Interpretation of _on_p_state for "drop the diamond on the table"

The first parse of "drop the diamond on the table" has the below tree as a solution. I’m struggling to interpret _on_p_state. Since it is modifying the event introduced by drop_v_cause, I am assuming it is modifying that verb.

Thus, does it mean: “drop the diamond, and do the dropping while you are on the table”?

                    ┌_table_n_1:x14
  _the_q:x14,h16,h17┤
                    │                  ┌pron:x3
                    └pronoun_q:x3,h5,h6┤
                                       │                 ┌_diamond_n_1:x8
                                       └_the_q:x8,h10,h11┤
                                                         │   ┌_on_p_state:e13,e2,x14
                                                         └and┤
                                                             └_drop_v_cause:e2,x3,x8

Logic: _the_q(x14, _table_n_1(x14), pronoun_q(x3, pron(x3), _the_q(x8, _diamond_n_1(x8), and(_on_p_state(e13, e2, x14), _drop_v_cause(e2, x3, x8)))))

On a related note: So far, I’ve been treating prepositions with *_p_state as synonyms for the same preposition with *_p_dir at least in the cases where they have the same argument signature. e.g.:

_to_p_state(e, e, x) and _to_p_dir(e, e, x)
_through_p_dir(e, e, x) and _through_p_state(e, e, x)

Is there somewhere that describes the difference?

If you go to the http://erg.delph-in.net/logon and paste the sentence, the second analysis is what you want, right? The PP attachment was wrong in the case you described above.

Yes, that top-ranked parse has the reading you understood it to have, and the one you likely want, using the entry for “drop” that is like “put”, is sadly ranked lower.

I don’t see the description we should have on the ERS pages explaining the difference between stative and directional preposition senses. In brief, a locative preposition such as “under” can have either sense, illustrated in the two readings of “The mouse ran under the table” meaning either “the mouse ran around while it remained in the region of the floor covered by the table” (stative) or “the mouse ran from some point away from the table to a point that was covered by the table” (directional).

2 Likes

@Dan, very clear, thanks. That explains why my mapping _to_p_state to be the same as _to_p_dir has worked so well for me. I’ve never wanted the stative interpretation so far and this has masked it when ACE has produced it. I’ll have to fix that and see how things work.

@arademaker: Short answer: I was hoping for the interpretation “drop the diamond and make sure it drops on the table”.

Longer answer:

I’m confused by the output of http://erg.delph-in.net/logon because it just uses _on_p without any suffix at all (I expected _state or _loc).

_on_p also doesn’t show up as an option in surface.smi which I thought was exhaustive…

I also think I don’t understand what that page is showing me since it shows a maximum of two MRS outputs for the sentence (which are identical), while ACE gives me 8, even though I said show=all.

@Dan , where are the “ERS pages”? Just want to make sure I’m not missing a resource.

1 Like

OK, one more clarifying question on stative and directional prepositional phrases: I’m having trouble mapping your example of how to interpret stative and directional under to the preposition to.

“exit to Plage” generates only stative options like:
proper_q(x10, named(Plage, x10), pronoun_q(x3, pron(x3), and(_to_p_state(e9, e2, x10), _exit_v_1(e2, x3, i8))))

I have the same confusion about through_p_state.

It is easy to imagine the state of something being under a thing while doing something, but hard to imaging the state of something being to or through a thing while doing something…

Is to/through an example where the stative and the directional really are the same because of their nature?

Holding off for a second on whether “to” and “through” can ever be other than directional, the problem with the verb “exit” is that it is currently lacking an entry of the type v_pp_dir_le, for verbs of motion that select for a directional PP (and specialize the PRED name to a directional one). The missing entry is like this:
exit_v3 := v_pp_dir_le &
[ ORTH < “exit” >,
SYNSEM [ LKEYS.KEYREL.PRED “_exit_v_1_rel”,
PHON.ONSET voc ] ].
If you add that entry to your ERG’s lexicon.tdl file, you should see the directional sense for “to”.

As for whether “to” and/or “through” can ever be non-directional, maybe not as VP modifiers, but these prepositions head phrases that can also modify nouns, and in “the letter to Sandy” or “the tunnel through the mountain” it seems reasonable to view the preposition’s sense as non-directional.

Dan

1 Like

@EricZinda I believe @Dan is referring to this documentation: http://moin.delph-in.net/ErgSemantics

Thanks @Dan! I’ll enter an issue here as well for this.