zompist bboard

I still don't understand it people. Could someone explain with examples, but not getting too technical...

Thanks,
Ghost

Notice how, for each word in this passage, the word either contains only vowels from the group { a, i, o, u }, or only vowels from the group { e, i, ?, ? }, never from both groups:

erte(#) urida(*) kabalik(*) balgasun-dur(*) biraman-u(*) qamug(*) uqagan-u(*) j?il-d?r(#) mergen(#) bolugsan(*) sain(*) t?r?l-t?(#) kemek?(#) nigen(#) biraman(*) b?l?ge(#). ...

(*) first group { a, i, o, u }
(#) second group { e, i, ?, ? }

Interesting. Is there any pattern in the way the vowels are split into the groups?

Usually it's front vs back, round vs unround, or high vs low. But average people are not linguists, and they dont often follow the rules. For example, notice that /i/ is in both groups.

In bicoherent's example, the defining form is frontness/backness (though what is considered front and what is considered back can sometimes be something else, from what I understand). However, other forms include rounding and height harmony.

For round harmony, imagine a language has six vowels: a rounded series /y & u/ and an unrounded series /i a M/. Now, the rule is any morpheme will conform to the roundness of the stem: that is /ni/ when attached to /k&n/ becomes /k&p.ny/, or /ku/ to /pan/ becomes /pan.kM/.

Height harmony works in the same way: vowels will tend to conform to the height of the root vowel (or perhaps, to some other degree. In a language that is just root + 1 suffix, the root may harmonize with the suffix!). So if you had two phonemoic heights, /i u/ and /a Q/ then /kap/ + /ni/ would yield /kapna/, etc. (I'm not sure if three-height would work the same...)

Mercator wrote:Usually it's front vs back, round vs unround, or high vs low. But average people are not linguists, and they dont often follow the rules. For example, notice that /i/ is in both groups.

But that was a natlang example, so it looks like real languages don't follow the rules either. In which case, why are they rules?

Ahribar wrote:

Mercator wrote:Usually it's front vs back, round vs unround, or high vs low. But average people are not linguists, and they dont often follow the rules. For example, notice that /i/ is in both groups.

But that was a natlang example, so it looks like real languages don't follow the rules either. In which case, why are they rules?

But the rules are being followed. It's just that "i" is in both groups.

Usually vowels that are found from both series wary a bit. Eg. Finnish /i/ is a neutral vowel that can be in both back and fron words. Usually its sound is but close to back vowels it gets more backish, something like or even [1]. Same kind of alternation happens with the other neutral vowel /e/. The only thing that makes these allophones and not phonem pairs, like a-?, o-? and u-y, is that they are not distinktive.

Vowel harmony must not affect all the vowel phonemes. It can as well affect only to one phoneme. An example are the eastern dialects of North S?mi (though it's only allophonlic variation). In the standard language there is no vowel harmony and all of the vowels a (back a), ? (front a), e, i, o, u can occure together. But in the eastern dialects the ? varies quite much. Itself and near e and i it's [{] but with a, o and u it's [A] or even [Q]. Althoung this is only allophonic it has the same nature as the phonetic vowel harmony.

Another kind of vowel harmony is found in Igbo, IIRC; there are two groups called "tense" and "lax", each of four vowels. I think the tense set contains /i u e a/ and the lax set /I U E o/. A word can only contain vowels from one set, not both.

gach wrote:Vowel harmony must not affect all the vowel phonemes.

Very minor semantic quibble: I think you mean "... does not have to affect"; otherwise my Igbo example is invalid

I think it might be interesting and useful to gather together several types of vowel harmony, with examples. E.g. tense/lax, as in Igbo; front/back, as in Finnish; high/low, as in I don't know what; and any others I may have missed.

Mercator wrote:Usually it's front vs back, round vs unround, or high vs low. But average people are not linguists, and they dont often follow the rules. For example, notice that /i/ is in both groups.

Ah, but that is an error. The example was from Turkish and in the * group the i was supposed to be with out the dot, which means .

vegfarandi: It's Mongolian, actually.

bicoherent wrote:vegfarandi: It's Mongolian, actually.

Woops. But in Turkic, it's like I pointed out.

geoff wrote:I think it might be interesting and useful to gather together several types of vowel harmony, with examples. E.g. tense/lax, as in Igbo; front/back, as in Finnish; high/low, as in I don't know what; and any others I may have missed.

The other one mentioned so far that's not in your list is rounded/unrounded. Does anyone know any natlang examples of this?

good thread, I've always wondered how vowel harmony works

Ahribar wrote:

geoff wrote:I think it might be interesting and useful to gather together several types of vowel harmony, with examples. E.g. tense/lax, as in Igbo; front/back, as in Finnish; high/low, as in I don't know what; and any others I may have missed.

The other one mentioned so far that's not in your list is rounded/unrounded. Does anyone know any natlang examples of this?

in another thread, I wrote:Turkish has some rounding harmony in addition to front/back harmony:

The possessive form of 'eshek' is 'eshekin', but for 's?z' it's 's?z?n'.
Similarly, 'ac?' > 'ac?n?n' but 'kol' > 'kolun' (<?> is /1/).

