• WeSearch
• HankøSchedule

Background

As a supplementary component of the WeSearch project, there is funding for collaborative activities towards semantic interface corroboration, i.e. fixing (in various possible senses) salient properties of the downstream interface to the ERG (and, in principle, similar grammars).

As a central element of this line of work, the project team will host a by-invitation meeting of experts, starting in the morning of Monday, April 30 and lasting until the afternoon of Wednesday, May 2. At least some participants will arrive the preceding day and will gather informally at the meeting site starting sometime in the afternoon of Sunday, April 29, already.

The meeting will be held at Hankø Fjordhotell, close to the city of Fredrikstad and both Moss Lufhavn Rygge (RYG) and Oslo Lufthavn Gardermoen (OSL). Upon completion of the main meeting, sometime late afternoon on May 2, some of the participants will head on to Oslo and stay for another night or two. Stephan is prepared to make accomodation arrangements for everyone, both at Hankø and in Oslo.

Participants

The following have confirmed particpation:

Stephan will make arrangements to meet those flying into Gardermoen (OSL) and will then drive down to Hankø. We expect to arrive at the hotel between 15:00 and 16:00 on Sunday. Rebecca and Angelina will take the train to Fredrikstad and make arrangements to meet up with Ann somewhere along the way. Lilja will head straight for the hotel a little later on Sunday.

Tuesday is a national holiday in Norway, so it seems only fair that we take part of the day off and leave our island for an excursion into the old town of Fredrikstad sometime in the afternoon. We will have dinner there and return to the hotel in the evening, for a late-evening session (Lilja needs to return to Oslo Tuesday night, to attend a seminar the next morning).

Programme

In the tradition of earlier meetings, participants will determine the programme jointly, possibly over drinks already on Sunday evening. Further below on this page, we will jointly (and incremetally) construct the agenda for the meeting.

As one source of inspiration, say to derive a list of (semantic) phenonema that are either exceptionally well analyzed in the current ERG or maybe deserve further thinking, we suggest the collection of sentences compiled by Yusuke Miyao and colleagues for the 2008 Workshop on Parser Evaluation Across Frameworks. A gold-standard Redwoods treebank for these sentences is available as part of the ERG trunk. UiO will make available graphical renderings of the syntacto-semantic annotations in other frameworks prior to the meeting.

Candidate Topics

• makes notes of guiding principles in semantic design, e.g. paraphrases

• towards a catalogue of stable semantic analyses (and documentation)

• a list of 'known bugs' (e.g. degree specifiers on quantifiers; scopal arguments in coordination: she can and must sing, this city john admires and visits.)

• revisit (unrealized) plans from the past, e.g. fewer quantifiers, type distinctions for 'unbound' variables, non-restrictive relatives
• updates on information structure (emily)
• negation in a cross-linguistic perspective (emily)

• anaphor links (ICONS) (ann)

• underspecification of modifier attachment: is it desirable? if so, how could it be done?

• the continuum between logical-form semantics vs. (syntacto-semantic) dependencies

  • predicate symbols underspecified for arity

• requirements of composition vs. interface aspects

• specification of co-reference: *''I bit her tongue'' (francis, heartfelt)

• apposition (non-restrictive relative clauses)

• look at analyses in PEST sentences

• parentheticals
• correct characterization (in PET)
• contentful and technical: SEM-I representation

• predicate inventories, 'optional' arguments, scopal vs. non-scopal arguments: This came up at least in the PEST examples and coordination discussions

• are we proud of our current analysis of coordination (in terms of the handle vs. index duality)

• internal (semantic) structure of proper names: British Broadcasting Corporation, [Frankfurt and] New York Stock Exchange => see PEST examples discussion

• guiding principles in deciding on predicate names, e.g. closed class (that vs. those) => see PEST examples discussion

• fine points of comparatives (how proud are we of our current analysis)? => see PEST examples discussion

• whatever else comes up

Photos

photos Password: oslofjord