Ecological influences on phoneme inventories?

[alice]

I don't know if this counts as quackery or not; if it isn't, it's something to consider in your conlanging.

Maddieson and Coupe’s[9] study on 633 languages worldwide observed that some of the variation in the sonority of speech sounds in languages can be accounted for by differences in climate.

See under "Ecological patterns in sonority". I was under the impression that this was nonsense and has been debunked, but it's in Wikipedia, so it must be true.

[Creyeditor]

I don't think many linguists believe that there is a causal relationship. You can find some strange correlations if you compare the right variables.

[Ares Land]

An interesting test would be to see if any correlation can be found between climate and dialects of widely-spoken languages, such as Spanish or English.

Such as, is Spanish spoken in Patagonia less sonorous than in Colombia? How about Canadian English versus Southern American English?

[Rounin Ryuuji]

The most standard varieties of British and Australian English are quite similar; the Island of Great Britain itself seems to contain some of the most widely-varied English. The English where I live is also not similar to the Japanese of Honshuu, though the two have similar climates, if I understand right.

[Travis B.]

I'm not even going to entertain this one...

[Rounin Ryuuji]

I'm quite sure it's merely a protracted joke.

[Kuchigakatai]

I think it's quite possible, but I don't know if it has been studied seriously. Ares Land's experiment would be interesting to carry out, even if the results were negative. Does altitude affect how people speak in high La Paz (12000 feet above sea level) and Bogotá (8700 feet) versus coastal Santiago and Caracas?

Although, in Spanish, there's the problem highland vs. lowland dialects are indeed quite distinct as they are, partly due to higher later amounts of migration from Spain happening in the latter areas... Maybe a better experiment would be to take people from the coast to La Paz, without letting them speak to locals or listen to the local media (in order to avoid normal dialectal contamination), just letting them talk to themselves to see if their articulation changes.

[WeepingElf]

I am mildly sceptical about all this; there may be a tendency towards this, though. It wouldn't be the first time Wikipedia was wrong

[Nortaneous]

Even leaving aside the usual issues with statistical studies, this looks like it's dominated by macro-areal effects.

[xxx]

It's funny, I used to tease my co-workers native from the north of the country for their accent by telling them that they could open their mouths wider without the risk of seeing their tongues freeze and relax their jaws too accustomed to the muscular contractions due to the cold...

[bradrn]

Sorry for the slightly late reply, but I’d like to point out that there actually appear to be quite a few areas where ecology and phonology seem to be related:

1. Most strikingly, implosives are found almost exclusively in the tropics (no source, but just look at the WALS map)
2. Ejectives are most commonly found at higher elevations (Everett 2013)
3. Tone is most common in hot and humid areas (Everett & Blasi & Roberts 2015)
4. ‘Languages in Drier Climates Use Fewer Vowels’ (Everett 2017)

With this evidence, I’m not prepared to rule such correlations out entirely, though I’m still sceptical. (I also find it a bit suspicious that most of these studies come from the same person, though there could easily be other reasons for that.)

[Travis B.]

Is this the Everett we all know and love?

[bradrn]

Is this the Everett we all know and love?

Caleb Everett, not Daniel Everett.

[zompist]

I looked very briefly at one of these papers,enough to see that its database is WALS, with 527 languages.

This sort of this is likely to be garba--- ah wait, that's not very academic-- garbled, because languages are not randomly distributed. They are not randomly taken from the sea of possible languages; they are not independent bits of data. They are highly interrelated, they group up in families, they are highly affected by historical processes that are themselves dependent on geography, they are subject to areal effects, the best studied languages (i.e. WALS) are not representative of the whole 6000 or so in the world.

At the same time the sheer number of languages is so high that some wacky random hypotheses will end up "statistically significant". Ignoring the fact that statistical significance is defined for random independent events, which languages are not. "Statistically significant" doesn't mean that something is likely to be true; it's an evaluation that (say) you'd only expect that result in 5% of cases. It's often misused, because people forget that that means of 100 wacky random guesses, 5 will meet that threshhold by random chance.

The really interesting study would be of Everett's recycle bin. How many wacky hypotheses does he throw out compared to those that he gets into print?

[bradrn]

I looked very briefly at one of these papers,enough to see that its database is WALS, with 527 languages.

This sort of this is likely to be garba--- ah wait, that's not very academic-- garbled, because languages are not randomly distributed. They are not randomly taken from the sea of possible languages; they are not independent bits of data. They are highly interrelated, they group up in families, they are highly affected by historical processes that are themselves dependent on geography, they are subject to areal effects, the best studied languages (i.e. WALS) are not representative of the whole 6000 or so in the world.

