Small vowel inventories in North America

[Frislander]

Has anyone notice that despite the relatively small vowel inventories in the languages of North America, there is still a large amount of unusual inventories? I'll provide a few here to show you what I mean.

First off, the typical four-vowel system of most Algonquian languages or Navajo (admittedly it's not that unusual relatively speaking, but the lack of u is particularly noticeable)

Code: Select all

i
e   o
  a

Then there's the more unusual inventory of both Cheyenne and Quileute lacking all high vowels

Code: Select all

e   o
  a

Then there's Blackfoot and the Muskogean languages which drop e instead

Code: Select all

i
    o
  a

There's Arapaho, which lacks low vowels and where u is in nearly complementary distribution with i

Code: Select all

ɪ   ʊ
ɛ   ɔ

Then there's Wichita's weird system where o may or may not be present, leaving a highly atypical three-vowel vertical system (note that the system is unusual not so much for the fact that it is vertical but for the actual qualities of the vowels)

Code: Select all

ɪ ~ i ~ e
ɛ ~ æ      (o)
          ɒ ~ a

Even weirder is the typical Uto-Aztecan system. This seems to be extremely stable, being reflected in a large number of the daughter languages.

Code: Select all

i  ɨ  u
      o
   a

Then there's Hopi.

Code: Select all

i  ɨ
 ø  o
ɛ  a

There's also the thing in the Pacific-Northwest where many of the sonorants have vocalic allophones, thus potentially leaving https://en.wikipedia.org/wiki/Nuxalk_la ... la%20Coola with a single vowel a by analysing i and u as the vocalic allophones of the glides.

Is this a specifically North American thing or are small inventories often unbalanced in this manner?

[Frislander]

Silly me, I forgot some of the athabaskan inventories with more back than front vowels.

For instance Koyukon

Code: Select all

Full       Reduced
i    u         ʊ
            ə
            ɞ
æ    ɔ

Or Sarcee

Code: Select all

i    u
  a  ɒ

Note though that these systems are not necessarily typical of the family, with plenty of others having a simple five-vowel plus/minus schwa.

[Zaarin]

Note that most NA vowel systems described as /i e a o/ could just as easily be described as /i e a u/, since [o] and are generally allophonic in such languages.

[Chengjiang]

Honestly, I've always thought the linguistic "universal" that languages distinguish more front than back vowel qualities was pretty dubious, even as a tendency rather than an actual universal. There just seem to be too many exceptions.

[Frislander]

Note that most NA vowel systems described as /i e a o/ could just as easily be described as /i e a u/, since [o] and are generally allophonic in such languages.

Yeah, Ojibwe seems to be like that, while definite /i e a u/ cases include East Cree (coupled with a length distinction which may verge on quality), Pawnee and the Masset and Kaigani Haida. It's also sort of the case in Tlingit, though there's some quality weirdness going on with its length distinction.

Other systems with more back than front vowels include Valley Yokuts, and the Salishan systems with no rounded vowels also include Tilamook and (for the most part) Halkomelem.

[----]

/e a o/ is a very misleading vowel inventory for Cheyenne, since the vowel spelled <e> is almost never pronounced [e] in any context. It's usually [ɪ].

This is also the issue with languages that have the /i e a o/ inventory. In most cases the 'e' vowel is significantly lower than [e] and the 'o' vowel is significantly higher than [o]. So it's like a typical square vowel system except sort of 'tilted'.

[Sumelic]

Well, from what I understand canonical IPA [e] is actually quite close to [ɪ]. But as you say, the phoneme transcribed /e/ in most languages is usually a fair bit lower, often [e̞] or even [ɛ].

[zompist]

Well, from what I understand canonical IPA [e] is actually quite close to [ɪ]. But as you say, the phoneme transcribed /e/ in most languages is usually a fair bit lower, often [e̞] or even [ɛ].

Er, what? [ɪ] is lax; the cardinal vowels are supposed to be as tensed as possible.

When small vowel inventories are reported, I think it's important to go back to language-specific sources before drawing too many conclusions. It's usually convenient to just use the simple Roman letters rather than overprecise IPA values. (E.g. Southern Quechua has /a i u/, but none of them have the IPA values.) If there's only one mid front vowel, there's nothing wrong with calling it /e/.

