ErgSemantics_Fragments

ESD Test Suite Examples

Hey!
Dogs.
Arrived.
Because Abrams left.
Probably Abrams.

Linguistic Characterization

Many well-formed utterances in context are not complete sentences, but instead consist of just a word or phrase (or a short sequence) which we interpret as an argument or modifier of an implied eventuality we call 'unknown' which can be supplied by the context.

Motivating Examples

• Maybe tomorrow, but probably not with Kim.

• Who's going to send out the announcement?/Maybe Kim, but not on Tuesday.

MRS Fingerprints

Nominal phrases are treated as an underspecified argument 'ARG' of the 'unknown' relation:

h0:unknown(ARG0 e1, ARG x2)
_(ARG0 x2)

Intersective adverbials and prepositional phrases are treated as intersective modifiers of the 'unknown' relation:

h0:unknown(ARG0 e1)
h0:_(ARG1 e1)

Adjectives are treated as intersective modifiers of some instance which is an argument of the 'unknown' relation, but for compactness, the adjective's relation identifies its handle with that of the 'unknown' relation, as an abbreviation for adding an 'unknown_n' relation (FIX?):

h0:unknown(ARG0 e1, ARG x2)
h0:_(ARG1 x2)

Scopal adverbials are treated as scopal modifiers of the 'unknown' relation:

h0:unknown(ARG0 e1)
h1:_(ARG1 h2)

{ h2 =q h0 }

Verb phrases are treated as propositions missing the argument that would have bee supplied by the subject:

h0:_v(ARG0 e1)

Sequences of an adverb and a nominal phrase like 'probably Kim' are treated as a modifier and an argument, respectively:

h0:unknown(ARG0 e1, ARG x2)
_(ARG0 x2)
h1:_(ARG1 h2)

{ h2 =q h0 }

Interactions

• Fragments can also be constituents of a larger utterance, including arguments of verbs of saying, and parts of run-on sentences.

Reflections

Open Questions

• For verb phrase fragments, how do we distinguish between e.g. "Hired Kim" and "Kim was hired"?

Grammar Version

• 1212.

More Information

• ErgSemantics main page

• Inventory of semantic phenomena (to be) documented

• How to cite this work