I think there is a lot of confusion regarding cases :-) I'd rather remind once for all what are the 3 linguistically proven options (A) and explain again the solution (B) to make Carlos, Christophe and Pablo happy, and at least not shock Herman and Nik too much :

A. There are only 3 pure, therefore ideal, case systems in the world :

1. Nominative/accusative system (NA)
Pure verb-stemming predicate. Predicate is a process. Is axed on humanized experiment of process :
I lost wallet =
nom-me  acc-wallet to-lose
I see wallet =
nom-me acc-wallet to-see
I run =
nom-me to-run
I fall =
nom-me to-fall
I raise the hand =
nom-me acc-hand to-make-rise
Unlike agent/patient and erg/abs languages, nom/acc languages incorporate whithin the verb causative tags. This is called 'applicative' insert.

2. Agent-Patient system (AP)
Tends towards noun-stemming predicate. Predicate is usually the noun state, instrument or result of process. Is axed on control over process OR attribution of instrument or result  :
I lost wallet =
patient-me agent-wallet loss/lost-thing
I see wallet =
patient-me agent-wallet image
or : agent-me patient-wallet eye/vision
I go =
patient-me way
or : agent-me going
I run =
agent-me running
I fall =
patient-me falling
I raise my hand =
causative-me agent-hand rising/rising-one
or : agent-me patient-hand rising/rising-one
both are possible depending on whether 'to rise' is considered either a basically autonomous move or else a prompted move in the language studied.

3. Ergative/absolutive system (EA)
Halfway between 1. and 2. Tends from noun- towards verb-stemming predicate. Predicate is therefore stative, i.e. it is a stative verb experienced by the noun being (like 'the content') or featuring (like 'the wounded') the result of process. Axed on proximality :
I lost wallet =
ergative-me absolutive-wallet to-be-lost
or else : unergative-me absolutive-wallet to-be-lost
I see wallet =
ergative-me absolutive-wallet to-be-seen
or : unergative-me absolutive-wallet to-be-seen
both are possible depending whether the language cares more about cause or rather about control (i.e. whether it tends towards nom/acc or towards agent/patient)
I run =
abs-me to-run
I fall =
abs-me to-fall
I raise my hand =
erg-me abs-hand to-rise

Of course no language is exactly NA, AP or EA mainly because of the causative issue I solve below in 3 lines.
Since Carlos would prefer noun-rooting predicate (as I do) , this language very naturally tends towards AP.Then 'to be red' could derive either from a passive state ('the redness') or a result ('the red colour'), or an instrument ('the red paint') while 'to loose' could derive from either an active result ('the lost thing'), a passive result ('the loss'), a direct passive state ('the fact of being lost') or an indirect passive state ('the fact of having lost something'). So how choose the right root ?
We need a RULE like 'root = 1.result; if no 1. then 2.instrument; if no 2 then 3. state; if no 3 then... etc.

I suggest :

Trend towards noun-rooting predicate, so predicate is usually the tangible result or instrument so case (1) would be the AGENT who would be the one either being the result of process (1.1) or the instrument thereof (1.2).

When noun-rooting predicate above not possible, then state-rooting predicate.

ABSOLUTIVE case (2) (=former 'undergoer') would be the one either featuring the active result (2.1), or experiencing the instrument (2.2) or experiencing that state (2.3).

Regardless whether the root-predicate is either an instrument- or result-noun or else a state, the ERGATIVE (3) is the indirect cause of result (3.1), the user of instrument on someone else (3.2) or the direct cause of state (3.3).

PATIENT (4) is the one featuring the passive result (4.1), or subjected to the instrument (4.2) or suffering the state (4.3).


1.1. age-apple fruit (apple grows)
1.2. age-house shelter (house shelters)
2.1. abs-tree fruit (tree grows fruit)
2.2. abs-man seat (man sits)
2.3. abs-man dead/death (man dies)
3.1. erg-man wound (man wounds)
3.2. erg-man hammer (man hammers)
3.3. erg-man rise (man raise)
4.1. pat-man wound (man is wounded)
4.2. pat-man hammer (man is hammered)
4.2. pat-man dead/death (man is killed)

TO SUM UP we would have :

(i) 4 base-cases

1. Agentive
2. Absolutive
3. Ergative
4. Patient

(ii) with a trend from result (bite) towards state (bitten/biting),

(iii) from tangible result (wound) towards intangible result (image) : example : if 'bite' is a tangible mark, then 'kjak' is a result and 'to bite' result-rooted; if 'bite' is no tangible mark, then 'kjak' could be either an intangible result, a passive state ('to be bitten') or an active state ('to bite') : we have to decide on a case-by-case basis.

(iv) from passive state ('to be bitten') towards active state ('to bite').

I suggest we call the system STM (St Thomas' Meathod).

(v) Then we also need a causative-factitive case 'to have/make someone do' : you can then put the negative before the case and spare MUCH trouble on designing factitive :
I have him raise the hand =
caus-me erg-he abs-hand rising
I don't have him raise the hand =
not-caus-me erg-he abs-hand rising
I have him not raise the hand =
caus-me erg-he abs-hand not-rising
I have him raise the hand but he doesn't =
caus-me not-erg-he abs-hand rising


See the original message at
Free e-mail group hosting at