Open AccessLearnability and falsifiability of Construction Grammars
Author(s) -
Jonathan Dunn
Publication year - 2017
Publication title -
proceedings of the linguistic society of america
Language(s) - English
Resource type - Journals
ISSN - 2473-8689
DOI - 10.3765/plsa.v2i0.4009
Subject(s) - falsifiability , learnability , grammar , set (abstract data type) , natural language processing , artificial intelligence , computer science , linguistics , programming language , philosophy , epistemology
The strength of Construction Grammar (CxG) is its descriptive power; its weakness is the learnability and falsifiability of its unconstrained representations. Learnability is the degree to which the optimum set of constructions can be consistently selected from the large set of potential constructions; falsifiability is the ability to make testable predictions about the constructions present in a dataset. This paper uses grammar induction to evaluate learnability and falsifiability: given a discovery-device CxG and a set of observed utterances, its learnability is its stability over sub-sets of data and its falsifiability is its ability to predict a CxG.