A guide to small vowel inventories.

[Birdlang]

This is where we post small vowel inventories.
Arabic has /a i u/. Adyghe has /a a: @/.
Amuesha has /a e o/ plus /1/ as a marginal phoneme.

[KathTheDragon]

Hint: This isn't what L&L Museum's for. A thread like this belongs in L&L, and would be moved here if it became a very useful resource. Besides, this sort of thing has been done before: see this thread.

[Birdlang]

Hint: This isn't what L&L Museum's for. A thread like this belongs in L&L, and would be moved here if it became a very useful resource. Besides, this sort of thing has been done before: see this thread.

I know. I now want to see small vowel sets.

[KathTheDragon]

See the thread. It's very comprehensive.

[Dezinaa]

There seems to be a bit of a pattern here, but I can't put my finger on it.

Adzera: /ɑ i u/
Aleut: /a i u aː iː uː/
Aymara: /a i u/
Bella Coola (Nuxalk): /a i u/
Caddo: /a i u aː iː uː/
Diyari: /a i u/
Dyirbal: /a i u/
Garawa: /a i u/
Greenlandic (Kalaallisut): /a i u (aː iː uː)/
Haida (Skidegate): /a i u aː iː uː/
Hupa: /a ɪ u/
Kayardild: /a i u aː iː uː/
Kutenai: /a i u/
Nunggubuyu: /a i u aː iː uː/
Paumarí: /a i u/
Pitjantjatjara: /a i u aː iː uː/
Rama: /a i u aː iː uː/
Tashlhiyt: /a i u/
Yanyuwa: /a i u/
Yidiny: /a i u aː iː uː/

[Nortaneous]

There's a list of vowel systems on Anthologica, but Anthologica is down.

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/; in Moloko, there's an epenthetic schwa and such extreme vowel harmony that all features but low/high are analyzed as suprasegmentals.

Arrernte has /a @/; it used to have a rounded/unrounded contrast in its vowels (probably */a i u/), but it offloaded that onto its consonants. Some Chadic languages have the same system, but the high vowel there is usually called /1/. Proto-NWC probably had something like /3 1/; /a:/ in most NWC comes from absorption of radical consonants into the low vowel.

Then /a i u/, /a e o/ in Amuesha and Cheyenne, /a i o/ in Piraha, /a @ 1/ in Marshallese (it was /a 3 7 1/ but the mid vowels merged), /a: 3 1/ in NWC, /a e i/ in Wichita. Arapaho almost has /e o i/ but /i u/ aren't in *completely* complementary distribution.

/a @ i u/ in Rukai, /a e i u/ or /a e i o/ in various languages, /{ A i u/ in Seri, /a 3 7 1/ in older Marshallese.

Also, search UPSID, but take it with a grain of salt -- I'm pretty sure Tigre has more than two vowels.

[WeepingElf]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

This appears to have been the case in pre-ablaut PIE, too, though */a i u/ is a viable (and typologically less odd) alternative.

[Nortaneous]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

This appears to have been the case in pre-ablaut PIE, too, though */a i u/ is a viable (and typologically less odd) alternative.

Where'd /e o/ come from then?

[Hallow XIII]

There's a list of vowel systems on Anthologica, but Anthologica is down.

No longer!

[----]

/a e o/ in Amuesha and Cheyenne

I would prefer an analysis of Cheyenne vowels as /ɪ a o/, seeing as that matches the surface pronunciation more accurately. Only in vowel sequences /ɪɪ/ does it really approach [e] but those are rather rare.

[Pogostick Man]

/a @ i u/ in Rukai

That's also the reconstructed system in Proto-Eskimo-Aleut.

[Yaali Annar]

/a @ i u/ in Rukai

That's also the reconstructed system in Proto-Eskimo-Aleut.

That's also the reconstructed system in Proto-Austronesian

Some reconstruction of Proto Japanese is also claimed to have that vowels.

[Nortaneous]

Some reconstruction of Proto Japanese is also claimed to have that vowels.

Do they have diphthongs, or is it just coalescence of vowel clusters?

[WeepingElf]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

This appears to have been the case in pre-ablaut PIE, too, though */a i u/ is a viable (and typologically less odd) alternative.

Where'd /e o/ come from then?

Split of *a under the rules of ablaut. What exactly those rules are, is still unknown, though. That is one of the matters we are discussing in this thread.

[gach]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

This appears to have been the case in pre-ablaut PIE, too, though */a i u/ is a viable (and typologically less odd) alternative.

Though with such systems you have to ask yourself is there really any difference in the language between analysing /j w/ ~ or /i u/ ~ [j w]. If there isn't, you are able to choose either way, but including such results in a typological study will bend it towards analysing someone's personal tastes and not just the languages included.

Sometimes apparent extreme vowel systems don't stand closer inspection at all. The Ndu family is conventionally thought to include languages with extremely simple vertical vowel systems with frontness and rounding imposed allophonically by surrounding consonants. Iatmul in particular has been analysed as having /a ə ɨ/ or even /a: a/, taking the distinction between the lower vowels to be primarily of length and claiming that the high vowel is a fully predictable epenthetic vowel and thus qualifies as unphonemic. Later research just happens to totally disprove this picture requiring at least 11 distinct vowel phonemes at seven distinct qualities. Even in the old picture many words would seem to require phantom /j/s or /w/s to explain the surface vowels. Not counting the fact that these often contradict the currently known phonetic reality of the language, I can't understand to start with why would someone consider for example /ɨj/ a better analysis for than /i/.

