SoylentNews is people
SoylentNews
from the intelligent-content-might-be-significantly-lower dept.
Human speech may have a universal transmission rate: 39 bits per second
Italians are some of the fastest speakers on the planet, chattering at up to nine syllables per second. Many Germans, on the other hand, are slow enunciators, delivering five to six syllables in the same amount of time. Yet in any given minute, Italians and Germans convey roughly the same amount of information, according to a new study. Indeed, no matter how fast or slowly languages are spoken, they tend to transmit information at about the same rate: 39 bits per second, about twice the speed of Morse code.
"This is pretty solid stuff," says Bart de Boer, an evolutionary linguist who studies speech production at the Free University of Brussels, but was not involved in the work. Language lovers have long suspected that information-heavy languages—those that pack more information about tense, gender, and speaker into smaller units, for example—move slowly to make up for their density of information, he says, whereas information-light languages such as Italian can gallop along at a much faster pace. But until now, no one had the data to prove it.
Scientists started with written texts from 17 languages, including English, Italian, Japanese, and Vietnamese. They calculated the information density of each language in bits—the same unit that describes how quickly your cellphone, laptop, or computer modem transmits information. They found that Japanese, which has only 643 syllables, had an information density of about 5 bits per syllable, whereas English, with its 6949 syllables, had a density of just over 7 bits per syllable. Vietnamese, with its complex system of six tones (each of which can further differentiate a syllable), topped the charts at 8 bits per syllable.
From the Abstract:
"Language is universal, but it has few indisputably universal characteristics, with cross-linguistic variation being the norm. For example, languages differ greatly in the number of syllables they allow, resulting in large variation in the Shannon information per syllable. Nevertheless, all natural languages allow their speakers to efficiently encode and transmit information. We show here, using quantitative methods on a large cross-linguistic corpus of 17 languages, that the coupling between language-level (information per syllable) and speaker-level (speech rate) properties results in languages encoding similar information rates (~39 bits/s) despite wide differences in each property individually: Languages are more similar in information rates than in Shannon information or speech rate. These findings highlight the intimate feedback loops between languages' structural properties and their speakers' neurocognition and biology under communicative pressures. Thus, language is the product of a multiscale communicative niche construction process at the intersection of biology, environment, and culture."
(Score: 2) by PiMuNu on Friday September 06 2019, @03:52PM (4 children)
> roughly the same amount of information
Pi
(Score: 0) by Anonymous Coward on Friday September 06 2019, @04:23PM (3 children)
It takes Italians nine syllables to say the same thing Germans can say in five syllables. Risolvilo di nuovo Tony (vs) Gebrochen.
(Score: 2) by PiMuNu on Saturday September 07 2019, @01:03AM (2 children)
My point is twofold.
Pi, as an irrational number, requires an infinite number of bits to express, and hence in one second it is possible to represent an infinite number of bits in words.
But the word pi will transfer a hugely different amount of information depending on the audience and the context. A non-mathematician won't get any information out of that at all, where as a mathematician may glean deep insight. What it all means in terms of number of bits is undefined.
The same could be said for a word like e.g. Vauxhall Corsa. The object that is a Vauxhall Corsa is very complex. Perhaps the authors would say that "Vauxhall Corsa" is represented in binary as a pointer to a car object. Are they passing by value or by reference?
So I call bullshit, this is some social science nonsense.
(Score: 1, Insightful) by Anonymous Coward on Saturday September 07 2019, @06:06AM
You're conflating message with protocol.
See why you didn't get it? It's not about what symbols can be exchanged at what bit-length representations. A given language is kinda like a prefix code system, but what the code words stand in for doesn't matter - they're arbitrary codewords which could be assigned to arbitrary meanings.
The point of this is that, when broken down into differentiators within a language, the bitrates between languages as spoke are similar. So maybe an inflection language might have fewer phonemes per second than a language without inflection, etc.
I don't know how you got the idea they were claiming things in the real world like Vauxhall Corsas could be defined with finite bit counts. No reputable scientist would publish a claim of a total coverage mapping from discrete to continuous.
(Score: 2) by maxwell demon on Saturday September 07 2019, @09:04AM
Wrong. Pi can be completely specified in a finite number of bits. It is true that if you try to encode it as digit string you'll need infinitely many digits to do so, but that just means that a digit string is a very inefficient method to represent pi.
