to sacrifice a lot of childrens play time to rote learning?), and they compensate in other ways?

lmkg• 17 years ago

Spoken Japanese may be less compressed than English, but written Japanese is considerably more terse.

hegemonicon• 17 years ago

I don't know about Japanese, but in Chinese, all the numbers have words that can be said extremely quickly - si, qi, etc. Because our working memory is implemented as a 2 second loop of audio, this allows Chinese people to have a working memory that's closer to 10 items instead of the famous 'seven plus or minus two' that English speakers have. In 'Outliers' Malcolm Gladwell posits this as part of the reason why asians often seem to be better at math.

lionhearted2• 17 years ago

Good comment. That said...

> 'Japanese words tend to be longer than English words. So how do the Japanese communicate as effectively as we do?'

I speak some Japanese. Japanese is actually a bit harder to communicate in than most other languages. The emphasis on social unity and modesty has evolved to make the language incredibly nuanced and complex. Nowadays, there's some factual statements that are almost literally impossible to say in Japanese because the words/phrases/grammar to say them don't match up.

Feynman touched on this in "Surely You're Joking" - there's different verbs for the exact same action based on the social situation and how well respected the person you're talking to is. So if you want someone to look at your garden, you say something like, "Would you please take-a-pasing-glance at my barely suitable garden?" But if you want to look at someone else's garden, you might say, "May I look-with-admiration upon you're beautiful garden?" If it's an official figure or a temple, it'd be something like, "May I quickly-glance-my-eyes-over-reverently on your most majestic of gardens?"

Different verbs. All mean "view" - but with different levels of politeness and appropriateness. One of the HNers who lives in Japan could probably comment more, I haven't spent all that much time there. Fascinating and beautiful language, but tricky to say things sometimes.

retroflexzy4• 17 years ago

> Nowadays, there's some factual statements that are almost literally impossible to say in Japanese because the words/phrases/grammar to say them don't match up.

Care to elaborate? As a English/Mandarin/Japanese speaker, I am certainly acutely aware of the limitations of translating from one language in to another, but find such an assertion hard to accept without concrete examples.

req2• 17 years ago

You're mincing words to tell us about Japanese. Your examples seem to indicate that Japanese more naturally conveys social situations and basic descriptions at the same time, making it more effective, not less. Russian has no future first person singular conjugation for 'to win'- this doesn't make Russian 'less effective' at communication, nor does English's use of "I will be able to" in place of "I will can".

If you want to make an argument on the effectiveness of communication, please use a quantitative or otherwise qualified argument, rather than exotic anecdotes.

christopherdone• 17 years ago

I did a ten-week course in BSL (passed with full marks!) and your assessment matches up with my experience, which was that it was like exaggerated enthusiastic telling of a story. It's like miming, too. Hearing people talk with their hands, bodies, faces, reactions, energy, rhythm, speed, etc. all the time already. They get along fine in clubs with loud music thumping away. Just watch someone chatting on the phone, they still make a tonne of physical gestures. And we've seen that textual interpersonal communication (i.e. pure English syntax and grammar) is hard, it's great for technical details or (with some effort) describing stories but sucks at that.

joe_the_user• 17 years ago

One obvious point that somehow isn't mentioned in the article:

The speed of human language isn't limited by the ability of "the language" to encode a stream of information. It is limited by the human ability to create and understand the bits of language information.

And that limit is much lower than the human ability to take in other streams of information. Vision lets you take in mega-bytes of information in seconds. Speech processing involves far less because language processing is such a hard problem for the brain.

Luria's the "The Working Brain" mentions that most brain damage degrades speech in some fashion and that's because such a large portion of your brain works on the speech recognition problem when you are speaking or listening.

And processing language is a hard problem for people (and computers!) because a language statement involves answering (at least implicitly) global questions about your store of information - "are all men are mortal", "do black swans exist" etc.

Hexstream• 17 years ago

"Vision lets you take in mega-bytes of information in seconds. Speech processing involves far less because language processing is such a hard problem for the brain."

I've read that people with extraordinary memory or mental arithmetic ability often visualize shapes and perform operations on them to get at the answer. And I guess "normal" people use "subvocalization" instead. In light of your comment, that makes sense.

vlisivka• 17 years ago

> And processing language is a hard problem for people (and computers!)

Processing of some human languages (like English)...

cbr2• 17 years ago

Calling asl "signed english" is usually a good indication that the writer doesn't know much about the subject. Then they go on to say:

> "It turns out that the information content of handshapes is on average just 0.5 bits per handshape less than the theoretical maximum. By contrast, the information content per phoneme in spoken English is some 3 bits lower than the maximum."

