This paper presents the Recursive Adaptable Grammar (RAG) formalism. RAGs allow arbitrary Turing-powerful language analysis to be described entirely in terms of a single level of "context-free" derivation. RAGs superficially resemble a limited form of Extended Attribute Grammars (EAGs); but while the EAG derivation step relation may entail arbitrary subsidiary computations, the normal RAG derivation step relation is elementary. The RAG formalism is introduced and defined; basic properties of the formalism are shown; and a well-behavedness property called strong answer-encapsulation is developed.
Shutt, John N.
(1999). Recursive Adaptable Grammars. .
Retrieved from: http://digitalcommons.wpi.edu/computerscience-pubs/199