Yonagu by Nancy - vowel systems

[alice]

http://gesc19764.pwp.blueyonder.co.uk/v ... stems.html

Here's a start. Let me know if anything's obviously wrong, or if you have any other vowel systems which might be appropriate for inclusion.

[finlay]

I don't like the vowel diagram that comes from Wikipedia. It's too... busy.

I want to point out that the diagrams you have are idealised, anyway... in real situations vowels tend to cover an area of the space and aren't necessarily equally spaced and all that. I'm also always a bit disgruntled by the disclusion of diphthongs - for english in particular this makes it look like there's only one mid front vowel. However, I do know that you're more displaying possibilities rather than making a list by languages, so...

Also... "by Nancy"?

[Jipí]

You took half of German's vowels away. Why? The standard language's got [iː ɪ yː ʏ eː (ɛː) ɛ øː œ ə ɐ aː a ɔ oː ʊ uː], that's 17 (or 16 if you merge [ɛː] with [eː]), not 8. Even if you merge the 'long' and 'short' varieties in order to be able to classify it basically, I think you still should include [ɐ] even though it's an allophone of /r/. Besides, I'm curious whether [ɐ] and [a] will merge within my lifetime in the same way [ɛː] merges into [eː] for many people now, myself included.

[roninbodhisattva]

I want to point out that the diagrams you have are idealised, anyway... in real situations vowels tend to cover an area of the space and aren't necessarily equally spaced and all that.

This is of course really true about any kind of discussion of phonology, general or of a particular language...of course there is variation and such. You just have to talk about idealized systems to talk about systems at all. Also, I don't know if yours was really meant as a criticism, so if mine seems hostile, I didn't mean it to be. You just can't talk about all the mess all the time.

[Radius Solis]

Here's a start. Let me know if anything's obviously wrong, or if you have any other vowel systems which might be appropriate for inclusion.

There's another interesting 5-vowel system you've missed, which can be found in some Uto-Aztecan languages; it is also reconstructed for proto-Uto-Aztecan: /i ɨ u a ɔ/. The system is cool because there are essentially only two heights distinguished, with two degrees of front/back on the lower tier and three degrees of it on the upper tier. For a couple examples, see Shoshone and Tohono O'odham. (The latter is listed with an additional schwa on wikipedia, but this is not listed as an independent phoneme in the dictionary and grammar we have.)

The basic system is shared by Tübatulabal and Kawaiisu but with the addition of an /ɛ/, forming a six-vowel system you also don't list: still only two degrees of height, but three degrees of front/back for each tier. Many other Uto-Aztecan languages have variations on the same theme, but as far as I'm aware, it does not exist outside of this family.

[Radius Solis]

Oh also, there is another two-vowel language you don't mention, Arrernte, in Australia. Like Ubykh, its two vowels are /ə a/, but unlike it, there is almost no conditioned allophony, but rather a lot of free variation - /ə/ can be realized as any highish vowel and /a/ as any lowish vowel, regardless of context.

[Ulrike Meinhof]

I only noticed one error. On the very last line, it says "S14C is my idiolect of English" where what is obviously meant is "S14C is English as the LORD intended".

[Magb]

Your claim that /a i u/ is the extent of the short vowels of Old Norse and Icelandic is perplexing to say the least. Maybe you were thinking of Proto-Norse? But even that had at least short /e/ in addition to those.

[AnTeallach]

I agree that the description of German is over-simplified. Languages like German (and many other Germanic varieties, including many forms of English) are probably best analysed as having sub-systems of long and short vowels.

Other comments:
- On the first browser I saw this on (a version of IE), the IPA symbols in the text were messed up. I'm now on Firefox and they're fine on that, but I'd check them.
- In the "S4" system, it might be worth explaining why you've got /a/ and /e/ so far from their positions on the IPA chart. I'm not saying you're wrong to have them there, just that it could be confusing to the uninitiated.
- Where are you getting those analyses of English dialects from? Here's Rhondda English, which only seems to match your "Welsh English" if you ignore length and indeed the tense/lax distinction for the high vowels, but then for Australian and Irish English you seem to be treating the lax vowels separately and assigning different qualities to some pairs where the distinction is mainly length.
- Scottish Gaelic (/i e ɛ a o ɔ u ɤ ɯ/ plus length) is quite an interesting system (variant of S9).
- For some more "big" Germanic systems, there's the 13 long vowels of one variety of Austrian German (five heights, with back, front rounded and front unrounded distinguished at four of them), and the similar long vowel system of Weert Dutch.

