Phonology
This won't be complete, but I want to include enough to give some context to the points I'm least sure about. (You can skip to the end to see the particular issues that right now strike me as most up in the air.) And anyway, it's long enough even if it's not complete.
In case you're interested, significant bits of this were directly inspired by the Mayan language Mam, via the grammar by Nora England.
There'll be three main parts to this post: the segmental inventory, vowels and syncope, and suprasegmentals. It's the suprasegmentals I'm least sure about, though lots of it is a bit outside my comfort zone.
The segmental inventory
The phonemes:
Code: Select all
p t ts tɬ k kʷ
pʰ tʰ tsʰ tɬʰ kʰ kʷʰ
p' t' ts' tɬ' k' kʷ' ʔ
f s ɬ xʷ
m n ɲ ŋʷ
mˀ nˀ ɲˀ ŋʷˀ
r l ɫʷ
j w
ȷ̃ w̃
i ɨ u
e ə o
a
All that's using the IPA, except that I've skipped the ligature diacritic for digraphs (e.g.,
ts instead of
t͡s).
Some of those phonemes might end up being analysed away, as sequences rather than individual phonemes. The morphophonology will often invite this analysis, with a morphologically produced /t/ + /s/ (for example) ending up indistinguishable from what the table above calls
ts. Syllable boundaries will lead to exceptions, though, maybe frequent ones; syllabification isn't yet nailed down.
I'll probably end up with an orthography that departs with the IPA in some ways, and I'll be soliciting suggestions about that at the end of the post.
It's natural to think of Kwa̰ Mɨɨrts ejectives and glottalised nasals as sharing a feature, maybe [+glottal], because within an onset or a coda this feature will move so that it is realised next to the vowel (maybe most saliently by a bit of creak on the vowel itself). This means, for example, that
mat' would end up not with an ejective but with a preglottalised stop. Aspiration works the same way, so that
mˀkatʰ would end up being more like
mk'aʰt, with an ejective and a preaspirated stop.
The glottal stop will be predictable in onsetless syllables but will be contrastive otherwise, either colouring a preceding vowel or merging with a following consonant. For example, a glottal-stop-final verb like
tɬeaʔ will convert the first person subject marker
-tsɨ to
-ts'ɨ, yielding
tɬets'ɨ (the stem's unstressed vowel has also deleted).
The only phonetic detail I feel sure of is that
p' will usually end up as the implosive [ɓ] rather than an actual ejective.
Vowels, stress, and syncope
There's a verb (I don't know yet what it means) whose stem alternates between
ŋʷet and
ŋʷat; another that's either
piw or
pwa; and a third that's always
fkɨt. When emphasised, these can instead be
ŋʷeʔət or
ŋʷəʔat,
piwə or
pəwa, or
fəkɨt; or even
ŋʷeʔat,
piwa, and
fekɨt. What to say about cases of these sorts (which are legion)?
One thing you can say is that some verbs have two stems, and you can insert epenthetic vowels for the purposes of emphasis. (Like English "strange" can become something like [stəˈrejndʒ], for example.)
Another thing you can say is that the underlying forms of these verbs are
ŋʷeat,
piwa, and
fekɨt, and that unstressed vowels delete. (The first two would have stress on the first syllable, the other one on the second syllable.)
I'm going to adopt the second sort of description here, though it's got its problems. It allows me to say that Kwa̰ Mɨɨrts syllables are maximally CVC, which is nice. It also allows a nifty explanation of the alternations mentioned above: most words have stress on the root's initial syllable, but there are lexical exceptions like
fekɨt; and some suffixes attract stress to the syllable that precedes them.
The main disadvantage of this approach is that the posited underlying forms are often underdetermined by the surface forms. Underlying
fekit never gets stress on its first syllable, so we don't really know what that vowel should be. In cases like that I'll take the orthographically simplest route and choose
e.
A bit more detail. Assuming these underlying forms, one syllable in the root, usually the first, will get primary stress; then every second syllable before or after that will have secondary stress. It's the vowels with neither primary nor secondary stress that delete. Vowels with secondary stress centralise: they go to
ɨ,
ə, or
a, depending on height.
Again you get underdetermined underlying vowels in syllables that never receive primary stress. This is the case for example with vowels in affixes, which (so far at least) can never get primary stress. Take the first person subject marker, which surfaces either as
ts or as
tsɨ: is the underlying vowel
i,
ɨ, or
u? As above, I'll side with simple orthography, and assume
i.
For a simple example, take the free first person pronoun,
tsʰī, and its plural,
tsʰēk. There's reason to suppose that underlying these forms we have
tsʰie[+long] (see below for suprasegmental vowel length) together with a suffix
-k that attracts stress to the syllable preceding it. Without the suffix, the
i is stressed and therefore hosts the [+long] feature, while the
e drops, yielding
tsʰī. With the suffix, stress, along with the vowel length, moves to the
e while the
i drops, yielding
tsʰēk.
For a sneakier example, take the morpheme (or whatever) that I'll write as
E. It's a vowel, but it also attracts stress to the previous syllable, which means that it itself is unstressed and therefore deletes. Thus
piwa +
E +
sa (
sa cross-references a first person object) gives you
pwasa.
Like I said above, you could probably do all of Kwa̰ Mɨɨrts morphophonology just by talking about allomorphs, without positing unifying underlying forms. In fact I'm inclined to say that in the fictional present newly borrowed words no longer undergo this process. We'll see.
Suprasegmentals
I just gave an example involving a suprasegmental feature, [+long]. Let's go over how that works, before talking about creak and tone.
