More short vowels than long vowels?

[alice]

AFAIK, if a language has long and short vowels, there are never more sort vowels than long vowels. Are there any counterexamples?

[Soap]

I'm surprised, I thought it would be the other way around. Japanese is a counterexample if you consider what is commonly written ē ō to be diphthongs, and/or consider ā (which only occurs over historical morpheme boundaries) to be just a sequence of two short a's. Icelandic might work too, since most of its historical long vowels have changed into diphthongs, and likewise for other early stages of Germanic languages when we still had vowel length but it was beginnning to degrade.

[Hakaku]

I think a better example than Japanese would be Okinawan, since all mid vowels raised and merged with their high counterparts, while mid long vowels persist as they arise from other sources. So /e/ /i/, but /ai/, /ae/ /e:/. Although, due to some regional borrowing and exceptions, Okinawan still does possess short mid vowels, but they constitute less than 0.8% of the vocabulary from what I recall.

But anyway, depending on the analysis, Khmer is a likely candidate.

EDIT: I misinterpreted it as the opposite, wooops. Tariana is one example with 6 short vowels, 5 short nasal ones, and only 4 long vowels. French, namely Canadian French, is another example, having only /E:/ as a contrastive long vowel.

[Zhen Lin]

RP English, in some sense: /ɪ e æ ʌ ə ɒ ʊ/ vs /iː ɜː ɑː ɔː uː/.

[alice]

It might be worth discounting schwa, since otherwise some interpretations of Slovene and Irish would also qualify.

[Åge Kruger]

It might be worth discounting schwa, since otherwise some interpretations of Slovene and Irish would also qualify.

Bad science.

[Bedelato]

Are we talking about raw phoneme-inventory counts or segment-frequency statistics?

[alice]

It might be worth discounting schwa, since otherwise some interpretations of Slovene and Irish would also qualify.

Bad science.

Just checking you're awake

I'll restate the question then: "Disregarding schwa, which languages have more short vowel phonemes than long vowel phonemes?"

[Zhen Lin]

It might be worth discounting schwa, since otherwise some interpretations of Slovene and Irish would also qualify.

I think a better reason to disregard the case of English would be that vowel length isn't really contrastive, except maybe /ə ɜː/.

[Qwynegold]

Huh, it's the opposite of what I would've tought. So how about frequencies then? To me it seems like short vowels are more frequent than long ones in languages that contrast them, but Persian looks like an exception to me. (But I don't really know anything about that language so I could be wrong.)

[Radius Solis]

I don't really see how that universal can work without limiting it to languages with robust and systematic length distinctions that are intact - i.e. ones that have not undergone subsequent changes. There are certain common types of sound changes that reduce long vowel counts (dipthongization) or increase short vowel counts (conditioned quality splits like reduction to schwa that could become contrastive; also epethetic schwas or other epenthetic vowels), and the "universal" is not, realistically, going to bar such things from happening just because a language already has an equal number of long and short vowels. One example would be the Great Vowel Shift, in which a formerly systematic length contrast lost two members of the long set by diphthongization, /i: u:/ /ai au/, while the short vowels stayed put.

And in a language with only short monophthongs plus diphthongs, the "universal" is not going to prevent monophthongization of certain diphthongs into long monophthongs, either; consider Southern American English, which for some speakers has only one long vowel, /a:/, contrasting with /a/ mainly by duration and corresponding to Gen. Am. /ai/. That's hardly a robust and systematic length contrast! But it does result in a system of nine short vowels and one long vowel.


Whether situations like this tend to stay stable is a different question, and it's very possible they don't. But as we can see, they are, at least, not incapable of showing up for a while.

[linguoboy]

AFAIK, if a language has long and short vowels, there are never more sort vowels than long vowels. Are there any counterexamples?

Standard German As She Is Spoke. A lot of younger speakers have lost the distinction between /a/ and /a:/ although they maintain length distinctions for all other vowels.

(And that's not even getting into the fact that StGer has short close vowels in unaccented syllables in borrowed words, even though in the native system length closeness correlates with length. That's the chief reason for the collapse of /a/ and /a:/: there is no quality distinction between the chief allophones.)

[tezcatlip0ca]

Tariana

Who called it the kitchen-sink natlang? Because it definitely is.

[Tropylium]

Firing up the ol' UPSID…

In all langs with a single long vowel (eg. Mien), it is /aː/.

Two long vowel system reported: French as having separate /ɛː oː/, and Atayal as /iː uː/ + /i u ɛ ɔ a/.