[alice]

Plenty of food for thought here. Let's see.

You took half of German's vowels away. Why? The standard language's got [iː ɪ yː ʏ eː (ɛː) ɛ øː œ ə ɐ aː a ɔ oː ʊ uː], that's 17 (or 16 if you merge [ɛː] with [eː]), not 8. Even if you merge the 'long' and 'short' varieties in order to be able to classify it basically, I think you still should include [ɐ] even though it's an allophone of /r/. Besides, I'm curious whether [ɐ] and [a] will merge within my lifetime in the same way [ɛː] merges into [eː] for many people now, myself included.

I did in fact wrestle a bit with my conscience about this. Perhaps I could present German and Swedish this way as part of a "Very Big Vowel Systems" section, although in the context of the page I'm not sure how useful it would be.

There's another interesting 5-vowel system you've missed, which can be found in some Uto-Aztecan languages; it is also reconstructed for proto-Uto-Aztecan: /i ɨ u a ɔ/. The system is cool because there are essentially only two heights distinguished, with two degrees of front/back on the lower tier and three degrees of it on the upper tier.

The basic system is shared by Tübatulabal and Kawaiisu but with the addition of an /ɛ/, forming a six-vowel system you also don't list

I'll put those in; they're certainly interesting enough.

Oh also, there is another two-vowel language you don't mention, Arrernte, in Australia. Like Ubykh, its two vowels are /ə a/, but unlike it, there is almost no conditioned allophony, but rather a lot of free variation - /ə/ can be realized as any highish vowel and /a/ as any lowish vowel, regardless of context.

This too; I think the stuff I wrote about two-vowel systems might be a bit dogmatic

I only noticed one error. On the very last line, it says "S14C is my idiolect of English" where what is obviously meant is "S14C is English as the LORD intended".

Ha ha ha ha ha! Thank you; I've just won a bet as a result of you saying that.

Your claim that /a i u/ is the extent of the short vowels of Old Norse and Icelandic is perplexing to say the least. Maybe you were thinking of Proto-Norse? But even that had at least short /e/ in addition to those.

It comes from the chapter in Routledge's The Germanic Languages. I'll double-check it against another source.

I agree that the description of German is over-simplified. Languages like German (and many other Germanic varieties, including many forms of English) are probably best analysed as having sub-systems of long and short vowels.

Therein lieth the problem: in the context of the page, is it better to ignore the differences in quality and treat the two systems as essentially the same, or ignore the differences in quantity and show all the vowel qualities?

- On the first browser I saw this on (a version of IE), the IPA symbols in the text were messed up. I'm now on Firefox and they're fine on that, but I'd check them.

The encoding is specified in the file as utf-8, so I don't know what IE thinks it's doing.

- In the "S4" system, it might be worth explaining why you've got /a/ and /e/ so far from their positions on the IPA chart. I'm not saying you're wrong to have them there, just that it could be confusing to the uninitiated.

Yes.

- Where are you getting those analyses of English dialects from?

Wells's Accents of English. You've found a mistake: that should be the *short vowels* of Welsh English.

- Scottish Gaelic (/i e ɛ a o ɔ u ɤ ɯ/ plus length) is quite an interesting system (variant of S9).

Yes.

- For some more "big" Germanic systems, there's the 13 long vowels of one variety of Austrian German (five heights, with back, front rounded and front unrounded distinguished at four of them), and the similar long vowel system of Weert Dutch.

Just how many Big Germanic Vowel Systems can fit on one page, anyway?

[Travis B.]