[Frislander]

My question is probably better phrased as: are theses mall, atypical inventories typical only of North America or are they more often found elsewhere? How many languages outside the North-American continent have /i e a o/, for instance?

[Pole, the]

Er, what? [ɪ] is lax; the cardinal vowels are supposed to be as tensed as possible.

But this kind of an opposition is English-specific, isn't it?

[Sumelic]

Er, what? [ɪ] is lax; the cardinal vowels are supposed to be as tensed as possible.

"Lax" is a relative term, and it's often used phonologically rather than phonetically anyways. Cardinal [e] is described as similar to [ɪ] by Geoff Lindsey here: The British English vowel system

[zompist]

Er, what? [ɪ] is lax; the cardinal vowels are supposed to be as tensed as possible.

But this kind of an opposition is English-specific, isn't it?

Not at all-- you can hardly understand Latin or Greek vowels without it.

Lax vowels can easily be defined either in articulatory or acoustic terms-- see Labov's books for instance (where they are defined acoustically on a precise F1/F2 chart).

I'll admit that Jones's [e] sounds a lot like [ɪ]! But the Wells/Meier recording is much more what I'd expect for [e/ɪ]; it's odd that Lindsey says he can't tell them apart.

[gach]

I'm assuming the tense/lax contrast is used to mean something more than simply a distinction in length and height. Since we don't have native speakers of Ancient Greek or Latin around any more, we can't do close studies of their pronunciation. What's the reason then to assume that some other parameter connected to tenseness, like centralisation, was also vital for understanding their vowel systems? From the top of my head, I can't remember reading of any, though I haven't studies the language that closely either.

[zompist]

Try reading Allen's Vox Graeca and Vox Latina for what we know about Latin/Greek vowels and how we know it. (Hint: it does not depend on tape recordings.)

The importance of laxness in sound changes in various languages is described in Labov's Principles of Linguistic Change: Internal Factors.

[gach]

I've skimmed bits of Vox Graeca now and then but a condensed account of the evidence by someone well informed would be nice. I'm not that interested on phonetics myself that I'd realistically take time to read a whole pile of books just to answer a single question like this.

If you know of page numbers to sections in Vox Graeca or Vox Latina that specifically deal with tenseness, those would be nice as well.

[zompist]

OK, see Vox Latina p. 47. The diagram is the most informative, but I'll quote the text:

"As is commonly the case the long-vowel subsystem occupies a larger periphery of articulation than the short, the short vowels being in general more laxly articulated and so less far removed than the corresponding long vowels from the 'neutral' position of the speech organs."

The corresponding Greek diagram is on p. 62 (3rd ed.), though from the text it's clear that laxness is far less important in Greek.

As for Labov, he resists summary, though the general argument is "Certain vowel shifts produce paradoxes as usually described, because contrary movements would produce mergers. These mergers are avoided if we distinguish tense and lax vowels, and peripheral/nonperipheral movement tracks for vowel shifts."

[Vijay]

I'm assuming the tense/lax contrast is used to mean something more than simply a distinction in length and height.

Advanced tongue root?

[gach]

@zompist: Thanks.

Advanced tongue root?

I imagine it might sometimes be reasonable to talk about those as well under the labels of tenseness.

[cromulant]

My question is probably better phrased as: are theses mall, atypical inventories typical only of North America or are they more often found elsewhere? How many languages outside the North-American continent have /i e a o/, for instance?

Those aren't all small. WALS defines a "small" inventory as one with under five vowels, which is reasonable as five is the most typical inventory size. But none of them are large either. The defining feature seems to be that they are 'skewed,' i.e. distributed in a suboptimal pattern in which the vowels are not maximally distinct.

South America has a few of these as well: Hixkaryana /e æ ɯ u ɔ/, Waorani /i e æ ɵ~o~ɤ a/ + nas., Piraha /i a o/, Lakono Arawak /i e a e ɨ u/ IIRC. Of these, only Piraha is "small."

EDIT: Forgot Wari, with the world's most skewed vowel inventory: /i y e ø a o/. Six vowels, only one of them back!

