agreement and pronouns

[Mark Donohue]

Just a few more comments on agreement and gender.

Kanum, a New Guinea language, has similar issues in its grammar as do the 
Amazonian languages that Dan Everett has brought up. Basing a response on 
my familiarity with Kanum grammar, I have a few qualms with treating 
agreement on the verb as pronominal. The problem is this:

Kanum absolutive pronouns distinguish the following categories:

1SG			ngka^
1PL			ny
2SG/2PL			mpw
3SG/3PL			py

yet the pattern of object prefixes on the verb is as follows (in today's
past tense):

1SG/3SG.FEM		w-
2SG/1PL			n-
3SG.NON-FEM, 2PL, 3PL	y-
3SG.FEM			t-  (some conjugations)

The subject suffixes in that same tense are:

1/2/3SG			-i
1/2/3PL			-ns


These affixes are invariant for the position of the argument, left or 
right of the verb, or absent.

(and to be fair, I should mention the ergative pronominal paradigm:

1SG			ngkay
2SG			mpay
3SG			pyengkw
1PL			nynta
2PL			mpwnta
3PL			pynta

So they do differentiate all 6 person/number combinations somewhere)

It seems unlikely to me that the verbal affixes alone can be treated as 
pronominal representatives in this case: not only do they underrepresent 
the features associated with the free pronouns, but they also introduce 
new  overt features not found elsewhere (e.g., feminine/non-feminine in 
3SG).

Is it reasonable to assume that these are pronominal representations? We 
could always argue for a fully specified paradigm everywhere, such as 

Revised Absolutive paradigm:	PERS	GENDER	NUM	
1SG			ngka^	1	-	SG
2SG			mpw	2	-	SG
3SG.FEM			py	3	fem	SG
3SG.NON-FEM		py	3	nfem	SG
1PL			ny	1	-	PL
2PL			mpw	2	-	PL
3PL			py	3	-	PL
 
but then we lose a lot of economy of the process; it seems better to 
assume that each occurrence of, for example, py is not specified for 
gender. Unlike Dan's data, the gender appears on Kanum verbs only for 
objects; both transitive and intransitive subjects take verbal suffixes, 
with no possibility of gender being marked.

Oh, P.S., there's also possessor ascension in Kanum. An example paradigm:

mpw-ne  swa   pyengkw  y-erm-y
2SG-DAT	hand  3SG-ERG  3SG.NON-FEM.OBJ-shoot-SG.SUBJ.TODAYSPAST
'He shot your hand.'
(male/female addressee)

mpw-ne  swa   pyengkw  n-erm-y
2SG-DAT	hand  3SG-ERG  2SG..OBJ-shoot-SG.SUBJ.TODAYSPAST
'He shot your hand.'
(male/female addressee)

mpw-ne  swa   pyengkw  t-erm-y
2SG-DAT	hand  3SG-ERG  3SG.FEM.OBJ-shoot-SG.SUBJ.TODAYSPAST
'He shot your hand.'
(female addressee)

This clearly shows a gender feature shared from the possessor of the NP 
to the verb; yet there is not any overt indication of gender on mpwne. 
Further, the optionality of possessor ascension would indicate that it is 
sensitive to topicality, despite there not being an overt topic position 
on the verb.

Well, I won't stay on forever. 

Mark Donohue