Archival static copy of the original website

Compounds and Multiword Expressions (MWEs) - for example, nominal compounds ("frying pan"), verb-particle constructions ("take off"), and idiomatic expressions ("break the ice") - constitute a highly frequent phenomenon in natural language, yet one that provides a challenge to the traditional division between lexicon and grammar in linguistics and language technology.

The goal of this workshop is to discuss current activities and outstanding problems in the area of compounds and MWEs from a theoretical and/or applications-oriented perspective. Reports about on-going work and original research are thus equally welcome. We believe that the relevant area can be broadly structured as follows:

• Identification: Although much compound identification can be carried out with high accuracy, identifying MWEs in running text is still a challenging problem, which might require combinations of traditional techniques such as static lists and rules with novel techniques based on machine learning.
• Interpretation: Semantic interpretation of compounds and MWEs is a central problem. Supervised approaches might be based on specifying the semantics of compounds and MWEs using a pre-defined, static set of semantic relations, whereas unsupervised approaches might be based on charting their distributions in word space.
• Disambiguation: Most compounds and MWEs are ambiguous in various ways. A key problem for MWEs is to determine whether it is used non-compositionally (idiomatically) or compositionally (literally) in a particular context.
• Applications: Identifying compounds and MWEs in context and understanding their syntax and semantics is crucial for many natural language applications. For example, in the area of terminology, nearly all instances of "unknown" words turn out to be compounds. In information retrieval, an important issue is the relative informational contribution of compound components. In translation, one issue is the choice of whether a word sequence constitutes a MWE and should be translated as such, or if the component words should be translated literally.

Note: This workshop has the support of the (organisers of the) international workshop on Multiword Expressions (see http://multiword.sourceforge.net/PHITE.php?sitesig=CONF). We would like to underline that, although both overlap contentwise, the SLTC Workshop is a locally organized event where reports about on-going work as well as completed results are equally welcome, and from which only abstracts will be published.

Submission:

Submissions are invited as two-page abstracts, following the formatting guidleines of SLTC. Submissions should be emailed as a pdf-file to .

Registration:

The workshop is part of SLTC, and registration is performed through the SLTC conference webpage. No separate registration is needed for the workshop. The registration deadline is October 10, 2010.

Program:

13.15	Organizers	Introduction
13.20	Marion Weller and Fabienne Fritzinger	A Hybrid Approach for the Identification of Multiword Expressions
13.45	Karin Friberg Heppin	Decomposition of Compounds and the Effect on Search Key Effectiveness in Information Retrieval
14.10		Break
14.20	Robert Östling	A Construction Grammar Method for Disambiguating Swedish Compounds
14.45	Lidia Pivovarova and Elena Yagunova	Collocation Extraction and Text Analysis: Different Types of Collocations and Different Genres
15.10-15.30	Organizers	Final session

Location:

SLTC 2010, Linköping, Sweden.
Room: John von Neumann (map)
For travel directions, see the SLTC webpage.

Important dates:

Submission deadline: September 20
Notification: September 27
Registration deadline: October 10
Workshop: October 29

Organizers:

Magnus Merkel (LiU)
Magnus Sahlgren (SICS)
Sara Stymne (LiU)
Mats Wirén (SU)
Robert Östling (SU)

Contact: .