Short Sketch #5: Definiteness marking and definiteness types

Definiteness is a thing some languages do not grammaticalize to the extent English does, and on the other hand, some languages have it even more grammaticalized - see, e.g. how it influences both adjective and noun inflection in Swedish, how it affects verbs in Hungarian, and so on.

Let us try out a slightly more complex thing. Some languages have three levels of definiteness, e.g. known to listener, known to speaker, and indefinite. 

Example, based on sample sentences from some pidgin, I don't recall the source where I saw these:

1. I am looking for a car → any car will do.
2. I am looking for one car → there is a specific car, but the listener is not expected to know which.
3. I am looking for the car → the listener too knows which car you are looking for.

(Also notice how "this" is acquiring a similar usage in colloquial English, c.f. any google result for "so this girl|dude|chick|guy|... walks|says|..." This is clearly not a deictic use of this, but rather a pragmatic use where it marks 'there is a specific girl that did it, but you probably don't know who she is, but I might refer to her in the future so she's significant enough for the narrative.)

Now, it seems rather natural to me that different constituents of a sentence will be differently likely to be marked for different types of definiteness, and this could easily lead to different kinds of arguments having very different marking strategies as well as a different number of distinctions.

An obvious example are languages where the object can be marked for definiteness by differential object marking, but other constituents have nothing of the sort. Usually, there is good likelihood that the subject is definite - even more so for the topic, but as topics and subjects often coincide, well, there you go.

A draft of this kind of a system would be something like:

• definite non-subject (2 or 3) topics trigger extra auxiliary verb that only carries congruence and has no semantic content

• definite subjects (2 or 3) trigger congruence in the verb (alternatively, see short sketch #1 for another strategy that could be used here)

• definite objects (3) trigger congruence in the verb and accusative case
• definite objects (2) trigger accusative case
• indefinite objects (1) take neither

• other definite (3) non-topic nouns trigger congruence on adpositions
• other non-topic nouns (that is, (2 or 1) functioning by virtue of case marking as an argument of the verb) do not trigger any congruence strategies
• different definiteness correlates with different case systems, something along these lines:
• 3:     nom - acc - dat - lat - abl - part - gen - instr
• 2:     nom - acc - dat        - abl          - gen
• 1:     nom                  - lat -         abl
• All topics, however, no matter definiteness, use the full (3) case system
• The lative for indefinitive nouns is identical in form to the dative for definite (2/3) nouns.

Other parameters could affect different systems: maybe animates have a different number of levels of definiteness from inanimates or whatever, ... however, something along these lines seems fairly intricate and interesting imho.