I don't think these unusual systems are "typical" of the Americas, but yeah, they do seem to be much better represented there than elsewhere. Outside of the Americas, there are of course the vertical two-vowel systems of some NE Caucasian languages (also found, perhaps, in Iatmul, though this is contentious), and the four-vowel vertical system of Marshallese. /a e i u/ is found in Malagasy and was found in Etruscan, and possibly Sumerian. /a i u/ is the prototypical vowel inventory of Australian languages, but this is as normal and skewless as a three-vowel inventory can be.

The most skewed, non-American, non-large (i.e. below 7) vowel systems I know of are Nivkh /ɪ ɪe æ u ɤ o/ and Big Nambas /i e ə a u/.

[Curlyjimsam]

I've skimmed bits of Vox Graeca now and then but a condensed account of the evidence by someone well informed would be nice. I'm not that interested on phonetics myself that I'd realistically take time to read a whole pile of books just to answer a single question like this.

It's only two books (and reading just one of them would probably give you the answer you want), and both are pretty short.

[gach]

It's only two books (and reading just one of them would probably give you the answer you want), and both are pretty short.

Yeah I know, but I already fill my days with technical literature, much of it from my own field. Linguistics is just a hobby for me.

[----]

The most skewed, non-American, non-large (i.e. below 7) vowel systems I know of are Nivkh /ɪ ɪe æ u ɤ o/ and Big Nambas /i e ə a u/.

Supposedly that Nivkh vowel inventory is false and it's a little bit of a mystery where it even came from. The 'right' inventory for the language is fairly banal, but I can't remember what it is.

[Frislander]

Those aren't all small. WALS defines a "small" inventory as one with under five vowels, which is reasonable as five is the most typical inventory size. But none of them are large either. The defining feature seems to be that they are 'skewed,' i.e. distributed in a suboptimal pattern in which the vowels are not maximally distinct.

South America has a few of these as well: Hixkaryana /e æ ɯ u ɔ/, Waorani /i e æ ɵ~o~ɤ a/ + nas., Piraha /i a o/, Lakono Arawak /i e a e ɨ u/ IIRC. Of these, only Piraha is "small."

EDIT: Forgot Wari, with the world's most skewed vowel inventory: /i y e ø a o/. Six vowels, only one of them back!

I don't think these unusual systems are "typical" of the Americas, but yeah, they do seem to be much better represented there than elsewhere. Outside of the Americas, there are of course the vertical two-vowel systems of some NE Caucasian languages (also found, perhaps, in Iatmul, though this is contentious), and the four-vowel vertical system of Marshallese. /a e i u/ is found in Malagasy and was found in Etruscan, and possibly Sumerian. /a i u/ is the prototypical vowel inventory of Australian languages, but this is as normal and skewless as a three-vowel inventory can be.

The most skewed, non-American, non-large (i.e. below 7) vowel systems I know of are Nivkh /ɪ ɪe æ u ɤ o/ and Big Nambas /i e ə a u/.

From what I've read , Big Nambas seems to be unusual in having a length contrast only on /a/ and /u/.

I do agree with what you're saying, though.

[vokzhen]

The most skewed, non-American, non-large (i.e. below 7) vowel systems I know of are Nivkh /ɪ ɪe æ u ɤ o/ and Big Nambas /i e ə a u/.

Supposedly that Nivkh vowel inventory is false and it's a little bit of a mystery where it even came from. The 'right' inventory for the language is fairly banal, but I can't remember what it is.

The two grammars I have use /i ɨ u e a o/ and /i ə u e a o/. Which *is* pretty close to the inventory Wikipedia gives if the latter were in more standard notation, but neither of my grammars gives a detailed description of the vowels.

[CatDoom]

Other systems with more back than front vowels include Valley Yokuts

I'm pretty sure /a/ in Valley Yokuts is a front vowel, though it would naturally be more central than /i/, so the system is actually very symmetrical. Considering that /uː/ is not usually phonetically distinguished from /ɔː/, you could even make the argument that the language has more front vowels than back.

Speaking of Californian languages, Hupa has a vowel system that's basically the same as what Thetha describes for Cheyenne. There are also some languages in the south with /a i u/ (Salinan and the Yuman-Cochimí family), and some in the north with /i e a o~u/ (Klamath, Shastan, and most California Athabaskan languages). Most Californian languages have /i e a o u/ (or, in the case of Yuki, /i ə a o u/), however, and a fair number have larger inventories.