Verb complexes

Mirja verbs can be extremely complex, and verb complexes can include a significant number of potentially separate lexical roots. In effect, Mirja seems to be in a state where its previous multi-word verb complexes were analysed as single words, resulting in not only significant amounts of noun incorporation but also multimorphemic sequences that straddle the boundary between one verb with suffixes and multiple serialised verb roots. Some subsequent changes have taken place since that wholesale reanalysis, though; the most notable of which is that noun incorporation has been restricted to only generic / nonspecific objects and specific objects must be separate words. It's also the case that Mirja verb complexes can now only have one marking for most grammatical categories (which now comes after any incorporated nouns); something that may not have been the case previously.

Basic verbs

Despite the relatively high ceiling on verb complex complexity, the most simple permissible sentence in Mirja is a single verb root. There is no need for a verb to carry any inflection at all—there's no agreement morphology, and all other categories either have an unmarked default or are allowed to be fully unspecified.

e

e

eat

e

eat

'[subj] is eating [it]'

VERB INFLECTION TABLE ETC

Inverse marking

Serialisation

Mirja verb complexes can contain a number of apparently separate verb roots, or things strongly resembling them, within a single word.

simallhanakovijeta

sima-llha-nako-vije-t

drive.car-to-be.able-easily-PFV

sima-llha-nako-vije-t

drive.car-to-be.able-easily-PFV

'was easily able to drive to [it]'

Here, -llha behaves mostly as an applicative and is less clearly an independent verb, though it seems to be derived from one—it can take 'its own' incorporated objects independent of the head verb:

simallhamyljata

sima-llha-mylja-t

drive.car-to-store-PFV

sima-llha-mylja-t

drive.car-to-store-PFV

'drove to the store'

-vije is one of a class of manner-providing 'roots' that never head their own verb complexes, but are difficult to distinguish from verbs that simply never appear outside of serialisation. Nako is in a class that is the most clearly separate, as while it provides an apparently very grammatical potential mood meaning, it can appear on its own in short exchanges when the main action under discussion is already highly discourse-active, such as nakowwe 'can they?' nako 'they can' and similar.

There are effectively four categories of 'verb roots' in Mirja, based on their behaviour within verb complexes. (It's up in the air how hard these boundaries are at the moment.)

• Main verbs: These are verb roots that usually head verb complexes.
• Applicatives: These add oblique-like objects, and are the most likely to have their own separate incorporated objects. They're also the most phonologically reduced, and often have forms that would not be valid root shapes.
• Other grammatical-property-like elements: These include things like 'be able', causation, numerals, and so forth. These are the most likely to appear independently when the action of the verb they would otherwise follow is already highly discourse-active.
• Manners: These add information about the manner the head verb is done in. They're the least apparently grammaticalised (besides main verbs), but rarely (if ever?) appear independently.

Mirja verb complexes are different from prototypical serialisation situations not only due to all being one phonological word, but also because the roots that usually head them rarely appear as non-heads inside complexes headed by other verbs in modern Mirja. This makes the serialisation analysis a bit more questionable—is it really serialisation if you can't take two clearly independent roots and serialise them? Probably the best way to analyse this is that Mirja is in the process of converting all of its serialised roots to simple bound morphology—it's at the point of closing off the class of 'serialisable roots' but still has a very large such class and still allows certain roots to have their own objects independent of the head. I still think the serialisation analysis is the most helpful way to think about it even synchronically, but future development may force the bound morphology analysis eventually.

Incorporation

Generic or nonspecific objects can be incorporated into verb complexes. Unlike independent objects, these incorporated nouns come after their head verb root, which suggests that pre-Mirja had a primarily SVO main clause order before reanalysing its verb complexes as single words.

iimata

e-ima-t

eat-morning-PFV

e-ima-t

eat-morning-PFV

'ate (some) breakfast'

Non-incorporated objects imply a specific reading:

ima eta

ima

morning

e-t

eat-PFV

ima e-t

morning eat-PFV

'ate the breakfast'

Serialised verb roots and similar things may have their own incorporated objects:

iimaheeteta

e-ima-he-ete-t

eat-morning-at-house-PFV

e-ima-he-ete-t

eat-morning-at-house-PFV

'ate (some) breakfast at home'