3

not a linguist. I was just wondering if the degree of robustness of a language to environmental noise is somehow measured or studied. I presume not every language is equally robust, right?

1

This is a currently non-existent but potentially interesting specialization in linguistics. The design of the experiment would not be very difficult: make good quality recordings using a head mic, standardized recording parameters and multiple speakers, speaking normally (according to norms of that culture, which is the main thing that matters) – don't normalize amplitudes (since that would obscure a feature of interest). Then overlay environmental noise of differing characters, and see what the effect is on listener comprehension. If Bella Coola turns out to be harder to hear than Cantonese, Bella Coola speakers can adjust by talking louder, so there's no insurmountable problem regarding noise.

Apart from cultural norms regarding how one should speak, a language like Hawaiian is probably one of the most robust against noise because it has only open syllables and a rather restricted phonemic inventory, so less chance of noise overriding the signal. A language with significant consonant clusters and lots of consonant types would be most at risk: candidates include Berber, Kartvelian and Salishan languages, Polish and Mongolian. In the case of Berber, some of the languages have epenthetic vowels and others don't, where the vowels can make the individual consonants more audible. Another possibility would be languages like Vietnamese and Hmong, which rely heavily on phonatory differences on vowels as part of their tonal system (vowel phonation contrasts are not very robust w.r.t. noise).

  • Thanks for the answer. Very interesting. As I said I am not from the field so don't know those languages you talk about. But just thinking in the languages I know, I'd say French is more sensible to noise compared to Spanish or Italian. Is it not true that they have more vowels [kind of more "density" in the same range of sounds]? I always thought this can make French language a bit more compact at the price of being forced to define better the sounds. But maybe it's only a perception. Not an expert :D – user2042652 Jan 20 '18 at 18:46
  • In fact, the same concept could be applied to written languages. Kanjis are more compact than alphabet, they say more in less space. The price is: 1, they need more visual resolution and 2) if they lose a grapheme they lose more information. – user2042652 Jan 20 '18 at 18:49
  • Yes, French has more vowel distinctions so it is in principle more vulnerable. – user6726 Jan 20 '18 at 18:51
  • 2
    This experiment would be just pseudo-science, to be honest. It's mostly just a function of practice, nothing to do with a language per se. – Adam Bittlingmayer Jan 20 '18 at 19:11
  • 1
    I don't understand why this experiment should be pseudoscience. Two languages being spoken. Random noise at increasing level over both. Maybe one of the languages fails to communicate before the other one. Why should this be wrong? Are the phonems equally important in all languages? Isn't there a different degree of redundancy in different languages? [Again, I'm just asking, not an expert] – user2042652 Jan 20 '18 at 21:34
1

In the mathematical theory of communication, which is an enterprise of engineering mostly electronic, noise reduction, detection and correction of errors, robustness (to use your word) under conditions of noise, all these are very important, maybe even the most important issue in the human designing of engineering artifacts (codes) to facilitate machine communication.

But, for whatever reason, all that very important research, while discussed philosophically by intellectuals in an informal manner, never really had a discipline transfer to linguists. There is surely some research into confusion of phonemes and other phenomena within particular languages, but I don't know of much into comparing different languages.

This sounds similar to comparisons of language length: the size of a text that has been translated into a number of languages, and which is the shortest and which the longest. There may well be succinct and, likewise, verbose human languages but the experiment to do so will be difficult because of the bias and artifice involved in both human composition of narratives and in the translation process. Average document length is actually one possible technical metric for comparing coding mechanisms so this is not just an analogy but also one very narrow example of what you are looking for.

So to your first question, no, I don't think there is any established technical literature (or much of it) that experimentally compares spoken language and content transfer under noisy circumstances. I could easily be wrong; it's just that linguistics, despite being a natural science, is not as interested in experimental methods as say psychology.

To your second question, very informally I would guess that it all depends on under what context you're comparing robustness. If you are comparing yelling but/sell commands in a noisy stock market environment, one could experimentally determine which language is more accurate, but still allow that there may actually be either 1) No significant difference, or 2) a statistically significant difference that in the long run has little substantial effect.

I think there are too many social variables to account for before saying that one human language is more accurate under noise than another.

-1

It's been said that the Piraha language could be understood even if it's just whistled in the air. If that's true, it's very robust to noise.

  • 1
    There are many auxiliary whistle languages that have developed for communication over long distances, especially in mountainous regions. They often use the tones of local tone languages, but not always. Pirahã does have tones but does not have a whistle language. – jlawler Feb 19 '18 at 16:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.