V-initial order, multiple adjuncts, and the in-situ rule

Apologies for another thread which will probably result it “just live with this ambiguity” :).

Consider (1):

(1) Идет кто куда?
Idet kto kuda?
goes who where
"So, WHO goes, WHERE?" [rus]

(I added intonation to indicate that this word order actually isn’t canonical. I just want to note in parentheses that in much literature, this word order will be dismissed as not possible but I think it is entirely possible, and accepting it has been an important assumption on my part…)

(Note also the comma in the translation. Perhaps such intonation indicates that there is some coordination going on there. In that case, I should not be after this tree…)

But! Assuming I care enough about (1), is there anything sensible I could do about the following ambiguity:

Screen Shot 2020-05-04 at 10.47.56 AM

Both S-S in the left tree are the in situ rule. Saying that the in situ mother is hasmod does not help because the head-adjunct rule (the one that attaches the ADV in both cases) of course doesn’t care about that; more than one adjunct should be allowed, by definition.

(Perhaps that’s an indication that these V-initial orders are weird and should not be included, after all.)

One possibility would be to block modification of constituents licensed by the in-situ rule. I’m not sure this is fully general (could something else combine with them and then that’s modified?) but you could maybe leverage the MODIFIED feature or something similar to rule out the tree on the left.

Yeah, I mean, there can be a modifier there, nothing wrong with saying Who sleeps in the forest? or something like that…

Based on my understanding of this thread and Ginzburg and Sag’s analysis, it wouldn’t make sense for anything to attach after the in-situ rule. For Who sleeps in the forest?, the modifier would presumably attach before the in-situ rule.

1 Like

Exactly – in order for my proposed fix to fail, there would have to be some reason that the modifiers must attach high with respect to in situ or there would have to be something else above in situ and then modifiers above that.

1 Like

Sounds good, I will try and report what/if something failed. I just think that I’ve tried it but perhaps I am mistaken…

Oh, so, it probably had to do with word order. In Russian, something like The character arrives where in that film? But I will try nonetheless.

To reiterate, it might be worth reconsidering the role of in situ wh-adjuncts, if they tend to create a lot of problems. This word order may well be special. There are some very common phrases which use it, e.g. You where worked last summer? – so common that I really want to cover that but despite the commonality, the intonation that comes with this WO is special, and it is of the sort where absolutely anything is possible in Russian, including the wonderful sentence from Antonenko 2008 that I showed to Emily the other day:

(2) Ты  доктор видела когда подъезжал?
    Ty  doctor videla kogda pod'ezzhal?
    You doctor saw    when was.arriving
"Did you see when the doctor arrived?" 

I don’t really know what is going on in (2) but I suspect that You there has almost no importance, while doctor and see are of great importance, and finally when arrived is somewhat important but in a different way than doctor and see (for example: two neighbors are concerned about their elderly neighbor who’s been sick, and they see the doctor’s car but the husband doesn’t know when the doctor arrived, was it recently or has the doctor been there long. Depending on that, he can establish how well the neighbor is doing, and so he asks his wife whether she actually saw the doctor arriving). And that somehow licenses this order. Idk. What’s particularly odd is the initial position of You, and one thing to add here is that You will be almost ``swallowed’’, will get almost no emphasis and will be uttered very quickly and with no pause after. Anyway. What I’m saying is perhaps there is a lot going on in such orders.

I am not sure how to do that. The head-adjunct rules, because they are general, do not have an opinion about their head daughter’s MODIFIED other than that it is the same on the non-head daughter. It is possible to use MODIFIED in the in-situ phrase and in the extracted-adjunct phrase, to block adjunct extraction from the in-situ rule, but I don’t see how to use if with the general head-adjunct, without blocking multiple adjuncts generally?

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.

It seems surprising to me that the head-adj rules would identify MODIFIED between the two daughters. Here’s what I see:

head-mod-phrase := head-nexus-phrase &
  [ SYNSEM [ LOCAL.CAT.VAL [ SUBJ #subj,
                             SPR #spr,
                             COMPS #comps ],
             MODIFIED hasmod ],
    HEAD-DTR.SYNSEM [ LOCAL.CAT.VAL [ SUBJ #subj,
                                      SPR #spr,
                                      COMPS #comps ],
                      NON-LOCAL [ REL 0-dlist ] ] ].

And:

basic-head-mod-phrase-simple := head-mod-phrase & binary-headed-phrase &
  [ SYNSEM [ LOCAL.CAT.MKG #mkg,
             NON-LOCAL [ SLASH [ LIST #first,
                                 LAST #last ],
                         REL 0-dlist ] ],
    HEAD-DTR.SYNSEM
           [ LOCAL [ CAT [ HEAD #head,
                           VAL #val,
                           POSTHEAD #ph,
                           MC #hmc,
                           HC-LIGHT - ],
                     CONT.HOOK #hdhook & [ ICONS-KEY.IARG1 #clause,
                                           CLAUSE-KEY #clause ],
                     AGR #agr ],
             NON-LOCAL #nonloc &
                   [ SLASH [ LIST #middle,
                             LAST #last ] ],
             LIGHT #light,
             MODIFIED #modif ],
    NON-HEAD-DTR.SYNSEM
           [ LOCAL [ CAT [ HEAD [ MOD < [ LOCAL local &
                                              [ CAT [ HEAD #head,
                                                      VAL #val,
                                                      POSTHEAD #ph,
                                                      MC #hmc ],
                                                AGR #agr,
                                                CONT.HOOK #hdhook ],
                                          NON-LOCAL #nonloc,
                                          LIGHT #light,
                                          MODIFIED #modif,
                                          OPT - ] > ],
                           VAL [ COMPS olist,
                                 SPR olist ],
                           MKG #mkg ],
                     CONT.HOOK [ ICONS-KEY.IARG1 #clause,
                     CLAUSE-KEY #clause ] ],
             NON-LOCAL [ SLASH [ LIST #first,
                                 LAST #middle ],
                         QUE 0-dlist & [ LIST null ] ] ],
    C-CONT [ RELS <! !>, ICONS <! !> ] ].

That says that the modifier daughter’s MOD’s MODIFIED is the same as the head daughter’s … so the modifiers get to have an opinion about what they modify.

You could probably add a feature inside xmod [ INSITU bool ] and then have the INSITU rule set that to plus while the modifiers look for [ INSITU - ]. Or elaborate the types under xmod. Or add a HEAD feature, for that matter…

Yes, that’s right! I misread the AVM earlier.