Indeed, any computable number can be represented with a finite number of bits: A program that is able to calculate it to arbitrary precision.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by JoeMerchant on Friday September 06 2019, @04:19PM (9 children)
Back in the decades immediately following the invention of movies and TV, there was a widely accepted maxim that, in the US, Northerners spoke much more quickly than Southerners. This persisted as late as the 1970s, but mostly among the older generation who did not grow up glued to the broadcast networks.
Since the world has unified with instant communication and near-universal distribution of audio-visual media, it's not surprising that we're losing our accents and other local variations in communication, including socially normal data rates.
Watch some old Bogart movies and tell me what the data rate is when Lauren Bacall talks to Bogey... If I recall correctly, it depends quite a bit on the character she's playing.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 1, Informative) by Anonymous Coward on Friday September 06 2019, @04:32PM (8 children)
I can understand British English, New York English, Southern English, and West Coast English (me)... but I couldn't understand a single word from a back woods hillbilly from somewhere near the Ozarks. It wasn't a French accent like New Orleans, wasn't Southern drawl like Georgia, it was English like I've never heard.
(Score: 2) by JoeMerchant on Friday September 06 2019, @04:49PM
Florida Crackers have their own dialect as well - it's English, but not like any you've heard elsewhere.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 2) by DannyB on Friday September 06 2019, @05:08PM (5 children)
I have rarely had a few similar experiences in my life. Someone speaking English. But certain words / phrases were impossible to parse.
One example (while working pumping gas during college decades ago, this thing called 'full service').
What I think I heard: "do you have any candy smaps?"
What was intended: "do you have any kansas maps?"
(but I preferred questions like: how much is your 39 cent ice cream? Or what time is the 3 O'clock parade?)
In Judges 12:5-6 [biblegateway.com] is an example of how a difference in dialect could get you killed.
Amusingly the Hebrew letter ש can be pronounced with the "s" or "sh" sound. But modern markings in text indicate which pronunciation is intended so that you don't lose your life.
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @05:58PM
Take your fork handle and stuff 'em.
(Score: 2) by hemocyanin on Friday September 06 2019, @06:58PM (3 children)
When I was in HS, working at a grocery store in a small town in Vermont, I was facing shelves one day when a guy came up to me and we had this conversation:
Guy: "Do you have any ass?"
Me: "WHAT!!!???"
Guy: "Ass, do you have any ass?"
Me: [totally shocked look on my face]
Guy: "Ass -- you know, you put it in a cooler, put beer in it, makes it cold."
Me: "Oh -- ICE -- you want ice -- that's over there."
(Score: 0) by Anonymous Coward on Friday September 06 2019, @07:51PM (2 children)
Girl to cashier: Can I have a poke?
Cashier: ?
Girl to cashier: A bag.
A poke is slang for a bag.
(Score: 1) by Ethanol-fueled on Friday September 06 2019, @11:36PM
Vietnemese electrical engineers discussing Scottky diodes in English:
(Score: 0) by Anonymous Coward on Saturday September 07 2019, @05:56AM
Not slang, just archaic.
https://www.lexico.com/en/definition/poke_bag [lexico.com]
(Score: 2) by fliptop on Friday September 06 2019, @08:53PM
Oblig You like to see homos naked? [youtube.com]
And the Ozark hillbillies ain't got nothing on the Scottish [youtube.com].
To be oneself, and unafraid whether right or wrong, is more admirable than the easy cowardice of surrender to conformity
(Score: 2) by All Your Lawn Are Belong To Us on Friday September 06 2019, @04:41PM (1 child)
9 syllables per second is what they apparently average 39 bits per second to (5-8 bits per syllable), and then suggest this is twice as fast as Morse Code. Wonder how they made the leap between bits/syllables to words per minute. 20 words per minute (fast sending and receiving) is not really comparable to 120-150 average words per minute of spoken speech from several different sources. But 20 WPM is an arbitrary calculation and there are operators who can understand speeds in the 100-140 WPM range. The other shoe is that Morse isn't just hand sent / ear listened to anymore, and there are fast generators/receivers capable of better than speech speed interpreation and transmission, same as any other digital mode.
This sig for rent.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @07:09PM
This is reasonably straightforward estimation to make with some basic simplifying assumptions.
Claude Shannon showed way back in the early 1950s [princeton.edu] that written English has about 1 bit of information per letter.
