With the current valence change library, is it possible to model something like this Chukchi sentence:
(1) inen- req -ekwʔi
CAUS do.what -MYSTERY
`What are you doing to me?' [ckt]
I cannot decipher the last suffix yet, unfortunately, but suppose even we could decide on how to gloss it in a meaningful way :). Suppose it means person/number information.
My question is, is there a way currently to model this, or will it not work if I don’t have overt arguments? I couldn’t quite figure it out. The verb has two predications in it inherently and does not inherit from either transitive or intransitive lex-item (it doesn’t make sense for this verb).
Here’s what I did and with this I can’t analyze the word lexically. But maybe there is a way of doing it? @curtosis
I haven’t looked into the valence changing library in particular, but it appears as if your valence changing rule is constraining its input to be intransitive but you’re saying you haven’t marked your verb as intransitive, so I imagine the rule wouldn’t be licensed and subsequently the word wouldn’t be licensed because it’s not
infl-satisfied, leading to the behavior you describe.
Whether or not the arguments are overt is a completely separate question. What do you want the ARG-ST to be for the stem glossed as do.what, without the CAUS prefix? What change do you expect the CAUS prefix to make to the ARG-ST?
Then there’s the separate question of whether the valence change library has specific verb types hardcoded to go with “intransitive” and “transitive”, but I rather doubt that it does, or if it does, it’s a verb type your type also must inherit from…
It has a normal SUBJ element and an empty COMPS list. It has three preds however, one for do and two for which thing.
That’s what I am not sure about! In the examples for valence change, things like X lighted the fire turn into Y made X light the fire.
But here it is a bit different: the question is about what was done by 2SG to 1SG. So, normally 2SG would be doing something, the question would be what is it that 2SG is doing. With the CAUS suffix, 2SG is doing something to me, and the question is still about which thing that is. In these terms, this is more like a ditransitive verb, then, in which case I don’t model it via the Matrix.
Parr of my confusion I think came from the gloss, so let me now regloss it (based on Ivan Stenin’s opinion):
(1) ine- n- req -ekwʔi
2>1- CAUS do.what -THEMATIC
`What are you doing to me?' [ckt]
(The last prefix may be semantically empty(?). But the pernum information must come from the first prefix.)
I think this may be modeled by the direct-inverse library but I don’t know how, I only have this one example for now, of this sort.
I am still not sure whether I need the valence change (I thought so because of the CAUS gloss).
Right, and for this particular verb, the COMPS list is empty. Does it mean I need to choose “intransitive target”?
In theory, yes, but with your 2>1 prefix it’d be a little more complicated.
However, I think this is more like an object-adding affix, going from 2SG does X to 2SG does X to 1SG. “Do to Y” is more or less how I’d read a generic applicative.
To model it you might look at first applying the applicative to the Ø-COMPS verb to get the to-whom slot, then another applicative rule to add the do-what slot (chain the rules on the affix, in other words). The 2>1 prefix should then be able to constrain pernum on the right elements (Lakota uses pronominal affixes for everything non-3SG, for example.)
Thank you Chris!
Here’s what I managed to get (btw I did not need two lexical rules because this verb inherently has the second predication; that’s a new type I added recently).