-
Notifications
You must be signed in to change notification settings - Fork 4
ErgSemantics_Ellipsis
Didn't you?
It rained, didn't it?
English allows the complement of an auxiliary or of the infinitival marker to to be omitted, if it can be reconstructed from context. Sometimes called VP ellipsis, this construction is better identified as post-auxiliary ellipsis, since the missing constituent need not be VP if the auxiliary in question can take a non-VP complement (e.g. Who is at the door?/Kim is.). The interpretation of elided constituents is an interesting and complex area of English grammar, especially as it appears to be constrained by the linguistic context, though that can be from a prior sentence, including one uttered by a different speaker. [FIXME: Citation Needed]
From the point of view of the sentence-by-sentence processing done by the ERG, the interpretation is not given by the sentence itself. Accordingly, the elided element(s) are represented via one of two EPs. These characteristic EPs can be used to trigger the search for an appropriate antecedent, which will typically be not just a single EP but rather a subgraph of some MRS from some utterance in context.
The first, used when the subject is referential (non-expletive) has an ARG1 role.
_:ellipsis_ref(ARG1:x0)
This serves two purposes: First, it reflects the fact that the subject in such cases will be interpreted as playing some role within the reconstructed subgraph after the antecedent expression has been found. Second, it ensures that the MRS produced by the grammar in cases of ellipsis with contentful subjects will be connected graphs. Without that ARG1 position, the subject's semantic contribution would not be integrated.
The second characteristic EP appears when the subject is an expletive:
_:ellipsis_expl(ARG1:u0)
This EP also has an ARG1 role, but it is not linked to anything.
-
I can't get ellipsis_expl to show up with it, even in the example from the testsuite (1212).
-
I see two parses for Weren't there? One has be_v_there and one doesn't. The first is the only one I expected.
-
Why would ellipsis_expl need an ARG1?
1212
-
ErgSemantics main page
-
Inventory of semantic phenomena (to be) documented
Home | Forum | Discussions | Events