MWE Workshop

21st @multiword is going on in the Santa Ana room @naaclmeeting. @complingy is talking on "Meaning Construction at the Syntax-Lexis Nexus"

⚡️Final CFP: 21st Workshop on Multiword Expressions (MWE2025) ⏰Submit Your Papers By: Feb 13 at 23:59 (AoE) 👉More details: multiword.org/mwe2025/ @naaclmeeting @aclmeeting @insight_centre @UniDive_CA21167 @vgiouli @shashwatup9k #siglex

6PM: End of another #MWE workshop! A big thank you to the organisers, the program committee, and everyone who attended this year's #MWEUD2024 workshop, for making it such a special event. See you next year!

What about identification of genre and domain in text? Would additional annotation layers be valuable for a broad range of analyses? How are MWEs best represented in lexica? Some of the questions raised during the discussion.

Looking towards the future, Shared Tasks for both UD parsing and MWE identification are proposed, with interest voiced in opening Shared Tasks for semantic parsing and term identification.

5.30: Community Discussion begins with an announcement of the Best Paper Award, which goes to...

-Sparse Logistic Regression with High-order Features for Automatic Grammar Rule Extraction from Treebanks -Joint Annotation of Morphology and Syntax in Dependency Treebanks #MWEU2024 colocated at @LrecColing

-Domain-Weighted Batch Sampling for Neural Dependency Parsing -Redefining Syntactic and Morphological Tasks for Typologically Diverse Languages -UCxn: Typologically Informed Annotation of Constructions Atop Universal Dependencies

4PM: Poster session 2 will take place after the coffee break! ☕️🥐 Stop by the poster area to see presentations on: -Synthetic-Error Augmented Parsing of Swedish as a Second Language: Experiments with Word Order -MaiBaam: A Multi-Dialectal Bavarian Universal Dependency Treebank

Read the paper here: aclanthology.org/2024.mwe-1.27/ #MWEUD2024 colocated at @LrecColing

3.40: Qi Yu of University of Konstanz presents "Ad Hoc Compounds for Stance Detection"; German's system of compound creation allows for encoding attitudinal meaning (e.g. 'Merkel-Land'), however, these ad-hoc creations are not reliably identified or processed automatically.