DATE

Thursday 4 September 2014, as part of the 8th International Conference on Construction Grammar link

REGISTRATION

Through the main conference link

CONVENORS

Remi van Trijp and Luc Steels

DESCRIPTION OF THE WORKSHOP

Construction grammar is a family of linguistic theories that is rapidly gaining importance in all subfields of linguistics. Construction grammar has also enthused computational linguists, who see in construction grammar the potential to re-establish the lost connection between linguistic theory and natural language processing. Indeed, as argued by Schneider and Tsarfaty (2013: 451), the construction grammar “perspective presents a formidable challenge to the computational linguistics/natural language processing community”, and it may lead to “a more comprehensive account of language processing than our field is able to offer today.”

This workshop aims to identify the challenges of computational construction grammar and to bring together researchers who have been working on concrete solutions and operationalizations. In order to achieve this, an open call will be launched for research on three main themes:

(a) How to implement construction grammar? Construction grammar features some unique characteristics that distinguish it from other linguistic theories, such as its syntax-­‐ lexicon continuum, its free combination of constructions (including the possibility of coercion) and its handling of long-­‐ distance dependencies. To what extent can we reuse tools that have already been developed in the CL/NLP community, and which tools are missing? Some researchers have proposed extensions of well-­‐established probabilistic methods (e.g. Jurafsky, 1996; Bod, 2009; Chen et al., 2011) and/or typed feature logics (e.g. Bryant, 2004; Boas and Sag, 2012), whereas others have opted for a more radical recombination of language technologies (e.g. Steels, 2004; van Trijp, 2013).

(b) Case studies: The workshop also welcomes submissions focusing on specific case studies (ranging from a specific construction to a family of constructions). Case study submissions should focus on concrete solutions. Submissions that are accompanied by a demonstration are preferred.

(c) Experiments: Construction grammar has become popular in many fields of linguistics, ranging from language acquisition to corpus linguistics. In each subfield, computational models have emerged that argue for a construction grammar account, such as language learning (Chang, 2008), psycholinguistic modeling (Bryant, 2004), language evolution (Steels, 2012) and language grounding on robots (Steels and Hild, 2012).

PROGRAMME

Time	Speaker	Talk
10:30-11:00h	Luc Steels	Introduction to the workshop [wpfilebase tag=fileurl id=32 linktext=abstract /]
11:00-11:30h	Mark Steedman	Robust parsing and interpretation for construction grammars [wpfilebase tag=fileurl id=31 linktext=abstract /]
11:30-12:00h	Michael Ramscar	Production, comprehension and synthesis: a communicative perspective on language [wpfilebase tag=fileurl id=30 linktext=abstract /]
12:00-12:30h	Remi van Trijp	How constructions can Freely interact with Each Other [wpfilebase tag=fileurl id=33 linktext=abstract /]
12:30-14:30h		Lunch Break
14:00-14:30h	Stefan Müller	The CoreGram project: theoretical linguistics, theory development and verification [wpfilebase tag=fileurl id=26 linktext=abstract /]
14:30-15:00h	Ellen Dodge	A computationally implemented Embodied Construction Grammar. Analysis of Argument Structure Constructions [wpfilebase tag=fileurl id=28 linktext=abstract /]
15:00-15:30h	Natalia Levshina	A Radically Data-Driven Construction Grammar: English Constructions of Letting and Vector Space Models [wpfilebase tag=fileurl id=29 linktext=abstract /]
15:30-16:00h		Coffee break
16:00-16:30h	Johan Bos	Perspectives from Computational Semantics [wpfilebase tag=fileurl id=34 linktext=abstract /]
16:30-17:00h	Katrien Beuls	How computational construction grammar can revolutionize language tutoring applications [wpfilebase tag=fileurl id=27 linktext=abstract /]
17:00-17:30h		Concluding panel: Challenges for computational construction grammar