How does vowel harmony typically arise in a language?

Question

How does vowel harmony typically arise in a language?

Here's a definition of vowel harmony from the WALS chapter on Vowel Quality Inventories: http://wals.info/chapter/2.

"When a language is said to have vowel harmony this generally means that within a word, including any affixes, it is only possible to combine the members of certain subsets of the vowels together."

Here's an example of vowel harmony from Wikipedia's/Free Net Encyclopedia's article on vowel harmony: http://www.netipedia.com/index.php/Vowel_harmony#Features_of_vowel_harmony

"The vowel that causes the vowel assimilation is frequently termed the trigger while the vowels that assimilate (or harmonize) are termed targets. In most languages, the vowel triggers lie within the root of a word while the affixes added to the roots contain the targets. This may be seen in the Hungarian dative suffix:

Root    Dative  Gloss
város   város-nak   "city"
öröm    öröm-nek    "joy" 

The dative suffix has two different forms -nak/-nek. The -nak form appears after the root with back vowels (a and o are both back vowels). The -nek form appears after the root with front vowels (ö and e are front vowels)."

The article at Free Net Encyclopedia covers a lot of ground, but leaves me curious about how vowel harmony typically arises. Does it arise in languages with big vowel quality inventories, or do big vowel quality inventories arise in response to it? Does the consonant inventory play any role in its development? Does prosody play a role?

Answer 1

Probably the most common explanation of how vowel harmony starts is that it's a grammaticalization of the phonetic effect of coarticulation, where the properties of one segment influence how the speaker articulates surrounding segments.

For example, when an upcoming syllable contains a round vowel, it's common for speakers to begin rounding their lips partway through the preceding syllable or earlier (even two or three syllables earlier). In speech communities where speakers do a lot of vowel-to-vowel coarticulation, it's possible that learners in future generations will reanalyze what they hear as being contrastive rounding (or whatever) on more than one syllable.

There's probably no relation to the size of the vowel inventory. In the 1990s, Sharon Manuel proposed that languages with larger vowel inventories tend to allow less vowel-to-vowel coarticulation (so maybe we could hypothesize they'd be less likely to develop vowel harmony in the future), but her proposed relationship between inventory size and coarticulation hasn't really held up beyond the languages she first looked at.

Prosody should definitely have a role according to this story. Segments in prosodically strong positions tend to exert stronger coarticulatory effects on their prosodically weak neighbours than vice versa. So along with other tendencies (like right-to-left being more common than left-to-right), you'd expect more languages to develop harmony patterns where features spread from prosodically strong segments than to them.