It's hard to judge spoken English by the same metric because the actual words that come out of your mouth are only a small portion of the information conveyed by spoken communication. But if we completely ignore things like body language (let's imagine we are dealing exclusively with written transcripts of audio recordings taken out of their original context) then this is probably pretty similar to regular written English in terms of its information content per letter.
So by this, 39 bits per second is roughly equivalent to 39 characters per second or about 2.5k characters per minute.
Typists long ago standardized that a "word" is 5 characters, so 2.5k characters per minute is 500 words per minute. Suggesting this is merely twice as fast as Morse code seems within the realm of possibility. But it is a bit disingenuous as it assumes a morse code operator capable of over 200 words per minute, which is at world-record levels of proficiency...
(Score: 2) by LVDOVICVS on Friday September 06 2019, @04:47PM (4 children)
As court reporting student I wish this information would be more widely accepted. To pass the national test and officially be a reporter, one has to be able to take a five minute Q&A dictation at 225 words per minute with 95% accuracy. If you want to be extra fancy, there's a certification for being able to do 240 words per minute.
(Score: 2) by JoeMerchant on Friday September 06 2019, @04:51PM (3 children)
Court reporters have to deal with lawyers shouting over each other simultaneously...
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 2) by DannyB on Friday September 06 2019, @05:11PM
School teachers say: Put your pencils down, turn your paper over.
(but what if your exam paper is a mobius strip? what are you supposed to do?)
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 2) by LVDOVICVS on Friday September 06 2019, @07:23PM (1 child)
...deal with lawyers shouting over each other simultaneously... at 225 or more words per minute. Which is partly why you can still earn a good living as a court reporter in the age of voice recognition software. It still can't compete.
(Score: 0) by Anonymous Coward on Saturday September 07 2019, @05:38PM
At the risk of https://xkcd.com/793/ [xkcd.com]...
Seems like it should be amenable to hardware/software solution. Use a bunch of microphones in a phased array to identify the location of multiple sound sources, and compute separate audio streams from each source, feed each stream through its own instance of more-or-less normal voice recognition software, patched to tag the output with timestamps, and run all those text streams through a muxer/interlacer to format the output correctly.
A human is probably cheaper.
(Score: 3, Insightful) by Snotnose on Friday September 06 2019, @04:50PM (1 child)
My ex could transmit volumes at 0 bps.
I came. I saw. I forgot why I came.
(Score: 2) by JoeMerchant on Friday September 06 2019, @04:54PM
That's 0bps in the audible channel, I assume she was communicating with you visually... or by smell?
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 3, Interesting) by DannyB on Friday September 06 2019, @05:26PM (7 children)
<no-sarcasm>
Ever notice how various disciplines have their own jargon?
software dudes
plumbers
engineers
lawyers
NASA -- especially NASA. Everything is an acronym.
Does use of such jargon improve information transmission rate? That would seem to be its purpose. I have noticed this before in long jargon filled exchanges with others. It seemed like each jargon term was short but packed a lot of intellectual baggage and meaning.
</no-sarcasm>
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 2) by captain normal on Friday September 06 2019, @05:47PM (5 children)
Nope, just different syllables and rates, but still ~39 bits per second...if the article is right.
"It is easier to fool someone than it is to convince them that they have been fooled" Mark Twain
(Score: 3, Interesting) by JoeMerchant on Friday September 06 2019, @06:15PM (3 children)
So, the three syllables " D R D " do not communicate more efficiently than "the Department of Redundancy Department"?
The real question is: what is information? When you say: e equals m c squared, that can be shorthand for a tremendous volume of well known proofs, implications, etc.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 3, Interesting) by DannyB on Friday September 06 2019, @06:23PM (2 children)
That's my point. How much (pre-shared) information is "referenced" as I spout off the following acronyms or jargon:
AOC
SLS
SJW
MECO
RegEx
Variac
Boot
Compander
Log-log plot
Joule
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 2) by JoeMerchant on Friday September 06 2019, @06:26PM (1 child)
Link shorteners...
Also: Shaka, when the walls fell:
https://www.theatlantic.com/entertainment/archive/2014/06/star-trek-tng-and-the-limits-of-language-shaka-when-the-walls-fell/372107/ [theatlantic.com]
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 4, Interesting) by DannyB on Friday September 06 2019, @06:36PM
That's a great example.