*For all you Turkish speakers, I'm using <sh> for s-cedilla and <?> for dotless i... damn Unicode won't work again.

Ahribar wrote:

geoff wrote:I think it might be interesting and useful to gather together several types of vowel harmony, with examples. E.g. tense/lax, as in Igbo; front/back, as in Finnish; high/low, as in I don't know what; and any others I may have missed.

The other one mentioned so far that's not in your list is rounded/unrounded. Does anyone know any natlang examples of this?

Kyrgyz, another of the Turkic languages (like Turkish), makes extensive use of both high/low and rounded/unrounded harmony.

p@,
Glenn

In natlangs...

do they use 2 style harmony:.. High v Low or Front v Back

or

4 Part: High, Low, Front, Back

Hmm. I only know about Hungarian and Finnish harmony, which is essentially front-back.

For my conlang, of which the grammar can be found here, I use a system in which there exists a distinction between back, middle and front, with one vowel having three versions, one having a back and a middle/front version and one with a back/middle and front version. The rest remains always the same. Would such a system be plausible?

daan wrote:Hmm. I only know about Hungarian and Finnish harmony, which is essentially front-back.

For my conlang, of which the grammar can be found here, I use a system in which there exists a distinction between back, middle and front, with one vowel having three versions, one having a back and a middle/front version and one with a back/middle and front version. The rest remains always the same. Would such a system be plausible?

I looked over your phonology..

Your right, one has to read it carefully

doctrellor wrote:In natlangs...

do they use 2 style harmony:.. High v Low or Front v Back

or

4 Part: High, Low, Front, Back

This isn't a direct answer, but...

In the examples I've seen from Kyrgyz (mentioned above), the front/back and rounded/unrounded harmony is simultaneous; both dimensions apply together. That is, if a root word (or the last syllable of a polysyllabic or compound word) contains a rounded front vowel, then the grammatical suffixes that follow it (Kyrgyz, like most Turkic languages, is largely agglutinative and suffixing) will also contain rounded front vowels. The same applies to the other three categories (front unrounded, back rounded, and back unrounded), although I suspect that the full picture is more complex and depends in part on the particular suffixes in question.

I would think that a language with front/back and high/low vowel harmony might do something similar (this is only one possibility, though).

p@,
Glenn

That certainly puts some strict limits on the possible affixes.

Most languages with complex phonologies hardly use any of their possible sound combinations for affixes, prepositions, etc. anyway ... for example there is no final -c in Latin declensions, as far as I know. or -b, -f, etc.

Eddy the Great wrote:That certainly puts some strict limits on the possible affixes.

No, they change to fit the vowel harmony (Hungarian, this time):

olvas (read) -> olvasok (I read)
k?r (ask) -> k?rek (I ask)
s?t (bake) -> s?t?k (I bake)

Eddy the Great wrote:That certainly puts some strict limits on the possible affixes.

As a rule, each affix will have multiple forms, depending on the form of the root it follows.

I'll take a basic example from Kazakh, the language with vowel harmony that I know best. Kazakh makes use of front/back harmony and a degree of consonantal harmony with its grammatical affixes (there is no real rounded/unrounded harmony--Kazakh affixes generally contain only unrounded vowels). The plural ending in Kazakh takes the following forms, depending on the ending of the noun (or pronoun, for the 2nd and 3rd persons) being pluralized:

If the noun ends in a vowel or /r/, /j/, or /w/ = -lar/ler

(Example: bala = child, balalar = children; kisi = person, kisiler = people)

If the noun ends in /l/, /m/, /n/, /N/, /z/, or /Z/ = -dar/der

(kalam = pen, kalamdar = pens; kilem = carpet; kilemder = carpets)

If the noun ends in a voiceless consonant or /b/, /v/, /g/ or /G/ = -tar/ter

(aghash = tree, aghashtar = trees; mektep = school, mektepter = schools)

As can be seen, in addition to the difference in the initial consonant, the plural endings show front/back vowel harmony, with /a/ if the last syllable of the root word contains a back vowel, and /E/ (<e>) if it contains a front vowel. (In terms of root words, while many are harmonized internally, others--especially compound words or foreign borrowings--contain both front and back vowels; the final vowel determines the form of the affixes.)

Most noun and verb affixes in Kazakh and other languages with vowel harmony have multiple forms of this kind--at least two for each affix; for long agglutinative constructions, this applies to the entire string of affixes following the noun. (For a language with two dimensions of harmony, like Kyrgyz, the number of affix forms is doubled for affixes that exhibit both forms of harmony.) This may sound like a large number of forms, but the process of adding the right one is generally quite simple and even automatic.

(Note that irregularities also exist: in Kazakh, while all the other case endings follow vowel harmony, the instrumental ending (-men/-ben/-pen) does not; its vowels are always "front". In a different kind of irregularity, the plural of the 3rd person pronoun on ("he/she/it") is olar ("they"), rather than ondar.)

p@,
Glenn

EDIT: I see that pharazon summed up the point ahead of me, and far more concisely! Note, by the way, that Hungarian is another language that exhibits both front/back and rounded/unrounded vowel harmony.

Besides the aforementioned forms of vowel harmony, there are also some which are harder to specify, such as the advanced/non-advanced tongue root harmony of written Manchu.