The [+long] feature is generally supplied by the root or some affix, though they can also end up floating. A word is [+long] if one or more of its components is, and if a word is [+long] then its stressed vowel surfaces long.
So vowel length is a thing, but it's not happily represented by doubling the vowel. When I'm feeling a bit fancy I'll superscript the feature specification, as in
tsʰie[+long]. More often I'll use a macron, as in
tsʰī. (But I'll ask for advice about this part of the orthography below.)
Another example. There's a plural morpheme that shows up on some pronouns that might be represented as
-Em[+long]. This takes the third person feminine and masculine pronouns, underlyingly
nea and
tuwa, and yields
nām and
twām, respectively: stress goes to the stem-final syllable, which becomes long. (Their singular forms, with stem-initial stress, are
ne and
tuw.)
Actually the singular surface form of the masculine pronoun warrants a comment. It's
tuw, and I'd rather not claim that there's a phonetic distinction between this and a hypothetical
tū---so this is a case where a phonetic long vowel does result from a sequence of phonemes.
There's another suprasegmental feature that works about the same way as [+long]: a word gets it if any of its components have it, and if a word has it, then its realised on the word's stressed syllable. For the moment I'll call this feature [+creak], because of the creaky voice it yields. That's to say, for example, the 2s pronoun is
wòe[+creak], which ends up as
wo̰ (with the grave accent marking stress); its plural form is
woèm[+creak,+long], or
wḛm.
Oh! A complication: if a word is marked both [+long] and as [+creak], then the result is just as if it were marked only [+creak]; that's why you get
wḛm rather than a hypothetical
wḛ̄m. (The vowel may actually end up phonetically long, I'm not sure about that yet.)
Now, both ejectives and glottalised nasals tend to cause creak at the edges of an adjacent vowel. I'm going to say that if a vowel is surrounded by such consonants and therefore has creak coming at it from both ends, it's phonetically indistinguishable from one that's itself genuinely [+creak]. So (as with
tuw and vowel length) you again get phoneme sequences duplicating a suprasegmental.
And then there's [+high] (referring to tone, not vowel height). It works somewhat differently.
With [+long], the feature gets realised on the stressed vowel, and if that's already [+long] then nothing changes. And similarly with [+creak].
[+high] will also go to the stressed vowel, unless either of two conditions obtains:
- If the stressed vowel is creaky as a result of adjacent glottalic or glottalised consonansts, then the high tone will move to the right until it finds a suitable host or it falls off the end of the word (don't worry, it'll float). As an example, suppose you had a word surfacing as mˀamˀbə, and it acquired a floating high tone. The two glottalised nasals would make the a phonetically creaky, and it would resist the high tone, which would then come to rest on the final vowel. The result would be something like mˀa̰mˀbə́ (using the acute accent for high tone).
- If the stressed vowel is already [+high], in which case we must be dealing with a compound form of some kind, then the new [+high] feature will lodge on the morpheme that sponsors it, if it has a suitable vowel, and otherwise it will float. The word will now have two [+high] syllables. If no glottalic or glottalised consonant intervenes, then the two will spread, giving every intervening vowel a high tone. So there's another affix that marks some pronouns as plural, -Esi[+long,+high]. If you add it to (underlying) wɨat, the 1+2 dual pronoun, you get watsɨ[+long,+high], which is to say wātsɨ[+high] or wa̋tsɨ (with the double acute accent representing a vowel that is both high and long).
So unlike [+creak], [+high] combines freely with [+long], and in the way you'd probably expect.
That leaves the question of how [+high] interacts with [+creak]. The idea I pulled out of my ass some time back was that a [+high, +creak] vowel gets realised as nasalised. That would mean, for example, that if the 2s pronoun
wo̰ somehow acquired a floating high tone, then it would end up as
wõ. (A nice visual pun: the high tone raising the tilde from below the vowel to above it.)
Except---that's stupid, right? (Right?) If it's cool or fun I might stick with it, but I'd rather not go with something stupid.
I do know that [+creak, +high] won't just be high tone on a creaky vowel, and I don't want the two features simply to cancel each other out. But so far other than the maybe stupid nasalisation idea I don't have any thoughts what to do.
(Orthographic issue: I seem to be beset with fonts that don't nicely distinguish tildes from macrons, so I may just stick with
wó̰ regardless, or use the ogonek to represent nasalisation.)
Anyway, those are the main points of Kwa̰ Mɨɨrts phonology, as it currently exists. Have at me.
Particular questions
There are some things I'd particularly like feedback on, if you happen to have any thoughts.
- [+creak, +high] → [+nasal]. Fun? Stupid? Intriguing? Any good alternatives? And, is there a happier name for the [+creak] feature than [+creak]?
- Leaving that aside, does the behaviour of [+high] make sense? I feel like it should interact with consonants somehow, but am not really sure how. (There are languages in which voiced consonants block high tone movement, not sure about anything closer to what I've got here.)
- I also wonder whether, if I go with [+creak, +high] → [+nasal], I should let nasal and nasalised consonants interact somehow with [+creak]. (Like, do they tend to raise the pitch of an adjacent vowel while also lending it some nasality?)
And specifically about orthography:
- I'm actually liking v̰́ for [+creak, +high] the more I look at it, but I also have a soft spot for the ogonek. Does one of those options, or something else, seem clearly better?
- Is there a way to use y instead of ɨ even though it looks terrible with diacritics below it?
- Relatedly, I'd rather avoid using the apostrophe for ejectives, since it's nice to have it around to mark nonobvious syllable boundaries. Maybe ˀ, like the glottalised nasals? (Would it make any sense to mark both with a tilde, preferably below but above where there are descenders?)
Anyway, thanks for reading, and for any suggestions you might have.