Formation of vowel harmony?

[Eivuhekoi]

Greetings people!

I have an on-going world-building project, which involves having one language as the "proto-world" language, so that all languages in said project are descendant of said original language. I had the idea of having the various branches from this proto-world (hence forth referred to by its proper name Mist-Ashen or MA for short) resemble "earthian" language families. For example, one of them is going to be inspired by/inspired by the Celtic languages. However, where the question of vowel harmony becomes relevant, it that I plan to draw influence from either Uralic languages as a whole, or Finnic languages specifically. The reasoning for both being that the Finnic language family is something of a "special interest" for me, since I originally got into linguistics via an interest in the dialects of Finnish (my native language).

Vowel-Harmony is something that is featured in most Finnic/Uralic languages, of course there are exceptions such as Estonian. As such I would want this Finnic-inspired language family to have vowel harmony, but the problem is that the Mist-Ashen lacks said feature. Despite my native language having vowel harmony, I am not actually sure of the formation process and, from what I have researched, it seems that vowel harmony was something that was already present in Proto-Uralic. So I am kind of lost with how to impliment it. I have a rough idea in my head of how to divide the vowels into the two groups that I want, front and back vowels, but I have no idea as to how I can get the harmony part itself to emerge.

I'm also trying to prepare for a possibility in which this Finnic-inspired family emerges from a family that has affixes with vowels. Under vowel harmony they would have to appear in both back and front variants, but what if they only have, for example, only back-variants in this theoretical proto-language between Mist-Ashen and the Finnic-inspired languages. Can I just retroactively change the affixes to fit with the noun harmony?

[Creyeditor]

One hypothesis for a diachronic pathway to vowel harmony is reduction of unstressed vowels (such that the later harmonizing feature is not contrastive anymore) and the assimilation of the unstressed voweks to the stressed one.

[Eivuhekoi]

So, unstressed vowels get reduced and then later merge (or change) in a way that they align with a later developed vowel harmony system, if I understood this correctly?

So if the proto-lang had the word *kɑ.'lei but later developed front-back harmony, the process could co something like:
*kɑ.'lei => kɐ.'lei => kæ.'lei

[I used ɐ to mark a reduced vowel, I am not sure how IPA marks them. My only experience of reduced vowels is with the Uralic Phonetic Alphabet.]

[Flavia]

So, unstressed vowels get reduced and then later merge (or change) in a way that they align with a later developed vowel harmony system, if I understood this correctly?

So if the proto-lang had the word *kɑ.'lei but later developed front-back harmony, the process could co something like:
*kɑ.'lei => kɐ.'lei => kæ.'lei

[I used ɐ to mark a reduced vowel, I am not sure how IPA marks them. My only experience of reduced vowels is with the Uralic Phonetic Alphabet.]

Yeah, it could go like that. Also, here's a video by Artifexian on the subject, maybe it'll be helpful.

---

By the way, welcome to the CBB! Nice to see you here.

[Eivuhekoi]

So, unstressed vowels get reduced and then later merge (or change) in a way that they align with a later developed vowel harmony system, if I understood this correctly?

So if the proto-lang had the word *kɑ.'lei but later developed front-back harmony, the process could co something like:
*kɑ.'lei => kɐ.'lei => kæ.'lei

[I used ɐ to mark a reduced vowel, I am not sure how IPA marks them. My only experience of reduced vowels is with the Uralic Phonetic Alphabet.]

Yeah, it could go like that. Also, here's a video by Artifexian on the subject, maybe it'll be helpful.

---

By the way, welcome to the CBB! Nice to see you here.

omg I'm such an idiot. Why didn't I think of checking if Artifexian had made a vid on it . Makes my question rather pointless in hind-sight.

Also thanks for the welcome ^^

[Salmoneus]

I'd suggest looking past the labels ("vowel harmony"), and thinking about what actually happens.

What actually happens to create harmony is simple: features from one phoneme spread into nearby phonemes. In the case of vowel harmony, features of one vowel spread across into some or all of the other vowels in the word.

What makes this 'harmony' rather than simply 'umlaut' (or some similar term) is that in harmony the vowels produced in this way form a set that don't appear elsewhere.

To take a super-simple example: rounding harmony can be created simply by saying "if a word contains a rounded vowel, all other vowels become rounded". This means that rounded vowels can only appear in words containing other rounded vowels, and cannot appear in words not containing rounded vowels, while unrounded vowels cannot appear in words containing rounded vowels.

Complications include...

- do the features spread forward (all vowels after the 'trigger' vowel take on that feature) or backward (all vowels before the trigger take on that feature) or both (all vowels in either direction of the trigger take on the feature)?

- can the spread of features be 'blocked' by certain consonants? Some languages, for instance, will have spread across non-stops, but no spread across stops. Or across voiced consonants, or dorsal consonants, etc.

- are affixes affected by the spreading? In some languages they aren't, in some they are. In some languages only the affixes are affected, while roots themselves can be non-harmonic (usually then affixes must harmonise with the nearest vowel of the root, so suffixes harmonise with the final vowel, prefixes with the initial vowel).

- are all vowels initially influenced by the feature-spreading? For instance, maybe high vowels round when another vowel is rounded, but low vowels don't.

- do all vowels equally trigger the feature-spreading? FWIW, it's actually more common for mid-vowels to trigger harmony than for high or low vowels to do so, though that's not an absolute rule.

- have there been any mergers since harmony was established? This can create 'neutral' vowels

- have any vowels moved since then? This can create counter-intuitive patterns of harmony

- do consonants interfere in the vowel harmony patterns by imposing their own local influences on vowels? [eg a language with harmonic vowel rounding, but in which non-harmonic rounding also occurs after (past or present) labiovelars]

----------

But in general I think the key is not to think of vowel harmony as about a language having two different sets of vowels, as though given them as segregated sets since the dawn of time, but rather to simply think about sound changes that regularly tend to prevent certain vowels from occurring close together


-----------

Regarding Crey's suggestion of initial vowel reduction: this is possible but not necessary. Essentially, unstressed and reduced vowels will be more likely to be influenced by nearby vowels; this is particularly useful if you want total harmony (where unstressed vowels end up exactly matching nearby stressed vowels). But stressed vowels can be harmonised anyway if necessary.

[Creyeditor]

to everything Sal said.

Regarding Crey's suggestion of initial vowel reduction: this is possible but not necessary. Essentially, unstressed and reduced vowels will be more likely to be influenced by nearby vowels; this is particularly useful if you want total harmony (where unstressed vowels end up exactly matching nearby stressed vowels). But stressed vowels can be harmonised anyway if necessary.

Yes, this is true. Metaphony in Romance languages works like that, i.e. the stressed vowel assimilates to the unstressed vowel in at least some varieties (even though this is a bit more local than 'real' vowel harmony). I think ATR harmony in some Spanish variety also works like this but I might misremember.

[thethief3]

Final vowel triggered Umlaut + final vowel loss is a way i have come up to to get vowel harmony although i don't think i have used it.

So *patoki > *pädök and *patoku > *padok. This would lead to morpholigical alterations in previously suffixing languages but its great for prefixing languages.

[Man in Space]

You can also just happen to blunder into it. You can get harmony through sound changes, applied judiciously.

For example, in some Arabic dialects (like my professor’s), the emphatic consonants affected nearby vowels (particularly /a a:/), and this emphatic status went on to affect all vowels in a word with an emphatic consonant in it. The thing on the consonants that defined emphasis was later lost, but the vowels still retained their affected pronunciations.

Something along those lines, such as lag assimilation/dissimilation, is a totally plausible route.