I re-watched Jessica Coon’s presentation entitled Rethinking Aspectually Based Split Ergativity. It inspired some changes for Umu.

I learned that Mayan languages have two person-marking morpheme sets, traditionally labeled ‘Set A’ and ‘Set B’. Set A is used for ergatives and possessives—which are syncretic. Set B is used for absolutives.

I made Umu with vowel harmony. All words have two variations, based on their vowels. Words switch from one form to the other. Until now, I used this vowel shift to signal possession. But lets Mayan it up a bit.

Review

Vowel Harmony
Light vowels: a, o
Dark vowels: e, u
Neutral vowels: i, ö

By default, Umu words end in a, u, i and ö. The vowels i and ö are neutral. They don’t change and can be in either light words or dark words.

  • Words ending in a can only contain a, o, and i and ö.
  • Words ending in u can only contain e, u, and i and ö.
  • By default, words ending in i and ö can only contain an a or a u—but not both—and other neutral vowels.

All of this applies to a word’s default form. When a word undergoes vowel-shift:

  • The final a becomes e and any o becomes u.
  • The final u becomes o and any e becomes a.
  • Neutral vowels stay the same.

Basically light words become dark words and versa vice. It’s nothing new. I did it to squeeze more than four vowels into the language but didn’t really know how to use this feature. In the end, I settled for possession.

image
kama
waterimage
‘eru
childimage
mari
mother

image
iv (jövö)
fire

image
keme
water’s

image
‘aro
child’s

image
meri
mother’s

image
iv (still jövö)
fire’s

The examples above cover all the bases. As you can see, words with only neural vowels don’t change at all expert in writing. And that’s were we left off.

What’s New

If Mayan can squeeze ergatives in with possessors so can I. Henceforth, the subjects of transitive verbs will take the genitive a/k/a, d/b/a the ergative.

Unlike Mayan, where ergativity comes up as person markings on the verb, Umu will use straight up word case. …I think.

So before, where we’d have a sentence like:

image
image
image
image
Vi tja’eru mari.
vöjö töja ‘eru mari
PRFV bathe child mother.
The mother bathed the child.

We get:

image
image
image
image
Vi tja’eru meri.
vöjö töja ‘eru meri
PRFV bathe child GEN/mother.
The mother bathed the child.

Why? Well for one, now I can front the topic and have an indefinite agent. Remember that in Umu, preverbal arguments are indefinite:

image
image
image
image
Meri vi tja’eru.
meri vöjö töja ‘eru
GEN/mother PRFV bathe child.
A mother bathed the child

That’s cool. Additionally, it’s confounding. Reason enough to do it all the time. Though not all the time…

How Bout Aspect?

You might be wondering why I started out on aspect. If you’re a real nerd you already know. …Or you’re good with context clues. …Or you watched the video.

Well. It’s because ergative systems aren’t ergative all the time. They split along various illustrious factors such as person, animacy, lexical category, source, control, tense, aspect, and other words as well.

But with Mayan, and now Umu, the split involves aspect. In fact this whole mess started with my trying on aspect morphemes—some of which I shall introduce to you now.

wpid-wp-1411880308071.jpeg
vi
already completed

wpid-picsart_1411791740608.jpg
tam
just completed

wpid-picsart_1411690332483.jpg
zaj
in progress

wpid-picsart_1411792936109.jpg
hara
not yet completed

wpid-wp-1411793827540.png
kru
unreal or not yet existing

Nice huh? The punchline is Umu will use an ergative-absolutive pattern with aspect-free clauses and the perfective aspects, and a nominative-accusative pattern with nonperfective aspects.

  • ERG/ABS: no aspect morpheme, vi and tam
  • NOM/ACC: zaj, hara, kru

Perfectives & Aspectless Clauses

Here we see a tidy ergative-absolutive alignment. With the addition of case making we now find unambiguous infinite agents and patients. Though some of these forms are rarely used.

