Computational representation and annotation system for Cognitive Construction Grammar
Kai Chen, Dong Huang, Qingcai Chen, Chao Zhang · 2016
Construction Grammar (CxG) with strong explanatory power for language phenomena and language learning is still a stranger for most of natural language processing (NLP) tasks. The main reasons include challenges brought by the opening definition of construction, the lacking of a large scale construction knowledge base, the lacking of annotation tools and construction-annotated corpus which are big obstacles for using CxG in NLP. In this paper, we firstly present a flexible computational representation for Cognitive Construction Grammars (CCxG) which is based on the argument structure representation of CCxG. A CCxG definition and annotation system are then implemented. Through shallow parsing, this system provides the visualization of annotated results by the Box Diagram. By emphasizing the computable aspect rather than the cognitive and psychological aspects of the CCxG, we purposely provide NLP researchers and engineers an easily usable tool platform for building applicable construction knowledge base and large scale of training and testing corpus for CCxG parser. It is also useful platform for linguists to investigate and analyze new emerging language phenomena.