Just to make your job even more interesting, just about every Germanic dialect of any sort has its own Big Germanic Vowel System. Hell, even, say, the typical progressive Inland North dialect, for instance, has a different sort of BGVS than, say, conservative General American, despite the two not being all too far apart in the bigger scheme of things...

[AnTeallach]

I agree that the description of German is over-simplified. Languages like German (and many other Germanic varieties, including many forms of English) are probably best analysed as having sub-systems of long and short vowels.

Therein lieth the problem: in the context of the page, is it better to ignore the differences in quality and treat the two systems as essentially the same, or ignore the differences in quantity and show all the vowel qualities?

For Standard German and also for Australian and Irish English, I would take either the long vowels or the short vowels (whichever is more interesting I suppose) and say that's what you're doing, as you've already done for some of the other examples. That then probably makes those English dialects less interesting anyway, so one of them could be taken out, leaving more room for the really big Germanic long vowel systems.

- Where are you getting those analyses of English dialects from?

Wells's Accents of English. You've found a mistake: that should be the *short vowels* of Welsh English.

Wells presumably shows the Australian and Irish English vowels as having length contrasts, though?

[alice]

OK; I've made some changes. I'm guilty of fudging the issue in some cases by disregarding vowel length, but I don't think that should be too much of a problem.

[Nortaneous]

Ubykh supposedly has a third vowel /a:/, which is supposedly somehow qualitatively different than /a/. There's some NWC lang out there with just two, but damned if I can remember which one it is.

I think I sent you a link to a Wikipedia page about a lang with /a o i 1 u/ a while back; iirc that's a common feature of some family, which it inherited frmo the protolang, and I've never seen it anywhere else, so...

There are some langs with a variation on T6 where the extra vowel is /2/.

"The similar and unusual system T7a is found in Occitian, and derives from T7 in which /y/ fronted to /y/ and the other back vowels were raised." - Shouldn't the first /y/ be /u/?

T8Ra: I think what you have as /I/ is normally called /1/.

[Mecislau]

If you want to talk about Caucasian two-vowel systems, it might be better to reference languages like Abkhaz rather than Ubykh instead (or in addition to). Abkhaz is very closely related to Ubykh and also has only two vowels, yet it has well over 100,000 speakers to Ubykh's 0, and correspondingly has a lot more information available about it.

[Aurora Rossa]

I've always heard that Ubykh (and as Mecislau said, Abkhaz) has only two vowels, but I've also heard that those vowels sound like an ordinary full vowel system because of allophony with the palatalized and labialized consonants. In that case, why do we even describe it as having only two vowels?

Interesting and helpful webpage, by the way, although I recall seeing it somewhere before.

[zompist]

Length is kind of the big lurking monster in the discussion. I'd suggest making it clearer up front that what you're really plotting is idealized locations within the vowel space.

Perhaps you could have a section on length where contrasting systems are shown with and without length— it'd be interesting to see German and Latin contrasted that way. (It might be interesting also to have a couple analyses of French... if you put long and nasal vowels in, French starts to look positively Germanic in complexity.

Actually your treatment of Latin is inconsistent. Do you have (access to) Vox Latina? In terms of vowel space alone, it's really a nine-vowel system (all the short vowels but a are laxer).

Classical Greek is also a bit complicated. I can scan Allen's charts for you if needed.

You should be a bit more careful with specifying varieties— the Chinese you cite is presumably Mandarin, but I don't know if you're doing Brazilian or European Portuguese. It'd also be better to say Southern Quechua.

It's kind of you to cite one of my languages; I'd suggest also Elkarîl, which has a nine-vowel system with no rounded vowels at all.

[Mecislau]

I've always heard that Ubykh (and as Mecislau said, Abkhaz) has only two vowels, but I've also heard that those vowels sound like an ordinary full vowel system because of allophony with the palatalized and labialized consonants. In that case, why do we even describe it as having only two vowels?

Because what you just wrote there is pretty much the very definition of allophony?

[makvas]

Interesting to know that T5a is the system I'm using for my proto-lang, Kemacetan.