The aspect morpheme always precede the verb, regardless of where the arguments are.

image image image
Tja’eru meri.
töja ‘eru meri
bathe child GEN/mother.
The mother baths/bathed the child.

image image image image
‘Eru vi tja meri.
‘eru vöjö töja meri
child PERF bathe GEN/mother.
The mother bathed a child.

image image image image
Meri tam tja’eru.
meri tamö töja ‘eru
GEN/mother JUST.COMPLETED bathe child.
The child was just bathed by a mother.

image image image
Meri ‘erutja.
meri ‘eru töja
GEN/mother child bathe.
A child is bathed by a mother.

image image image image
‘Eru meri tam tja.
‘eru meri tamö töja
child GEN/mother JUST.COMPLETED bathe.
A mother just bathed a child.

More topical (older) information goes nearer the end. Newer, contrastive or unexpected info appears closer to the beginning.

In intransitive constructions, the sole argument remains in the absolutive—not the genitive/ergative.

image image image
Vi tja’eru.
vöjö töja ‘eru
PERF bathe child.
The child bathed.

image image image
Tam tjamari.
tamö töja mari
JUST.COMPLETED bathe mother.
Mom just took a bath.

This gives rise to a new, interesting ambiguity.

Remember. Arguments can be covert/dropped if understood, or for reasons stylistic or poetic. Compare the following sentences. The covert argument is marked in gray and glossed in parentheses.

image
image
image
image
Vi tja’eru (meri).
vöjö töja ‘eru (meri)
PREF bathe child (GEN/mother).
(The mother) bathed the child.

image
image
image
Vi tja’eru.
vöjö töja ‘eru
PREF bathe child.
The child bathed (itself).

This difference between agent and experiencer would have been ambiguous without the new case marking. In the first example, child is the patient of the covert mother’s action. The sentence could answer a question like, “What’s mother doing?”

In the second, the child experiences the intransitive action, answering questions like, “What’s that kid doing?” or “Who’s taking a bath?”.

The ambiguity is fun and easily navigable within its context. In both instances, we ultimately end up with a clean kid—which satisfies.

But it gets better:

image image image
Vi tja (‘eru).
PERFN bathe child
(The child) bathed.

image image image image
Vi tja (‘eru) (meri).
PERF bathe (child) (GEN/mother).
(The mother) bathed (the child).

Both here can be the affirmative response to similar questions.

Did the kid take a bath?
—Bathed (yes)

Did mom bathe the child?
—Bathed (yes)

In fact the shortest answer to any of it is:

image image
Vi.
PERF.
Yes/it happened/it’s done.

Or for no:

image image
Mna vi.
No PERF.
No/its not done.

That’s how to answer yes/no questions in Umu. A bare verb, a bare aspect morpheme, or both together for affirmatives. A negated verb, negated aspect marker, or a negated aspect marker and verb for negatives. But never just the negative mna on its own. We can’t get too lazy now.

Careful what you’re negating. Aspect markets are safer. Bare verbs can mean many things.

image image
Mna tja
No bathe.
Don’t bathe! OR
(She) didn’t bathe. OR
(It’s) unbathed.

Nonperfective Clauses

Here’s where we get crazy.

Going back to Jessica Coon, I learned that in the Mayan language Chol—and many ergative languages—nonperfective clauses show a nominative-accusative pattern. In Chol, this means that the subject of transitive verbs use Set A (ergative/possessive) morphemes like before. But now the subject of intransitive verbs also use Set A—where as before in perfective clauses they use Set B.

Why?

That’s what everyone’s trying to understand. Coon suspects something very sneaky. That in nonperfective clauses, the aspect morpheme isn’t a helping verb but the main verb itself! This main verb takes a single, intransitive argument which is a formally possessed nominalized predicate.