Talking about the information content of speech symbols is likely to be entirely bunk, but I'm going to go read the full article and try to find out whether the summary is bad or if the research is really this confused.

cbr2• 17 years ago

Just read the full article. It's no good. Analyzing ASL in terms of hand shapes only, on the grounds that they're the only feature that all linguists looking at ASL seem to agree on is really problematic. Computation of entropy is heavily dependent on the alphabet. If you say english has an alphabet at the syllable level, you'll get a very different amount of entropy than if you say the alphabet is phoneme level. If you go a level below phonemes and look at distinctive features (voiced/unvoiced, aspriated/unaspriated, front/back, high/low, ...) you'll get yet another entropy number altogether. These numbers will be wildly different from each other. Entropy calculations between languages must use comparably derived alphabets which cover the whole communication channel, or they will be meaningless. Argh. Junky article.

ars• 17 years ago

How would you calculate the theoretical maximum for a handsign? Or for spoken language for that matter.

mhb2• 17 years ago

What did they study? The accompanying photograph shows the alphabet, but I assume they aren't just considering using sign language to spell out words. Then they mention phonemes, but those letters aren't phonemes. Are there signs for phonemes? And ASL has signs for words.

jackfoxy• 17 years ago

I too find information theory a fascinating subject. Several commentors are disappointed by the trivial nature of the study (which I assume none of us have read). That is a problem with trying to apply theory to real world problems, You often have to make compromises in the quality of your interpretation of the problem in order to rigorously apply the tools of theory. I'm guessing the researches picked up some understanding of ASL during their research (if they didn't have some to begin with), but chose to frame their study so the parameters were easier to quantify.

sev2• 17 years ago

Along similar lines as the article, I've always thought about how we can be more efficient when speaking. If you notice, there is a tendency to be more and more efficient on the web, with abbreviations/acronyms. Imagine a world where we say "lol" just as though we type it (not hard to imagine). Now take it a few steps further, by considering the vast amount of different sounds/tones our vocal chords can create, imagine if we keep simplifying the spoken language to a point where it becomes like one of those star-trek civilizations with their clicking sounds when they speak.

I believe the trend is inevitible.

mhb2• 17 years ago

Based on how I, like, hear many people, like, talking, I'm, like, not so sure it's, like, inevitable.

sev2• 17 years ago

Like, such as, like, Miss Teen USA South Carolina, such as.

http://www.youtube.com/watch?v=lj3iNxZ8Dww

sethg• 17 years ago

There's a very nice sign in ASL to describe this article. It uses two hands, and one of them is a fist with the index and little fingers extended, like the horns of a bull.

(1) I'm pretty sure it was Klima and Bellungi's book The Signs of Language that pointed out that arm muscles are slower than vocal muscles, and therefore ASL does things that spoken languages can't do in order to maintain the same communication bandwidth. (That book was written ten years ago.)

(2) The handshape is only a very small part of what conveys meaning in sign language; one of the classic newbie mistakes in learning ASL is to look at your conversation partner's hands, rather than his or her eyes. A great deal of grammatical information is communicated purely by facial expression; for example, raised eyebrows can indicate a yes/no question, lowered eyebrows can indicate a wh-question, and just looking in one direction or shifting the body slightly can substitute for a pronoun. There are also movements of the mouth that act as adverbial modifiers for a sign, to indicate things like "almost", "carelessly", "with difficulty", "distant in time or space", and a whole bunch of other stuff.

(3) With regard to the hand and arm movements themselves, the location and movement of the signs are as significant as the handshape. The signs for "father" and "mother" differ only in location. The signs for "paper" and "cheese" differ only in movement. Skimming the article, it appears that the authors didn't bother taking location and movement because linguists disagree on how to categorize those other features. But that's no excuse for completely leaving them out of your analysis. That's methodological laziness.

(4) Modulation of movement also has grammatical significance which in English would be conveyed by modal verbs or adverbs. For example, a change in how you make the sign for "to be red" turns it into "to become red". The Klima and Bellungi book above has more of this kind of thing.

(5) There's also the ASL classifier system, which provides a concise way of using the relative position and motion of hands to indicate the relative position and motion of objects in physical or metaphorical space. I once saw a lecture at which a woman very eloquently used this to describe herself advancing through all four years of her college education while a friend of hers kept repeating her "prep" year. (Gallaudet has a pre-freshman year for students who, thanks to the ocean of suck that is the American deaf-ed system, don't arrive with adequate college preparation.)

There have been over thirty years of serious linguistic research into ASL, and judging from the references, these jokers didn't do more than strip-mine it for a list of handshapes. AAARRRGGGHH!

The Puzzling Paradox of Sign Language

Related Stories