[WeepingElf]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

This appears to have been the case in pre-ablaut PIE, too, though */a i u/ is a viable (and typologically less odd) alternative.

Though with such systems you have to ask yourself is there really any difference in the language between analysing /j w/ ~ or /i u/ ~ [j w]. If there isn't, you are able to choose either way, but including such results in a typological study will bend it towards analysing someone's personal tastes and not just the languages included.

Yep. Indeed, I prefer reconstructing pre-ablaut PIE with three vowels, */a i u/, with the restriction that */i/ and */u/ do not occur before resonants. That is IMHO better than positing */a ai au/, and corroborated by the Old European Hydronymy. But that discussion belongs to the Great Proto-Indo-European Thread.

[Tropylium]

in Moloko, there's an epenthetic schwa and such extreme vowel harmony that all features but low/high are analyzed as suprasegmentals.

This kind of a thing seems like a technicality, tbh. I've read an account of the closely related Muyang, and its "single vowel" struck me as more like a unit of morphophonology ("deep structure") than phonology ("surface structure"). And so a six-vowel inventory would seem like a more sensible level or analysis for the purposes of this topic: /a ɛ~e ɔ~o ɪ ʏ ʊ/.

BTW, has anyone run across a vowel system that distinguishes 3-4 qualities, no length, and at least one other feature (e.g. nasality, register?)

[Yaali Annar]

Some reconstruction of Proto Japanese is also claimed to have that vowels.

Do they have diphthongs, or is it just coalescence of vowel clusters?

It's reconstructed with diphothongs (ai, ui, ia, i forgot the others), yes.

[Birdlang]

There's a list of vowel systems on Anthologica, but Anthologica is down.

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/; in Moloko, there's an epenthetic schwa and such extreme vowel harmony that all features but low/high are analyzed as suprasegmentals.

Arrernte has /a @/; it used to have a rounded/unrounded contrast in its vowels (probably */a i u/), but it offloaded that onto its consonants. Some Chadic languages have the same system, but the high vowel there is usually called /1/. Proto-NWC probably had something like /3 1/; /a:/ in most NWC comes from absorption of radical consonants into the low vowel.

Then /a i u/, /a e o/ in Amuesha and Cheyenne, /a i o/ in Piraha, /a @ 1/ in Marshallese (it was /a 3 7 1/ but the mid vowels merged), /a: 3 1/ in NWC, /a e i/ in Wichita. Arapaho almost has /e o i/ but /i u/ aren't in *completely* complementary distribution.

/a @ i u/ in Rukai, /a e i u/ or /a e i o/ in various languages, /{ A i u/ in Seri, /a 3 7 1/ in older Marshallese.

Also, search UPSID, but take it with a grain of salt -- I'm pretty sure Tigre has more than two vowels.

What natlangs have /a e i u/? I was typing on an IPad so I could see the Unicode characters.

[vokzhen]

What natlangs have /a e i u/? I was typing on an IPad so I could see the Unicode characters.

A fairly large number of North American languages have either /a e i u/ or /a e i o/, and the difference between the two is often (though not always) a matter of transcription, with /u/ or /o/ varying on context somewhere between [u~ʊ~o]. Off the top of my head, Nahuatl's one of the big ones, usually transcribed with /o/ but varying. Pretty sure a few of the Salish and Caddoan languages have it too. Some Iroquoian languages have it in their oral vowels. Seri's got /a e i o/ but I think that's actual [o].

[Xephyr]

Akkadian. Probably Sumerian, though I understand there is some disagreement on that. Yup'ik has /a i u ə/, not sure if that counts.

[linguoboy]

What natlangs have /a e i u/? I was typing on an IPad so I could see the Unicode characters.

A fairly large number of North American languages have either /a e i u/ or /a e i o/, and the difference between the two is often (though not always) a matter of transcription, with /u/ or /o/ varying on context somewhere between [u~ʊ~o]. Off the top of my head, Nahuatl's one of the big ones, usually transcribed with /o/ but varying. Pretty sure a few of the Salish and Caddoan languages have it too. Some Iroquoian languages have it in their oral vowels. Seri's got /a e i o/ but I think that's actual [o].

You can add Omaha-Ponca, a Siouan language. It used to have a five-vowel system, like other Dhegiha languages, but /u/ was phonetically very fronted and eventually merged with /i/. Earlier /o/ probably once ranged from [ʊ] down as far as [ɔ] (as in contemporary Osage), but now I think is mostly .

[Pogostick Man]

Nuxalk and Moloko have one vowel, /a/. In Nuxalk, can be analyzed as /j= w=/

What's stopping them from analyzing [a] as a syllabic /h/?

[----]

Probably the fact that Nuxalk doesn't have /h/ except marginally, and that there is no reason to interpret it that way otherwise.

[Nortaneous]

Birdlang: http://anthologi.ca/?id=165580#%28%28T4F%29%29

Akkadian. Probably Sumerian, though I understand there is some disagreement on that. Yup'ik has /a i u ə/, not sure if that counts.

There's a lot of Austronesian languages with the Yup'ik system, and Old Japanese probably had it + some diphthongs.