Ditransitive verbs and free word order; SYNSEM.ATTACH constraints?

I am planning to support ditransitive verbs in the context of the new wh-question library, so I have them in my Russian dev grammar (to clarify: that’s a grammar which at various points can outgrow the customization system, but at the end there is a version which the customization system fully supports).

I am modeling Russian as having free word order for this version of the grammar.

(1) Ivan        Mashe      dal    knigu
     Ivan.NOM . Masha.DAT  gave . book.ACC
    `Ivan gave a book to Masha.' [rus] 

Currently I can’t parse (1).

For example, I can’t form a VP using the HCR at the top and the C2HR at the bottom; it appears that certain orders are excluded by the SYNSEM.ATTACH constraints which come from the customization system (the word order library):


head-initial-head-nexus := head-initial &
    HEAD-DTR.SYNSEM.ATTACH notmod-or-lmod ].

head-final-head-nexus := head-final &
  [ SYNSEM.ATTACH rmod ].

head-comp-phrase-2 := basic-head-2nd-comp-phrase & head-initial-head-nexus.

comp-head-phrase-2 := basic-head-2nd-comp-phrase & head-final-head-nexus.

Furthermore, I can’t form a VP using the C2HR at the top and the HCR at the bottom because that I think is not the intended use of C2HR (the meaning of that rule is that the more oblique complement, such as the dative one, attaches before the accusative, and so the accusative is still on the list):


basic-head-2nd-comp-phrase := basic-head-comp-phrase &
  [ SYNSEM.LOCAL.CAT.VAL.COMPS < #firstcomp . #othercomps >,
                                          REST < #synsem . #othercomps > ],
    NON-HEAD-DTR.SYNSEM #synsem ].

So I suppose I need to be approaching this from the first angle, looking at the SYNSEM.ATTACH constraints, relaxing them somehow? But I don’t really know what is behind them.

; Three-valued sort for distinguishing unmodified signs from both
; left-modified and right-modified signs PERIPH indicates whether this
; modifier is left- or right-peripheral in its phrase - e.g., "the IBM
; temporary employees" but "*the IBM five employees"

xmod := sort &
  [ PERIPH luk ].
notmod-or-rmod := xmod.
notmod-or-lmod := xmod.
notmod := notmod-or-rmod & notmod-or-lmod.

hasmod := xmod.
lmod := hasmod & notmod-or-lmod.
rmod := hasmod & notmod-or-rmod.

This is really hard to work on remotely, I find, but here’s an attempt. The head-2nd-comp rules are indeed intended to pick up the 2nd thing on the COMPS list while the first is still unrealized. However, that doesn’t mean you necessarily need them in this sentence.

You say:

I can’t form a VP using the HCR at the top and the C2HR at the bottom;

By this, do you mean a structure like [[ Mashe dal ] knigu ]?

I can’t form a VP using the C2HR at the top and the HCR at the bottom

But this is completely unexpected. For the bracketing [ Mashe [ dal knigu ]], assuming knigu is the first thing on the comps list, I expect head-comp for the lower one and the comp-head for the higher. What does interactive unification show you there?

Yes. And that is expected, right, because of what head-2nd-comp rules were designed for?

Oh, I see. I was missing this part (that head-2nd-comp are not always used where there is the second complement).

What is in the way for the regular comp-head rule to apply is actually the INIT constraint which probably comes from the clausal complements library :).

So, does it sound like I need to have an extra Head-Comp rule for prepositions and complementizers? (I used to have it anyway, in my other versions of the free word order grammar; this one is a new iteration, started from the GM).

Remove the INIT constraint from the general HCR, but add an extra HCR for HEAD +rp. Constrain other rules to be HEAD +nv?..

It’s unclear to me why you need an extra HCR. Isn’t it enough to constrain the comp-head rule to be [ HEAD +nv ]?

1 Like

Oh yeah! duh. :slight_smile: