Tutorial run by Katrien Beuls, Paul Van Eecke and Jens Nevens (Vrije Universiteit Brussel, Belgium)
Fluid Construction Grammar (FCG) is an open-source toolbox that allows linguists to operationalize specific linguistic knowledge in terms of construction schemas and linguistic pathways. FCG implements comprehension and production of utterances as linguistic pathways, being chains of consecutive operations (construction applications) over intermediary representations that hold all information that is known about an utterance at a certain point during processing (transient structures). A construction schema captures a pattern of language usage, and it expresses when and how a transient structure can be expanded. Constructions and transient structures are represented in the same way, through feature structures consisting of units and features expressing formal and semantic categorisations. In line with usage-based approaches, the categories that are used in FCG are dynamic, open-ended, continuously expanding, and shaped by learning. They obtain their function from the systemic structures in which they appear. Transient structures and constructions combine information from multiple perspectives, including the intonation and stress patterns, the decomposition of the words into lexical stems and affixes, the ordering of the words and their organisation into hierarchical phrase structure, the functions of words and phrases and their dependencies, the argument structures specifying who is doing what to whom, the logical structure (e.g. the propositions and how they are logically connected, quantification), the temporal structure of the events described in the utterance, and the information structure (what is foreground/background, what is the topic).
In this one-day, hands-on tutorial, participants will learn how the key concepts of construction grammar can be operationalized into working, computational models that can be used for comprehending and formulating utterances. In the morning, we will start with a short introduction on the basics of FCG (constructions, transient structures, pathways, search), after which the participants will write their own lexical and grammatical constructions. In the afternoon, we will explore more advanced features of FCG, including different visualisations, flexible construction application, and learning of constructions. We will also study the basic design principles of a larger grammar for Simple English, which is available through the FCG web service.
Summary of the Program
|Slot 1||Introduction to Fluid Construction Grammar||Hands-on exercises|
|Slot 2||Live-coding session: Aspectual Contours (Bill Croft)||The use of FCG in AI applications|
Slides and code
- Slides introducing FCG and the topic of the live coding session.
- An FCG grammar for English aspect (based on an analysis by Bill Croft in Verbs).
- Paul Van Eecke & Katrien Beuls. (2018). Syntax and Semantics of FCG. Github Wiki article, Evolutionary Linguistics Association
- Luc Steels. (2017). Basics of Fluid Construction Grammar. Constructions and Frames 9(2), 178 – 225.
- Remi van Trijp. (2017). A Computational Construction Grammar for English. The AAAI 2017 Spring Symposium on Computational Construction Grammar and Natural Language Understanding.
- Paul Van Eecke & Katrien Beuls. (2017). Meta-layer Problem Solving for Computational Construction Grammar. The AAAI 2017 Spring Symposium on Computational Construction Grammar and Natural Language Understanding.
- Luc Steels and Szathmary, Eörs. (2016) Fluid Construction Grammar as a Biological System Linguistics Vanguard 2(1).
- This hands-on tutorial is part of the International Construction Grammar Conference (ICCG-10) and will take place on Friday 20 July 2018 (9am - 4pm).
- Venue: ENS, Salle Dusanne, 45 rue d'Ulm, 75005 Paris, France.
- Participants must bring their own laptop with Babel2 installed (which includes the FCG software). Participants may receive a notification before the tutorial with a new download link if there are critical updates to the software.
- The registration fee is included in the ICCG10 conference fee (official registration for the tutorials will soon be opened)
- Pre-registration is required through firstname.lastname@example.org.
- More information: email@example.com.
About the Course Instructors
Dr Katrien Beuls received her MSc in Speech and Language Processing from the University of Edinburgh in 2009 and her MA in Linguistics from the University of Leuven in 2008. She defended her PhD in 2013 at the VUB on a new framework for Computer-Assisted Language Learning that is rooted in Computational Construction Grammar and uses fundamental techniques of symbolic AI such as student modelling and the representation of tutoring strategies to provide personalised feedback to the student. At present she is assistant professor in Computer Science lecturing in Natural Language Processing technologies and symbolic programming. She is Principal Investigator on the ESSENCE Evolutionary Semantics Marie Curie initial training network and prime coordinator of the ATLANTIS (Era-Net) project. Her main research interests lie in the applications of Computational Construction Grammar in diverse settings including language tutoring, formal grammars, music composition and evolutionary linguistics.
Paul Van Eecke obtained a Master of Science in Artificial Intelligence (summa cum laude) from the Engineering Science Faculty of the KU Leuven in 2013 and a Master of Arts in Linguistics (summa cum laude) from the Arts Faculty of the KU Leuven in 2012. Since 2014, he works as an assistant researcher at the Sony Computer Science Laboratory in Paris and is enrolled as a PhD student at the VUB. His main research interests are situated in the domain of computational construction grammar and its applications. Since 2014, he has been one of the main developers of the Fluid Construction Grammar (FCG) system and has developed algorithms for making construction application flexible and for learning constructions from data.
Jens Nevens obtained both Bachelor degree (magna cum laude, 2015) and Master degree (summa cum laude, 2017) in Computer Science, Artificial Intelligence profile, at Vrije Universiteit Brussel (VUB). Since then, he is enrolled as a PhD student at VUB. His main research interests are evolutionary linguistics, language tutoring and language grounding in robots.Top