So instead of saying, say, “The mother is bathing” and interpreting mother as the agent a/k/a the subject a/k/a nominative, what’s really being said is something like, “The mother’s bathing is happening” where mother’s-bathing is now the experiencer. It looks like a nominative-accusative pattern because the Set A morphemes are used for both ergatives and possessives. But really it’s just the (possessed) absolutive argument of an intensive verb. So the split isn’t a split at all but rather just possession in disguise.

If your head’s in a spin, watch Coon explain it. My powers and knowledge have very real and hard limits. It took me a few goes to unpack everything myself. But it is in truth a thing of beauty. Wait till you see how I fuck it up.

Umu Nonperfectives

Umu and I mangle everything we touch. And sadly, years of gifted and accumulated academic research is not beyond our grasp. But happily for Coon, in my world her hypotheses is demonstrably correct. It’s a simple, beautiful world I live in.

Recall the aspect morphemes zaj, hara and kru. How lovely they sound to our ears and our minds. These are our nonperfective morphemes which means they let us know the action under discussion isn’t done.

As before, I’ll give you some examples and I’ll dance some arguments around the verb and we’ll see if you notice anything different.

image image image image
Zaj hapame’uwre.
zajö hapa me’u ‘öre
PROG punch shark GEN/man.
The man is punching the shark.

image image image image
Mna hara nnama’o.
möna hara nöna ma’o
NEG NOT.COMPLETE eat GEN/shark.
The shark hasn’t eaten yet.

image image image
Kru zanöre.
köru zanö ‘öre
will stand GEN/man.
The man will win.

You’ll first notice that now the sole arguments of intensives take the genitive case, compared to perfectives:

image image image
Vi zanöra.
vöjö zanö ‘öra
PERF stand man.
The man won.

This is the main difference. But it gets crazier.

The “subjects” of these verbs are actually possessing what we assume is the main verb—or the main verb and its object in the case of transitives. These possessed verbs and their victims are in fact the sole augments of the aspect morpheme, the real main verb of the sentence.

Well so what, you say. Well it’s important because of word order. Remember all that dancing around with definites and indefinites? That’s gonna be different here. Let’s start with the intransitives.

image image image image
Nnama’o mna hara.
nöna ma’o möna hara
eat GEN/shark NEG NOT.COMPLETE.
A shark hasn’t eaten yet.

The shark owns the eating, so they’re stuck together. Nna ma’o. If post verbal nna ma’o means the shark’s eating, then pre verbal nna ma’o means a shark’s eating.

This happens because the aspect marker hara is actually the main verb of the sentence. Contrast this with vi—an aspectual morpheme that isn’t the main verb:

image image image image
Me’u mna vi nna.
shark NEG PERF eat.
A shark hasn’t eaten.

The two sentences in every way are identical in meaning except the aspect. But their word order is very different. You might be thinking, “When would I need to say that?” Remember that Umu doesn’t like plurals that much so a more logical translation might be indefinite sharks or some sharks. So be careful because some sharks haven’t eaten yet.

Now for the man:

image image image
Zanöre kru.
zanö ‘öre köru
stand GEN/man will
A man will win.

To stand is an idiom for to win. Vuj li kru tan zan öré kru. There will come a day when a man will stand up. (Sometimes after consonants I spell wre as öré. They sound the same.) Compare this to a perfective. Again we see different word order and different case marking.

image image image
Wra tam zan.
man JUST.COMPLETED stand.
A man just stood up.

Now for the transitive. Let’s recopy the first example.—possibly Umu’s the default word:

image image image image
Zaj hapame’uwre.
zajö hapa me’u ‘öre
PROG punch shark GEN/man.
The man is punching the shark.

The man owns the punching and the shark. Post-punching sharks are definite: hapa me’u: the shark punching. Preverbally they’re indefinite: me’u hapa: a shark punching. Both are followed by their owner GEN/man. Hapame’uwre: the shark punching of the man. Me’uhapawre: a shark punching of the man. To make him indefinite, take the man and all his friends and stick him before the main verb, in this case the aspect morpheme zaj

Advertisements