Human ingenuity improving spoken communication efficiency when both parties are assumed to have a shared knowledge base.
Also consider this example, I think from an early BYTE magazine article on general AI:
P1: Hungry?
P2: How about McDonalds?
P1: Where are my keys?
P2: On the dresser
A lot was communicated in few words.
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 2) by DannyB on Friday September 06 2019, @06:15PM
In such conversations, the speech didn't seem to slow down as jargon laden sentences were spoken.
I would have to think this optimization (or compression?) gets some increase in information density?
Young people won't believe you if you say you used to get Netflix by US Postal Mail.
(Score: 1) by catholocism on Friday September 06 2019, @09:50PM
Unless the jargon is well mapped by both parties it still takes time to unpack.
I read AOC, and in my head I speak the words each letter represent. A word like schadenfreude however is pretty instant for me. Even then however, the form of the meaning is transmitted, but its full body still takes time to process out fully, rather than act as a known placeholder. Information transfer hinges on figuring out at what depth of meaning to engage, how much nuance is losable.
Travel outside your country, someone asks you where you are from, you name your country, maybe your state, possibly your city if its big or famous, but not your neighborhood, street etc. Someone from your area asks you the same question, you may directly reference your house.
Most information has this kind of stack of specificity, where all values are true, but the exactness varies. Jargon tends to pin the meaning n number of levels below the standard, though usually still not all the way down.
(Score: 3, Interesting) by nobu_the_bard on Friday September 06 2019, @06:32PM (1 child)
Does this research take into account nonverbal cues? I'm curious if that could be an unaccounted for element...
(Score: 0) by Anonymous Coward on Friday September 06 2019, @10:10PM
The answer seems to be no, noverbal elements are not included. TFA says they determined the rate of speech by recording people reading off a script, and timed how long they took to read it out loud. Doesn't sound like there was any measure of someone listening to the speaker at all.
(Score: 2) by Rupert Pupnick on Friday September 06 2019, @07:01PM (2 children)
39 b/s seems really slow or underutilized when the receiver has a bandwidth of around 15 kHz. So what’s the bottleneck? Is it limited by how fast the oral musculature can accurately make sounds? Is it limited by the speech recognition systems in the brain?
I note also that the paper distinguishes between information rate and speech rate. If the IR is roughly constant at 39 b/s, then the paper seems to imply that languages with more grammatical overhead (German versus Chinese, for example) are spoken more rapidly? Grammar functions to a very large degree like an error detection system, and it would be interesting to see comparisons between languages that consider overhead for error detection along with the error rate as you do in digital communications systems.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @10:56PM
For me, the limiting factor is definitely my ability to parse the incoming stream.
I note this often in TV ads. They get some motormouth going in the background. Takes him about two seconds to lose me.
I have to consider it like people who expect me to sign a contract with fine and poor contrast print on it.
It's been a pastime of mine, since retirement, that if bored, and have already been discouraged from purchase, have that salesman go over that contract line by line. He will bitch like all high heaven over how much time. When he does, I will nicely agree with him and line it out. Then he really gets worked up.
It may take a couple of hours for the most determined salesmen... The ones who won't take NO for an answer. How long did they spend to write this thing? And they expect me to read and be legally bound by it in a quick reading? As we part, I remind him I did everything I could to close the sale except agree to things I did not understand.
I've never had em call back.
(Score: 2) by deimtee on Saturday September 07 2019, @06:36AM
Personally I would guess the limit is speech recognition in the brain.
Normally in talking there is a lot of non-verbal information too, but its presence or absence does not change the rate at which I can listen. Given someone reading a transcript (with no non-verbal information) I would read it at least 3 or 4 times faster than I could understand a high speed reader reading it out loud.
No problem is insoluble, but at Ksp = 2.943×10−25 Mercury Sulphide comes close.
(Score: 2) by hemocyanin on Friday September 06 2019, @07:03PM (8 children)
It would be interesting to see a human language designed for optimal information flow. I wonder how much junk we have in grammar could be eliminated.
(Score: 2) by canopic jug on Saturday September 07 2019, @03:28AM (5 children)
It would be interesting to see a human language designed for optimal information flow. I wonder how much junk we have in grammar could be eliminated.
