In this thread, I want to discuss problems I’m encountering in the development of Mesak. Said language is at this point pretty well-developed, having been worked on over the course of at least 1.5 years and having been tested in several relays. Nevertheless every now and then I keep encountering some unresolved issues. Nowadays these are mostly about the semantics or pragmatics, but some “more usual grammar stuff” also comes up sometimes.
Today I want to brainstorm a solution to a pretty subtle but deep problem: The anti-interplay between the absolutive pivot and incorporation. This’ll require some background knowledge, which I’ll summarize real quick as I do not expect anyone to have it.
The Absolutive Pivot in Mesak
Mesak, being highly ergative both in the morphology as well as the syntax, has a constraint on clause combinations: in general (but there are exceptions), two clauses can only be chained together if their Internal Person¹ is the same. As a rule of thumb, if this constraint can be fulfilled then it will; otherwise alternative strategies might be employed. Thus for example, the sentence “The man shot at the deer and the deer ran away” would be rendered in faux-mesak² as
Meanwhile, the sentence “The man shot at the deer and (the man) ran away” would end up asman-ERG deer-ABS 3EP-shoot-try-3IP and ran_away-3IP
with an antipassive to feed the absolutive pivot. Now if, say, a third party fled (say a bird that got startled by the noise) then the pivot constraint cannot be fulfilled. In this case, the conjunction and can still be used, but the absolutive NP must be overtly stated in both clauses, even if one of them was clear from context.man-ABS deer-ESS shoot-try-ATP-3IP and ran_away-3IP
This is however, while grammatical, somewhat frowned upon in actual use and a fluent speakers would probably split this into two clauses.man-ERG deer-ABS 3EP-shoot-try-3IP and bird-ABS fly_away-3IP
Now, keeping the internal person of the verbs the same is a syntactic requirement of many constructions (some, like the conjunction and allow for exceptions as mentioned, albeit questionable ones only; other constructions such as relative clauses don’t work at all without the constraint fulfilled - the IP of a relative clause must be the same as the head noun), but there’s also discourse considerations: over the course of multiple sentences, it is often attempted to keep the pivot the same. This can lead to several sentences in a row featuring an antipassive to move the topic (which is most commonly underlying A) to the pivot.
So for example, the following little “tale” has the same NP in the pivot every time, allowing it to be omitted from the sentences except for the verb-marking:
Tendencies of incorporation in general, and in Mesak specificallyher-husband-ERG woman-ABS 3EP-saw-3IP. at home was-3IP. meat-ESS prepare-ATP-3IP, for later eat-ATP-INF
“A man saw his wife. She was at home. She was preparing meat, for her to eat later.”
So I assume a few here to be familiar with the hierarchy of incorporation laid out by Marianne Mithun in some apparently quite famous paper that I personally haven’t read (but I’m skimming it for exact definitions and examples). A summary of it however goes more or less as follows: Incorporation occurs in four different types which are in an implicational hierarchy, that is if a language has type III incorporation, then it’ll also have types I and II. The types go as follows:
- Lexical compounding, especially of type N+V → V. This is stuff like Bob is berry-picking, which exists in English, but rather unproductively so. Type I incorporation is valence-reducing.
- Similar to type I, but here the valence is not decreased. Thus this is either incorporation of an oblique argument, or if a direct object is incorporated then a non-core argument is promoted to object. You’d expect to find stuff like Alice hair-combed Bob here. This is often done specifically to promote an animate argument to a higher position in the syntactic structure, and if a language does not have this type then it might be a marked structure to not do so.
- Backgrounding of known information via incorporation: In languages with this type, incorporation can happen as a means to background NPs which are already known information. We’re now in the area of discourse pragmatics. An example I found here from nahuatl goes something like You never eat meat. — I meat-eat all the time! In the second clause, “meat” is now an established thing and can be backgrounded via NI.
- Classificatory Incorporation: In the rarest type of incorporation, more general nouns are incorporated into the verb to classify the kind of action. An example given from Caddo uses eye as a classier for roundish objects, leading to a sentence of the form Plums are eye-growing. (which of course just means “Plums are growing”, don’t get nightmares please)
- Incorporation of nouns or adjectives into the copula or similarly behaving verbs. For copulas this is pretty much mandatory; it can be done on other transitive verbs (e.g. “become”) and this decreases valence to intranstive.
- Incorporation of the classifier noun into verbs when the IP is a mass noun. This has no effects on valence.
The Problem
If you compare the use cases of type III incorporation with the requirements of the absolutive pivot, then you might notice that they pretty much act against each other - the pivot demands the restated NP to be absolutive (and then allows omission thereof), while type III incorporation would rather have be incorporate it as known information. Obviously doing both at the same time doens’t really work. Some possible solutions are:
- Don’t care. The language can have types I, II and IV but not III. Languages do break implicational hierarchies at times and I feel like I’d have a good reason for it - type III incorporation is simply overshadowed by the more useful pivot.
- Care too much. I don’t really want the language to break such a strong pattern, so perhaps I should just remove the type IV incorporation. However, I like it a lot. It’s quirky and adds a bit of redundancy to the pretty underspecified third person.
- Find other applications for type III incorporation. The internal person of the verb is pretty well-handled by the pivot. But sentences often have other things as well. Perhaps, if there are other recurring nouns in the sentences, those could be incorporated.
Are there any other recurring NPs than the wife? Yes! The meat in fact also shows up in two clauses, and in the second one (the infinitive clause) is rather clumsily wiped under the carpet by an antipassive (it could be overtly included with essive case marking). Instead, it could absolutely be incorporated here! The incorporation would also reduce the valence in this case, as we’re incorporating from the position of a direct object; this wouldn’t necessarily always be the case. The important thing is that it’s being incorporated because it’s restating of old information and not simply to reduce the valence.her-husband-ERG woman-ABS 3EP-saw-3IP. at home was-3IP. meat-ESS prepare-ATP-3IP, for later eat-ATP-INF
“A man saw his wife. She was at home. She was preparing meat, for her to eat later.”
And I think I am content with this. One problem dealt with. Next up: derivational morphology.her-husband-ERG woman-ABS 3EP-saw-3IP. at home was-3IP. meat-ESS prepare-ATP-3IP, for later meat-eat-INF
“A man saw his wife. She was at home. She was preparing meat, for her to eat later.”
¹A term I’ll be using for phenomena that unify S and P, in contrast to A, which is associated with the External Person. These terms are taken from Sadock’s grammar on West Greenlandic and make descriptions quite a bit nicer.
²I don’t think adding actual glossed examples would add anything of value here. Non-relevant markers such as numbers or tense omitted from the gloss.