Two remarkable studies at the Lyon Linguistics Laboratory have revealed some amazing aspects of human languages. A team of linguists posed an interesting question, given that some languages are spoken faster than others, asking: does this make them more efficient at transmitting information?
The French Les Echos website said in an article by Lean Verdeau that it is recognized that some people speak faster than others, explaining that these differences in the flow of words not only between one speaker and another, but also between language and another, as we do not need to bother to realize that Japanese and Spanish speak quickly.
A team of linguists from the University of Lumière / Lyon II language laboratory conducted an experiment published last month in Science Advanced, where linguists asked 170 speakers from 17 different languages to read a series of texts aloud, and applied to their recordings inherited analysis methods and tools. Of Claude Shannon's brilliant information theory.
The first observation was that the fact that some languages seem faster than others in the ear is fully justified, the rate of syllables spoken per second ranges from a certain number to twice, with 8.03 sybps per second in the Japanese language and about 5.25 syllables for the Vietnamese language to 4.70 syllables The Thai language, says the paper, noting that these differences are not related to geographical distribution because the Asian languages fall on both ends of the spectrum.
effectiveness
Since this does not illustrate the effectiveness of languages supposed to be greater in the transmission of information, learners have measured other criteria that are difficult to understand, such as the cross-sectional density of information that the linguist Francois Pellegrino, the lead author of the study, explains. Provides a bit of information in the sense that Shannon meant. By contrast, if it is difficult to predict, it offers a lot. "
To illustrate, Pellegrino cites the example of present-day French: "If you read the word" for ", this is a safe bet that the next passage will be" that ", and just the fact that you are able to predict it almost with certainty, meaning that its information density is almost non-existent.
Two strategies
Not all languages encode the same average amount of information measured in bits in each of their passages. “The average density of information varies across languages at an interval of 5.03 bits per section in Japanese and 8.02 bits for Vietnamese,” says Pellegrino.
This means that the average density of information in Japanese is about 5 bits per segment, meaning that the prediction of the next segment based on its predecessor amounts to making the right choice among 32 (25 probabilities); in Vietnamese it means making the right choice between 256 (28 probabilities), So it is 8 times easier in Japanese than in Vietnamese, in other words, the density of information in Japanese is 8 times lower than in Vietnamese.
The study showed that these criteria, ie the rate of sections measured by the number of sections per second, and the density of information measured by the number of bits in the section, are inversely proportional, as the large section rate is associated with low information density and vice versa.
The website explained that this proportionality results in a very distinct phenomenon is that the rate of information of one of the languages, defined by the above criteria, fixed in every corner of the globe, was established at about 39 bits per second, no matter how different, whether it seems fast or slow to Ear, that is, all languages spoken on the surface of the earth transmit in a certain period of time the same amount of information.
"To be effective in terms of information transfer, language has a choice between two opposing strategies," Pellegrino says. "It either prefers a high percentage of speech with a low density of information, or vice versa. In this regard, French is a" medium "language with 6.85 syllables. The second has an information density of 6.68 bits per second, which allows access to an information rate approaching 39 bps.
Optimal threshold
The fact that this last value is universal means that it owes nothing to chance, but, according to the site, is severely constrained by our cognitive abilities and the way the brain handles language. A language well below this 39 bits per second will not allow speakers to deal with the complexity. If it exceeds this rate, it will increase the cognitive load and the person cannot sustain the production or processing of such a lot of information permanently, and will face the same fate, so this threshold of 39 bits is considered biologically and culturally safe on Both determines the language portability area T. Human survival.
The site pointed out that this vision of languages is more impressive because the language - away from stillness - is constantly evolving with the passage of time, and as a previous study of a laboratory in the city of Lyon, there may be the emergence of new sounds, which doubles the total number of vowels owned by the language. Of sounds and more syllables increase the density of language information.
Noting that this increase could leave the language above 39 bits per second, Pellegrino said that is not the case because “our hypothesis is that whenever something changes a language from its cross-sectional density of information, this change also drives its speakers to reverse the speech rate in order to maintain On the optimal threshold of information so you don't die. "