Three long vowel systems reported:
— /a e i o u/ + /aː iː uː/ in Fur
— /a i ɔ u/ + /aː iː uː/ in Bardi
— /a i o ɯ/ + /aː iː oː/ in Adzera
— /a aˤ eˤ i oˤ u/ + /aː iː uː/ for Lak

Finally, for Tamasheq they have /æ ɐ e ə i o u/ + /aː eː iː oː uː/.

---

Somewhat more securely, early Proto-Finnic must've gone thru a stage where it had /ɑ æ e i o u y/ + /eː iː oː uː/ for a while. OTOH, fairly soon afterwards new long vowels /ɑː æː øː yː/ arose, bringing things back on the side of "more long vowels than short". So I'm going to second what Radius said…

Proto-Samic allows for an even wilder analysis: one long vowel /aː/, four short vowels /i ʌ o u/, and four difthongs /ea ie oa uo/ (but with unstress'd allophones [eː ɔː]).

[Travis B.]

Somehow this supposed "universal" seems to be pretty baseless - and is not merely a universal tendency (like, for instance, in a language with fortis and lenis stop series, the most likely fortis stop to be absent is /p/ and the most likely lenis stop to be absent is /ɡ/) but rather is completely without basis in the the first place.

[Travis B.]

AFAIK, if a language has long and short vowels, there are never more sort vowels than long vowels. Are there any counterexamples?

Standard German As She Is Spoke. A lot of younger speakers have lost the distinction between /a/ and /a:/ although they maintain length distinctions for all other vowels.

(And that's not even getting into the fact that StGer has short close vowels in unaccented syllables in borrowed words, even though in the native system length closeness correlates with length. That's the chief reason for the collapse of /a/ and /a:/: there is no quality distinction between the chief allophones.)

By all accounts, it seems like Standard German is headed in the direction that both many Anglic varieties and many continental Scandinavian varieties (including both Standard Swedish and Standard East Norwegian) headed, in that I would not be surprised if it in the near future killed off vowel length as a distinctive feature separate from vowel quality; in varieties lacking a native three-way distinction between /ɛ/, /ɛː/, and /eː/, after losing the distinction between /a/ and /aː/, all it would take then is merging short and long tense vowels in stressed syllables through simply nativizing borrowed short tense vowels. (Of course, there are High German varieties such as Central Bavarian that have already killed off any phonemic vowel length altogether, but that is another story.)

[zompist]

Yet another counterexample: Hindi, which has no long counterparts for short e and o.

It makes me wonder if the universal wasn't stated backwards; but no, then Classical Greek would be an exception, with 6 long and 4 short vowels.

[Travis B.]

Yet another counterexample: Hindi, which has no long counterparts for short e and o.

It makes me wonder if the universal wasn't stated backwards; but no, then Classical Greek would be an exception, with 6 long and 4 short vowels.

It does not work either way, really. For the backwards universal, later Middle English is a counterexample, with /a ɛ ɪ (y) ɔ ʊ/ versus /aː ɛː eː iː ɔː oː uː/, ignoring the nigh-countless and ever-changing diphthongs Middle English had at various points along the way.

[Hakaku]

Is this actually one of the listed universals, or was it just out of curiosity?

[Cedh]

By all accounts, it seems like Standard German is headed in the direction that both many Anglic varieties and many continental Scandinavian varieties (including both Standard Swedish and Standard East Norwegian) headed, in that I would not be surprised if it in the near future killed off vowel length as a distinctive feature separate from vowel quality; in varieties lacking a native three-way distinction between /ɛ/, /ɛː/, and /eː/, after losing the distinction between /a/ and /aː/, all it would take then is merging short and long tense vowels in stressed syllables through simply nativizing borrowed short tense vowels. (Of course, there are High German varieties such as Central Bavarian that have already killed off any phonemic vowel length altogether, but that is another story.)

But there are also new long vowels arising in some close-to-the-standard varieties, mostly from elision of coda /l ʁ/, intervocalic /h/, and (more rarely) intervocalic voiced fricatives. For instance, I merge /aʁ/ in _(C)$ positions into /aː/ myself (i.e. Bart and Bad are homophones: [baːtʰ]. I still have the traditional length contrasts /a ɛ/ :: /aː ɛː/ though), and I have long monophthongs [ɘː ʊː ɔː] for standard /ɪʁ ʊʁ ɔʁ/ [ɪɐ ʊɐ ɔɐ] in the same environment. /ʊʁ/ in particular is very close to merging with original /oː/, and /ɔʁ/ [ɔː] clearly contrasts with short /ɔ/ [ɔ] for me (Dorn [dɔːn] vs. Don [dɔn]).

[/yagpt]

[Dewrad]

AFAIK, if a language has long and short vowels, there are never more sort vowels than long vowels. Are there any counterexamples?