I'm not sure Esperanto would fall into the category of efficient. It is more like a simplified version of Spanish or Italian. The article places Spanish and English at about the same bit rate. Since Spanish is usually spoken much faster than English, from what I have heard, it would follow that it actually has a lower bit rate. Presumably Esperanto would be in the same situation. You do raise a good question about information-dense grammer. So I would wonder more about Ithkuil than about Esperanto. However, even natual language have not been heavily studied in this area yet.
Money is not free speech. Elections should not be auctions.
(Score: 2) by hemocyanin on Saturday September 07 2019, @07:36AM (4 children)
What I was thinking about was waste. In some languages things have genders which effect lots of other words in the sentence. Waste. Even in English, where we can use "the" instead of a gendered alternative, we don't really need to most of the time. Also in English, we waste a lot of effort on verb conjugation because we have so many exceptions to standard forms. Another waste of cognitive effort. A simple example:
Joe ran in the forest.
We could lose "the" here -- it really doesn't specify anything at all -- there are many forests in the world and so without actually naming one, or having one named in the context previously, it's a useless word. Then there is "ran". Now if Joe "walked" in the forest, the grammar nazis are cool, but do not say "Joe runned in the forest -- that's wrong. Why? Then there are those pesky double consonants which we could eliminate just by using the a "u" or "-" over "run" -- with the u-shaped bit, it sounds like a person exercising run. With a line over the top, it sounds like "rune" -- the norse characters. Or eliminate one of the pronunciation symbols and presume that one unless told otherwise. So we get, prsuming the short vowel sound since the symbol is absent:
Joe runed in forest.
Meh -- I'm no linguist. I'm just thinking that language has a lot of junk -- a lot of frills so people can sound pretty.
(Score: 3, Interesting) by deimtee on Saturday September 07 2019, @08:18AM (1 child)
Something about gendered nouns I hadn't considered until it was brought up here a few weeks ago is that they can be used to disambiguate pronouns. I don't know if that increase in efficiency compensates for the extra overhead but it is something to take into account.
Consider: "The rock fell on the car because it was in the wrong place" - Which was in the wrong place?
Contrast: "Le rock fell on la car because he was in the wrong place" - you know it was the rock in the wrong place.
No problem is insoluble, but at Ksp = 2.943×10−25 Mercury Sulphide comes close.
(Score: 2) by hemocyanin on Saturday September 07 2019, @03:46PM
Neat example -- I had never considered that.
(Score: 3, Interesting) by maxwell demon on Saturday September 07 2019, @09:29AM
Actually in speaking, there are many places where unnecessary stuff is shortened out. Like "it's" for "it is" in English. Actually your example sentence reminded me about that:
In English, that can't be shortened, but in German, the most common way to say this indeed includes a shortened form:
Here "im" is short for "in dem" (in the). But nodoby would say "in dem Wald" here, unless they want to stress "dem" (which makes it essentially "Jo ran in that forest").
On the other hand, note that some redundancy is essential in spoken language, since it has to work also in presence of some noise. Without any redundancy, the danger of misunderstandings would be too large.
Also note that in cases where maximum efficiency is required, people won't form complete sentences anyway. If the house is burning, you wouldn't shout: "The house is burning, therefore you should better leave it now." You'll shout: "Fire! Get out!"
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Rupert Pupnick on Saturday September 07 2019, @12:45PM
Calling it junk seems a little strong to me, but I think it is fair to ask how much value is there in all of the grammatical superstructure that permeates most languages. Again, I think of it as something like an error detection system, but how good is it really? The best you can hope for is that the listener asks the speaker to repeat because what was heard didn’t make sense. In many cases, though, you can commit grievous errors of grammar, but still get the message across.
Presumably all grammars came about from cultural/evolutionary pressures over long periods of time, but I don’t understand how.
(Score: 0) by Anonymous Coward on Saturday September 07 2019, @06:43AM
Well, I DDG'd "Heinlein SpeedTalk" which lead to
https://en.wikipedia.org/wiki/Speedtalk [wikipedia.org]
which lead to
http://ithkuil.net/faqs.html [ithkuil.net]
which I had never heard of before but which sounds interesting.
(Score: 2) by Bot on Monday September 09 2019, @05:10AM
>not considering greentext
wake up
Account abandoned.
(Score: 2) by Bot on Monday September 09 2019, @05:12AM
- hey you taliani speak fast, tatatata tatata hehehe
- no we don't, we just have longer words, herr fuehrer
Account abandoned.