I don't know if it's relevant but my dialect of English has only the six vowels /ɑ ɛ ɪ ʊ̜ æ ʌ/ under my analysis, and is diphthong-heavy with /ɑj ɛj ɪj ʊ̜w æw ɔj ʌw/. It has no rounded vowels, just diphthongs with /w/. I guess I'm not a professional linguist so that's probably not appropriate for your survey anyways

[Nortaneous]

Going off zompist's mention of Elkarîl, here are two natlang systems with no rounded vowels:
Saanich - /ɑ e ə i/
Matsés - /a e ɤ i ɨ ɯ/

And here's another batshit insane vowel system, this time not in Europe: http://en.wikipedia.org/wiki/Iaai_language (edit: oh wait Estonian has pretty much the same thing)

[Aurora Rossa]

And here's another batshit insane vowel system, this time not in Europe: http://en.wikipedia.org/wiki/Iaai_language (edit: oh wait Estonian has pretty much the same thing)

Heh, I had a conlang with that vowel system (although I think it had a couple more thrown in there) but I discarded it as unrealistic.

[Nortaneous]

Yeah, large vowel systems are a bitch to pull off properly. I generally avoid them (hell, my most developed lang has two vowels), although Arve is an obvious counterexample with /a ɛ ɞ ʌ ɔ e ø i y ʊ u/ + 10 diphthongs.

Also, WALS claims Yimas can be analyzed as having only two vowels, but everything I've found in ~2min of searching says it has four: /a i ɨ u/

[Bristel]

Going off zompist's mention of Elkarîl, here are two natlang systems with no rounded vowels:
Saanich - /ɑ e ə i/
Matsés - /a e ɤ i ɨ ɯ/

And here's another batshit insane vowel system, this time not in Europe: http://en.wikipedia.org/wiki/Iaai_language (edit: oh wait Estonian has pretty much the same thing)

Basically my Athanic conlang has a similar vowel system to Estonian's... only the /æ/ is an /a/, or something like that.

[ná'oolkiłí]

I've only ever seen two- or three-vowel systems in the analyses of Northwest Caucasian languages, and nothing to suggest that such analyses are too deep. The phonologies in The Indigenous Languages of the Caucasus, Vol 2: North West Caucasus give them thusly—

Abkhaz: /ə a (aː)/ The third is marginal, arising in native words synchronically words only in the sequences /ħa aħ aa/ and diachronically from */ʕa/ or */aʕ/ or deletion of intervocal */q'/. In loan words it occurs when medial consonants are lost.

Abaza: definitely just /ə a/, apparently

Kabardian: /ə a (aː)/ Many schwas are epenthetic, but others are definitely underlying. /a/ often reduces to a schwa, further complicating things. /aː/ can easily be explained away with /ah/ or /ha/ word-initially; the sequences /ɣa aɣ/ likewise can trigger this. Apparently many Kabardians in fact pronounce [aː] with a prosthetic [h], indicating this isn't an unfounded perspective. /o/ is a very marginal loan phoneme. If anyone's interested about the NWC phonetic vowels, Colarusso's Kabardian lists at least the following allophones: [ə ʌ ɪ ɛ a ɯ ɔ ʊ o]. They are conditioned by some very cool processes based on place of articulation of surrounding consonants.

Abzakh (a West Circassian dialect): /ə e a/ These specify for height only, not frontness as <e> may imply.

Ubykh: /ə a ạ/ I think this correspond pretty much to Abzakh's system, just with a different orthographic convention. These last two chapters are in French, so I unfortunately can't get too much information on Ubykh or Abzakh.

As for South Caucasian languages, IIRC both Laz and Svan have your SR8 and an additional /ə/ (and length distinction for all vowels). Georgian and Mingrelian aren't too interesting vowel-wise.

I haven't read much on North-East Caucasian languages, but they apparently also can have crazy vowel inventories.

[Nortaneous]

I haven't read much on North-East Caucasian languages, but they apparently also can have crazy vowel inventories.

Maybe not crazy except in the case of Chechen, but definitely interesting. Ingush has /{ A e @ o i u/ and nine diphthongs (including /ij uw/) and Lezgi has /{ a e i y u/.