GURT 2023 Program Schedule

This page gives the schedule of the GURT 2023 Program.

All times listed here are in Eastern Time (New York/Washington D.C.) Note: Daylight Saving Time in Washington DC will begin Sunday, March 12 at 2am, “springing us ahead” 1 hr! This site can help you sort out the time difference for your location each day.

Logistics:

  • Plenary sessions (talks, keynotes, panel) will take place in the ICC Auditorium. Remote presenters will receive a Zoom link.
    • Time allocations: long papers—15+7 (15 min. plus 7 min. for questions); short papers—10+5.
  • The welcome reception, coffee breaks, and the poster session will take place in the ICC Galleria.
    • Food: The welcome reception will include wine and light bites (including meat, seafood, and vegetarian options). Morning coffee breaks will include pastries. For meals, there are dining options on and near campus.
    • Poster boards will be 30 in. by 40 in., and can be oriented horizontally or vertically. (local poster printing options)

Thursday, March 9, 2023

10:00 Excursion to Planet Word museum (meet at the museum; make sure to reserve your tickets days in advance!)

15:00 Registration desk opens (ICC Galleria)

16:00 Keynote: Guy Perrier Why is graph rewriting interesting for computational linguistics?

Graph Rewriting is a computational paradigm, not widely used in Computational Linguistics (CL). And yet, all linguistics resources such as treebanks, corpora annotated with semantics and lexicons can be considered as graphs and many treatments of these resources boil down to graph matching or graph rewriting. Unfortunately, there is no standard model for graph rewriting. Bruno Guillaume, Guillaume Bonfante and myself have designed a model adapted to CL. We implemented it in the GREW tool (https://grew.fr/). After a description of the model and the tool, I will present various applications to CL. First, I will consider the component of GREW dedicated to graph matching: GREW-MATCH (http://match.grew.fr/). It can be used independently of GREW for the linguistic exploration of annotated corpora, for annotation correction and for grammar extraction from treebanks. GREW, with its rewriting component, can be used to convert an annotated corpus from one format to another, and to produce a representation at a certain linguistic level (e.g. semantics) from a representation at another level (e.g. syntax).

17:00 Welcome Reception 🍷🧀🍤

20:00 End of Day

Friday, March 10, 2023

8:00 Coffee & Pastries ☕🥐

8:30 – 12 Registration

9:00 Paper Presentations

10:30 Coffee Break ☕🥐

11:00 Paper Presentations

12:30 Lunch

13:30 – 16:30 Registration

14:00 Keynote: Joan Bresnan Cooccurrence probabilities predict English pronoun encliticization

English object pronouns that encliticize to their host verbs, as in get’em, stop’er, are common in conversational speech but rarely represented in orthographic texts. They have conflicting analyses in previous linguistic work, and have escaped corpus study. From two corpus studies of American speech, I will show that pronoun encliticization is predicted by the probability of cooccurrence of lexical verbs with their object pronouns. The higher the conditional probability of cooccurrence of individual lexical verbs with an object pronoun, the greater their likelihood of encliticization in ongoing conversations. This empirical finding is new, but it is just what would be expected in the hybrid formal and usage-based theory of Bresnan (2021), which combines a dynamic, exemplar-based lexicon with a lexical syntactic theory (LFG) of the co-lexicalization of adjacent words. This work implies that not only the grammar of English object enclitics but also their usage probabilities are part of English speakers’ implicit linguistic knowledge in active use during language production.

15:00 Coffee Break ☕🍪

15:30 Paper Presentations

17:00 End of Day

Saturday, March 11, 2023

8:00 Coffee & Pastries ☕🥐

8:30 – 12:00 Registration

9:00 Paper Presentations (remote presenters)

10:00 Coffee Break ☕🥐

10:30 Keynote: Joakim Nivre Ten Years of Universal Dependencies

Universal Dependencies (UD) is a project developing cross-linguistically consistent treebank annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning, and parsing research from a language typology perspective. Since UD was launched almost ten years ago, it has grown into a large community effort involving over 500 researchers around the world, together producing treebanks for 138 languages and enabling new research directions in both NLP and  linguistics. In this talk, I will review the history and development of UD, explore the UD research community through a bibliographic survey, and discuss challenges that we need to face when bringing UD into the future.

11:30 Lunch

13:00 Poster Session (ICC Galleria) ☕🍪

13:30 – 16:00 Registration

14:30 Coffee Break ☕🍪

15:00 Paper Presentations

16:30 End of Day

Sunday, March 12, 2023

8:00 Coffee & Pastries ☕🥐

8:30 – 12:00 Registration

9:00 Paper Presentations (remote presenters)

10:15 10:30 Coffee Break ☕🥐

10:45 10:55 Keynote: Jonathan Dunn Emerging Structure in Computational Construction Grammar

This talk focuses on the emergence of grammatical structure in computational CxG given increasing amounts of exposure (i.e., training data). We divide this process of emergence into three phenomena: (i) the increasing complexity of category formation in producing basic slot-constraints, (ii) the increasing level of abstractness as constructions migrate from item-specific to more generalized constraints, and (iii) the clipping together of first-order constructions into larger second-order constructions. These three types of scaffolded structure produce grammars of increasing complexity given exposure to more training data.

11:45 11:55 Lunch

13:15 13:30 Panel

13:30 – 16:30 Registration

15:00 Coffee Break ☕🍪

15:30 Paper Presentations

17:00 End of Day