Subject and Object Omission in Ŋʒädär

Despite the lack of person congruence on the finite verb, Ŋʒädär permits some amount of subject and object omission. Generally, the most recent subject and the most recent object will carry over.

Example 1: NP₁ NP₂ V₁. V_trans.

The same NPs will be considered subjects and objects of the next verb, and the next after that and so on until new explicit argumenst are present. If it is intransitive, the previous subject generally will carry over, except for a handful of verbs for which the object will be considered the subject. These generally encode reactions to the first verb, and there are some lexically determined things going on there. 

Example 2: NP₁ NP₂ V₁. NP₃ V_dir.
Example 3: NP₁ NP₂ V₁. NP₃ V_inv. 

Although the subject of V₁ is whichever of the two NPs whose ranking is preferred by the direct/inverse marking, the first noun phrase syntactically has some subject-like properties – it will carry over to the next verb - unless NP₃ and NP₁ are co-referent, in which case some other noun takes precedence - a previous subject or the other noun of the previous verb phrase, or the discourse topic. This is somewhat ambiguous, and depends on the nature of the verbs and the involved nouns.

If more than one third person argument is present, however, one will have to be marked for obviativity, which simplifies things a fair bit - the omitted argument will either be obviative or not, which will affect the marking of the next verb.

The Ŋʒädär branch lost a lot of the verb complex during its development out of Proto-Ćwarmin-Ŋʒädär. Most direct-inverse languages seem to have pretty rich verb complexes, with person congruence almost omnipresent in this typological class. Therefore, Ŋʒädär is a bit of an outlier - in fact, I'd dare say it's typologically unlikely altogether. The Dagurib branch is more typical of direct-inverse languages.