At the same time the sheer number of languages is so high that some wacky random hypotheses will end up "statistically significant". Ignoring the fact that statistical significance is defined for random independent events, which languages are not. "Statistically significant" doesn't mean that something is likely to be true; it's an evaluation that (say) you'd only expect that result in 5% of cases. It's often misused, because people forget that that means of 100 wacky random guesses, 5 will meet that threshhold by random chance.

The really interesting study would be of Everett's recycle bin. How many wacky hypotheses does he throw out compared to those that he gets into print?

I can’t say that I’ve looked at these papers to any great extent either, but I don’t think it’s quite as bad as you make it out to be: both Everett 2017 and 2013 have p < 0.001. (The 2015 paper has p = 0.02 and p = 0.07, which is indeed more suspicious — and not even ‘significant’.) But I do agree that this isn’t an independent sample (even though WALS provides one), and it would be nice to see his ‘recycle bin’).

[Nortaneous]

Hard to tell between ecological influence and feature diffusion macroareas. Most such macroareas are equatorial. Definitions are hard, but regardless of how you slice it, there aren't many in the frozen north - Finnish looks a lot like Inuktitut, and there are probably grammatical similarities. Unusually large case systems? Dunno.

But if ecological influence suggests that certain things hold in cold temperatures, the languages of the southern tip of South America should look, if not Hyperborean, then at least a little Borean. Shouldn't they? I don't think they do. Phonologically, Selk'nam and Kawesqar look much more Californian than Arctic. Yahgan looks a little like Chukchi, I guess, but Chukotko-Kamchatkan is clearly on the periphery of the West Coast macroarea anyway - Itelmen more so than Chukchi (ejectives), but the vowel harmony is similar to what must be assumed for an earlier stage of Sahaptian.

And if Yahgan looks like Chukchi, well, they're both peripheral, so maybe they inherited the effects of older diffusion processes and never had time to get the new ones. Like how a fortis/lenis contrast in resonants used to be common across Western Europe but today is only preserved in a few Gaelic dialects.

[bradrn]

Hard to tell between ecological influence and feature diffusion macroareas. Most such macroareas are equatorial. Definitions are hard, but regardless of how you slice it, there aren't many in the frozen north - Finnish looks a lot like Inuktitut, and there are probably grammatical similarities. Unusually large case systems? Dunno.

But if ecological influence suggests that certain things hold in cold temperatures, the languages of the southern tip of South America should look, if not Hyperborean, then at least a little Borean. Shouldn't they? I don't think they do. Phonologically, Selk'nam and Kawesqar look much more Californian than Arctic. Yahgan looks a little like Chukchi, I guess, but Chukotko-Kamchatkan is clearly on the periphery of the West Coast macroarea anyway - Itelmen more so than Chukchi (ejectives), but the vowel harmony is similar to what must be assumed for an earlier stage of Sahaptian.

And if Yahgan looks like Chukchi, well, they're both peripheral, so maybe they inherited the effects of older diffusion processes and never had time to get the new ones.

I don’t think anyone’s suggesting that two languages in similar climates should look identical — just that they should be similar in some ways. For instance, those correlations I mentioned above predict that polar languages should have no implosives, no tone, large(ish) vowel systems and possibly ejectives. Does that sound about right for the South American languages? I don’t know enough about Borean and Notean languages to assess this properly.

Like how a fortis/lenis contrast in resonants used to be common across Western Europe but today is only preserved in a few Gaelic dialects.

I’ve never heard of this before; do you have any more info? I’m surprised to hear that voiceless resonants were once common in the area.

[Nortaneous]

I don’t think anyone’s suggesting that two languages in similar climates should look identical — just that they should be similar in some ways. For instance, those correlations I mentioned above predict that polar languages should have no implosives, no tone, large(ish) vowel systems and possibly ejectives. Does that sound about right for the South American languages? I don’t know enough about Borean and Notean languages to assess this properly.

No implosives, probably no tone, 3-7 vowels depending on language, ejectives in some languages but not others. But you could predict most of that from which side of the Atlantic they're spoken on.

I’ve never heard of this before; do you have any more info? I’m surprised to hear that voiceless resonants were once common in the area.

Voice doesn't have anything to do with it - that's fortis as in long

[alice]

It would be more useful to be able to *predict* a phoneme inventory from its ecology. One can imagine, for example, a language spoken in Antarctica to make much use of the biabial trill.

[Richard W]

It would be more useful to be able to *predict* a phoneme inventory from its ecology. One can imagine, for example, a language spoken in Antarctica to make much use of the biabial trill.

And indeed, all the indigenous languages of Antarctica do use it!