In this paper a notion of a grammar is defined which is an extension of a context-free grammar by index productions of the form Aƒ → B and A → Bƒ. With these productions one can ‘compute’ information coded by strings of indices. We consider two modes of derivation which distribute this information over context-free productions in a different way. The two derivation modes yield the classes of indexed and type-0 languages respectively. The context-free-like structure of the grammar is used as a tool to investigate normal-form transformations, Dyck languages and homomorphic characterizations.
Read full abstract