Proto-Celtic had /a e i o u/ vs. /a: i: u:/ (possibly also /e:/, but that's still one phoneme less than the short vowels).

[Radius Solis]

Well, the universal - if it is even listed, which I don't know for sure that it is - does have some basis in principle. Firstly, that greater vowel durations make it easier to contrast more qualities because there is longer for the ear to hear them to tell them apart. And secondly, that in any vowel pair distinguished mainly by length, the shorter should be more prone, on average, to getting reduced or merged into something else... both directly, because of shorter duration giving less time to in which to discern quality distinctions, and indirectly, because short vowels are more likely to go unstressed and unstressed vowels are always more prone to mergers and reductions than stressed ones.

[Salmoneus]

By all accounts, it seems like Standard German is headed in the direction that both many Anglic varieties and many continental Scandinavian varieties (including both Standard Swedish and Standard East Norwegian) headed, in that I would not be surprised if it in the near future killed off vowel length as a distinctive feature separate from vowel quality; in varieties lacking a native three-way distinction between /ɛ/, /ɛː/, and /eː/, after losing the distinction between /a/ and /aː/, all it would take then is merging short and long tense vowels in stressed syllables through simply nativizing borrowed short tense vowels. (Of course, there are High German varieties such as Central Bavarian that have already killed off any phonemic vowel length altogether, but that is another story.)

But there are also new long vowels arising in some close-to-the-standard varieties, mostly from elision of coda /l ʁ/, intervocalic /h/, and (more rarely) intervocalic voiced fricatives. For instance, I merge /aʁ/ in _(C)$ positions into /aː/ myself (i.e. Bart and Bad are homophones: [baːtʰ]. I still have the traditional length contrasts /a ɛ/ :: /aː ɛː/ though), and I have long monophthongs [ɘː ʊː ɔː] for standard /ɪʁ ʊʁ ɔʁ/ [ɪɐ ʊɐ ɔɐ] in the same environment. /ʊʁ/ in particular is very close to merging with original /oː/, and /ɔʁ/ [ɔː] clearly contrasts with short /ɔ/ [ɔ] for me (Dorn [dɔːn] vs. Don [dɔn]).

[/yagpt]

German: still trying to be more like English.

[Xonen]

Somewhat more securely, early Proto-Finnic must've gone thru a stage where it had /ɑ æ e i o u y/ + /eː iː oː uː/ for a while. OTOH, fairly soon afterwards new long vowels /ɑː æː øː yː/ arose, bringing things back on the side of "more long vowels than short".

IIRC, that system is thought to to have developed as early as during the Proto-Finno-Ugric stage, though, which would mean that it actually lasted for quite some time (in the dialects that later became the Finnic branch, at least). However, there are still some unresolved mysteries in the historical phonology of the Uralic languages, so I wouldn't be too surprised if this reconstruction turns out to be less than 100% accurate.

Yet another counterexample: Hindi, which has no long counterparts for short e and o.

This might be one of those cases where it depends on whom you ask... The way I was taught (and which, incidentally, Wikipedia appears to agree with), Hindi has only three short vowels /ə ɪ ʊ/ and seven long ones /aː iː uː eː o: ɛː ɔː/ (Wikipedia adds the eight long vowel /æ:/ but notes that this is found only in loanwords from English). But then, Hindi appears to be one of those languages where vowels are actually distinguished more by quality than length.

[Tropylium]

Somewhat more securely, early Proto-Finnic must've gone thru a stage where it had /ɑ æ e i o u y/ + /eː iː oː uː/ for a while. OTOH, fairly soon afterwards new long vowels /ɑː æː øː yː/ arose, bringing things back on the side of "more long vowels than short".

IIRC, that system is thought to to have developed as early as during the Proto-Finno-Ugric stage, though, which would mean that it actually lasted for quite some time (in the dialects that later became the Finnic branch, at least). However, there are still some unresolved mysteries in the historical phonology of the Uralic languages, so I wouldn't be too surprised if this reconstruction turns out to be less than 100% accurate.

Yeah, it was originally reconstructed that far back, but once we look at the whole development from Proto-Uralic on, it's less that optimal to assume that something like *päxli "side" Mokša päĺ, or Mansi pääl (both coinciding with plain *ä) ever went thru a stage with /eː/. I'm not sure however if this has yet been pointed out in literature anywhere else than Jaakko Häkkinen's Master's thesis…

(Also, as you probably are aware, the existence of a Finno-Ugric stage at all is under suspicion these days. I'm sold on Häkkinen's "East Uralic" (Ugric-Samoyedic) as a better alternativ.)