aliensdrinktea wrote: ↑14 Oct 2020 22:57
I'm not well-versed on vowel harmony, but I'm trying to understand it. I think I have the gist of how it works, but I'm a bit confused about diphthongs. If a language has, say, rounding/backness harmony, could /ai/ coexist in a word with /u/, or /au/ with /i/? (Let's assume that /a/ is neutral.)
I'll caveat this with: I'm certainly not an expert on this topic, so take everything with a pinch of salt and I'm happy to be corrected. That said, I have read a few things about it and hopefully remember some of it, so I'll take a stab...
The first thing to say is that nothing is certain. Vowel harmony systems are very varied.
Second: I assume you're treating /i/ as front/unrounded. Counterintuitively, it's actually relatively common for /i/ to be treated as neutral - perhaps because it's so 'extreme' that it often resists umlaut effects? It may also be because lots of front/back harmony systems have actually started out as other sorts of harmony and have gradually shifted in a way that can lead to counterintuitive exceptions.
However, that said, assuming that in this language's harmony rules /a/ is neutral, /i/ is front, and /u/ is back...
- I believe diphthongs are often exempt from harmony rules, so in that case, yes, these things would be allowed, yes. Likewise, long vowels are often exempt.
- however, they're not exempt in all languages! One issue might be whether your language treats [ai] as underlyingly /aj/ (a unitary diphthong), or as /ai/ (phonemically a sequence of two vowels in hiatus, even if phonologically it collapses to a diphthong in pronunciation). The former is probably less likely to trigger harmony, and much less likely to be altered by it.
- it's also possible in some languages for vowel harmony to be 'blocked' by certain vowels - not only do they not change, but vowels before them (or after them, depending on which direction the harmony works) don't change either. Long vowels can block in some languages, and I wouldn't be surprised if diphthongs can do too.
[in general, as with so many language things: although people commonly talk about "vowel harmony" as though it were its own unique thing, totally different from other vowel changes like simple local umlaut, the reality is often a bit more blurry. There are pure, ideal vowel harmony systems... but there are also confusing, incomplete vowel harmony systems too.]
What if unstressed diphthongs have a tendency to merge into monophthongs? Is that permitted when the resulting monophthong would violate vowel harmony rules? Does vowel harmony have any influence on pronunciation if this is the case?
Anything is permitted!
I don't think it's likely that normal harmony would prevent monopthongisation per se, although it could do so indirectly in some cases (eg by turning an easily-reduced diphthong into one that resists monophthongisation more strongly).
Other than that, though, I think the key question is: how 'active' is the harmony? Harmony looks like a synchronic system - a set of rules the current language follows - but it's probably best thought of as the result of a sound change. Like all sound changes, it's active at a particular time - and for a particular length of time. So, you can have a perfectly regular vowel harmony language - and then introduce a bunch of loanwords and discover that the loanwords aren't altered to obey vowel harmony rules: in this case, it looks like there's a rule in the language, but it's not actually an ongoing sound change, so the new words aren't changed. On the other hand, another language, even if it doesn't look to have as perfect a vowel harmony system, may still have an active sound change that is gradually altering words, including loanwords, to obey vowel harmony rules.
So, in your case you have two soundchanges, or perhaps more accurately two sound change rules: one that reduces diphthongs; and one that produces harmony. But what order do they operate in?
If you have harmony first, and then reduction, then you end up with a seemingly irregular vowel harmony system - certain vowels just don't trigger (or aren't affected by) harmony, for no obvious synchronic reason (or are affected, but in a different way) , or you may even have vowels that trigger it in some words and not (or differently) in others (depending on where the vowel comes from). This isn't 'ideal' vowel harmony, it's messy, but as I say, that doesn't mean it doesn't happen in real languages!
But, depending on the rules of the reduction, you may still effectively have harmony working, because the diphthong may be altered first, and the altered diphthongs may lead to different vowels after reduction. Or not! If your harmony change is /au/ > /ay/, but then your reduction changes both /au/ and /ay/ to /a:/, then it looks like harmony doesn't affect /a:/, even though historically it did...
If you have the reduction first, on the other hand, and then the harmony, then the new monopthongs may be just as influenced by harmony as any other vowels (although if they're long, there's a chance they'll be immune anyway).
OR: you can have harmony, and then reduction, and then more harmony again (i.e. the harmony rule remains active the whole time, before and after the other change)....
There's a word I want to coin, /kiʔau/ (or /kiʔaw/, if you prefer). The stress in this word would fall on the first syllable, resulting in [ˈkiʔo]. But combining a front unrounded vowel and back rounded vowel is a no-no, at least at the phonemic level. So my questions are these:
1. Would /kiʔaw/ be permitted in the first place, given the above restriction?
Probably, yes. If /aw/ derives from an earlier /au/, that existed when there was already harmony, then it's possible that it would already have been altered, and hence the /aw/ would never have arisen. However, if the /au/ already reduced to /aw/ before harmony arose, or if it's from, say, /al/ or the like, then there's no reason it couldn't exist (although I think it's also possible that the language may alter it through harmony regardless of its origin! - but I also think that while that's possible, it's not necessary). Or even if it's from /au/, the vowel harmony rules may just be blocked by /a/ followed by a vowel anyway, or some similar rule...
2. If so, would it be realized as [o], or as another allophone that's less fronted and/or unrounded?
Whatever you want!
Basically, assuming you do have /ki7aw/ (sorry, can't be bothered to copypaste the glottal stop symbol...), and assuming that /aw/ does >/o/, then the speakers of the language will face the exact dilemma that you do: woah, this word sounds weird!
But what they do about it is up to them. I think there are four plausible options for them:
a) say "that's impossible! Harmony is all-important!", and continue to apply the harmony rule to the new vowel. They may say /e/ is the fronted version of /o/, therefore this word must logically be /ki7e/!
b) say "that's impossible! Harmony is all-important!", and continue to apply the harmony principle to the new vowel, but not be convinced that /o/ > /e/ should be an ongoing rule, and instead front it to a new front rounded vowel. You'd then have two different phonemes that produce /o/ in back contexts - one of which fronts to a front unrounded vowel, and one of which fronts to a front rounded vowel. Or, for a really fun version: maybe original harmony worked from left to right, but because it's perfect and exceptionless and there are no suffixes or prefixes (or they don't get involved in harmony), the speakers forget this, and apply the new harmony from right to left.... [this isn't exactly likely, but it wouldn't be freakishly bizarre]
c) say "oh, screw it" and give up on harmony. Most words would still look like they obey vowel harmony, but words with vowels descending from former diphthongs would just not obey it, and new loanwords and derivations wouldn't obey it either.
d) be indecisive, and try to find a middle way. In this scenario, they maintain harmony as a general rule, but accept that certain words are exceptions. So again you'd have two different sorts of /o/ - one which was involved in harmony, and one which wasn't.
In the long run, the language will either tend toward imposing exceptionless harmony, or else it will tend to gradually abandon harmony. But it's possible for it to be in an irregular intermediate condition for a long time!