How does Vowel Harmony develop?

[zuben]

This question has probably asked loads of times before, but I didn't find anything by searching so here we go: what are some ways vowel harmony can develop in a lang?

Asking because I'm thinking about incorporating it properly in my conlang. At the moment I've got a rudimentary front/back harmony system, where the final vowel of each word in the protolang elicits all the others to be either front or back in the daughter lang. I'm guessing this isn't very realistic, and it's more likely to be elicited by the stressed syllable, but since I know next to nothing about vowel harmony I thought I ask for some help.

[linguoboy]

This question has probably asked loads of times before, but I didn't find anything by searching so here we go: what are some ways vowel harmony can develop in a lang?

From what I've seen, it's essentially a form of non-adjacent assimilation that gets elaborated by analogy.

You might want to have a look at the history of umlaut in the Germanic languages. You essentially had a nascent vowel-harmony system in Common Germanic until it was disrupted by later developments.

[Count Iblis]

Asking because I'm thinking about incorporating it properly in my conlang. At the moment I've got a rudimentary front/back harmony system, where the final vowel of each word in the protolang elicits all the others to be either front or back in the daughter lang. I'm guessing this isn't very realistic, and it's more likely to be elicited by the stressed syllable, but since I know next to nothing about vowel harmony I thought I ask for some help.

That's actually perfectly realistic. It's known as regressive vowel harmony. Germanic languages had it even though stress was on initial syllables. It may have to do with the brain anticipating which vowel sound is coming up and then spreading certain features to the vowels the mouth is currently pronouncing.

[alice]

With vowel harmony the assimilation tends to be progressive rather than regressive, but the principle is much the same.

[Radius Solis]

With vowel harmony the assimilation tends to be progressive rather than regressive, but the principle is much the same.

I don't know that that's true. Rather the term vowel harmony seems to be used more when root vowels affect affix vowels than the other way around, or at least that's the impression I've gotten, and of course suffixes are more common than prefixes. But AFAIK assimilation direction is correlated with branching order: VO languages are more likely to have regressive assimilation, while OV languages are more likely to have progressive assimilation. (There are, of course, languages which do not obey such tendencies.)

[zuben]

With vowel harmony the assimilation tends to be progressive rather than regressive, but the principle is much the same.

I don't know that that's true. Rather the term vowel harmony seems to be used more when root vowels affect affix vowels than the other way around, or at least that's the impression I've gotten, and of course suffixes are more common than prefixes. But AFAIK assimilation direction is correlated with branching order: VO languages are more likely to have regressive assimilation, while OV languages are more likely to have progressive assimilation. (There are, of course, languages which do not obey such tendencies.)

Ah ok, that's good, it fits in with my conlang which is VO.

Of course, I guess if I go for regressive harmony then that will result in umlaut-esque vowel alternation in the stem with suffixes, whereas if I go for progressive then that will give two sets of suffixes to match the vowels of the stem a la Turkish. It's just a matter of deciding which I'd prefer in my conlang.

[linguoboy]

Of course, I guess if I go for regressive harmony then that will result in umlaut-esque vowel alternation in the stem with suffixes, whereas if I go for progressive then that will give two sets of suffixes to match the vowels of the stem a la Turkish. It's just a matter of deciding which I'd prefer in my conlang.

Nothing says that it can't go both ways. It would kind of depend how (and at what stage) your suffixes originate.

[Skomakar'n]

I don't know why, but Icelandic has a little vowel harmony, where, before <u> /ʏ/, <a> /a/ changes into <ö> /œ/.
This is not true for the masculine suffix -ur, which was just -r in Old Norse and Old Icelandic.

tala - to speak/I speak/they speak
tölum - we speak

gamall - old (masculine)
gömul - old (feminine)

kallaður - called (masculine) [masculine suffix -ur]
kölluð - called (feminine)

flatastur - flattest (masculine nominative singular) [masculine suffix -ur]
flötustum - flattest (masculine dative singular, all genders dative plural)

Faroese still has it to some extent, but less so than Icelandic, and it's a little less regular; while Icelandic always turns <a> into <ö>, Faroese turns it into either of <ø o> depending on the word, and sometimes it doesn't seem to change at all.

The following three are declined in the masculine nominative singular followed by the masculine singular and all genders plural dative (in Faroese this is also the neuter singular dative; in Icelandic this form has no -m).

long:
Icelandic: langur, löngum
Faroese:: langur, longum

flat:
Icelandic: flatur, flötum
Faroese: flatur, fløtum

glad, happy:
Icelandic: glaður - glöðum
Faroese: glaður - glaðum

The following are declined in the masculine singular nominative followed by the feminine singular nominative:

fair, beautiful:
Icelandic: fagur - fögur
Faroese: fagur - føgur

old:
Icelandic: gamall - gömul
Faroese: gamal - gomul

called (past participle):
Icelandic: kallaður - kölluð
Faroese: kallaður - kallað

Just to show you that the change can either be very regular, like in Icelandic, or pretty irregular, like in Faroese.
Another example, is that an Icelandic word such as stjarna (star) would decline into stjörn- in all other forms, since they end with -u, -ur or -um (except for the plural genitive, which is also stjarna), and we get forms like stjörnu (all other singular forms), stjörnur (nominative and accusative plural) and stjörnum (dative plural). In Faroese, by analogy, the only two forms not having gone through this umlaut have also gained this change, even though they still end in -a; in Faroese the nominative singular and genitive plural are stjørna, and the other forms are stjørnu, stjørnur and stjørnum.

[Jipí]

The thing in Icelandic is called a-umlaut IIRC. As Linguoboy said, umlaut is essentially a process that may turn into vowel harmony, since both are kind of an assimilation process.

[EDIT] Wikipedia calls it u-mutation: http://en.wikipedia.org/wiki/Germanic_u ... _Icelandic [/EDIT]

[Skomakar'n]

The thing in Icelandic is called a-umlaut IIRC. As Linguoboy said, umlaut is essentially a process that may turn into vowel harmony, since both are kind of an assimilation process.

[EDIT] Wikipedia calls it u-mutation: http://en.wikipedia.org/wiki/Germanic_u ... _Icelandic [/EDIT]

I know. I know. I just wanted to point this out, because it strikes me as vowel harmony as well.

[Magb]

The modern a > œ / _C*u change in Icelandic was originally (i.e. in Old Norse) a > ɒ / _C*u, then later /ɒ/ shifted to /œ/ unconditionally. The Old Norse system arguably makes more sense vowel harmony-wise than the modern Icelandic and Faroese system. In Old Norse you can see how the rounding of the following /u/ spread to the /a/ and made it [ɒ], which went on to become phonemic.

[Echobeats]

The thing in Icelandic is called a-umlaut IIRC. As Linguoboy said, umlaut is essentially a process that may turn into vowel harmony, since both are kind of an assimilation process.

[EDIT] Wikipedia calls it u-mutation: http://en.wikipedia.org/wiki/Germanic_u ... _Icelandic [/EDIT]

I know. I know. I just wanted to point this out, because it strikes me as vowel harmony as well.

Technically it's not called vowel harmony when the stem vowel changes, only when affixes harmonise with the stem. So Turkish counts but Icelandic doesn't.